diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10157 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9982222222222221,
+  "global_step": 1686,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.352941176470589e-07,
+      "loss": 1.4934,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 1.4308,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2058823529411767e-06,
+      "loss": 1.5656,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 1.4176,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6764705882352942e-06,
+      "loss": 1.3197,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.411764705882353e-06,
+      "loss": 1.1572,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.147058823529412e-06,
+      "loss": 0.9889,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.7601,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.61764705882353e-06,
+      "loss": 0.5436,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.3529411764705884e-06,
+      "loss": 0.5049,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.088235294117648e-06,
+      "loss": 0.4444,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 0.4241,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.558823529411764e-06,
+      "loss": 0.406,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0294117647058824e-05,
+      "loss": 0.3248,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1029411764705883e-05,
+      "loss": 0.3274,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.3797,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2718,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.323529411764706e-05,
+      "loss": 0.26,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3970588235294118e-05,
+      "loss": 0.4324,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4705882352941177e-05,
+      "loss": 0.2265,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5441176470588237e-05,
+      "loss": 0.3209,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6176470588235296e-05,
+      "loss": 0.3329,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6911764705882355e-05,
+      "loss": 0.2335,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 0.2215,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8382352941176472e-05,
+      "loss": 0.1997,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9117647058823528e-05,
+      "loss": 0.2286,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9852941176470586e-05,
+      "loss": 0.2931,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.058823529411765e-05,
+      "loss": 0.2361,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1323529411764707e-05,
+      "loss": 0.3129,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2058823529411766e-05,
+      "loss": 0.2784,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2794117647058825e-05,
+      "loss": 0.2612,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.3529411764705884e-05,
+      "loss": 0.3002,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.4264705882352942e-05,
+      "loss": 0.1918,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.5e-05,
+      "loss": 0.2737,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.5735294117647057e-05,
+      "loss": 0.3001,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.647058823529412e-05,
+      "loss": 0.2636,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7205882352941174e-05,
+      "loss": 0.2934,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7941176470588236e-05,
+      "loss": 0.3193,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8676470588235295e-05,
+      "loss": 0.3171,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9411764705882354e-05,
+      "loss": 0.2971,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.0147058823529413e-05,
+      "loss": 0.2394,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.0882352941176475e-05,
+      "loss": 0.294,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.161764705882353e-05,
+      "loss": 0.2625,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.235294117647059e-05,
+      "loss": 0.3554,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.308823529411765e-05,
+      "loss": 0.3236,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.382352941176471e-05,
+      "loss": 0.2163,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.455882352941177e-05,
+      "loss": 0.3002,
+      "step": 47
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.529411764705883e-05,
+      "loss": 0.3036,
+      "step": 48
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6029411764705886e-05,
+      "loss": 0.2182,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6764705882352945e-05,
+      "loss": 0.2003,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.3377,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8235294117647055e-05,
+      "loss": 0.1783,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.897058823529412e-05,
+      "loss": 0.2838,
+      "step": 53
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.970588235294117e-05,
+      "loss": 0.2711,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.044117647058824e-05,
+      "loss": 0.2823,
+      "step": 55
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.11764705882353e-05,
+      "loss": 0.3191,
+      "step": 56
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.1911764705882356e-05,
+      "loss": 0.329,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.2647058823529415e-05,
+      "loss": 0.1626,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.3382352941176474e-05,
+      "loss": 0.3719,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.411764705882353e-05,
+      "loss": 0.2546,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.485294117647059e-05,
+      "loss": 0.2373,
+      "step": 61
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.558823529411765e-05,
+      "loss": 0.2828,
+      "step": 62
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.632352941176471e-05,
+      "loss": 0.2987,
+      "step": 63
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.705882352941177e-05,
+      "loss": 0.3749,
+      "step": 64
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.7794117647058826e-05,
+      "loss": 0.2856,
+      "step": 65
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8529411764705885e-05,
+      "loss": 0.367,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9264705882352944e-05,
+      "loss": 0.3635,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5e-05,
+      "loss": 0.2624,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9999952874867166e-05,
+      "loss": 0.3697,
+      "step": 69
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.999981149964632e-05,
+      "loss": 0.2175,
+      "step": 70
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9999575874870454e-05,
+      "loss": 0.2936,
+      "step": 71
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.999924600142787e-05,
+      "loss": 0.225,
+      "step": 72
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.999882188056219e-05,
+      "loss": 0.3736,
+      "step": 73
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.999830351387237e-05,
+      "loss": 0.2763,
+      "step": 74
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.999769090331264e-05,
+      "loss": 0.3704,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.999698405119256e-05,
+      "loss": 0.3845,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.999618296017697e-05,
+      "loss": 0.275,
+      "step": 77
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.999528763328598e-05,
+      "loss": 0.331,
+      "step": 78
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9994298073895e-05,
+      "loss": 0.2806,
+      "step": 79
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.999321428573467e-05,
+      "loss": 0.2779,
+      "step": 80
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999203627289087e-05,
+      "loss": 0.1724,
+      "step": 81
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.999076403980475e-05,
+      "loss": 0.3406,
+      "step": 82
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.998939759127262e-05,
+      "loss": 0.3398,
+      "step": 83
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.998793693244601e-05,
+      "loss": 0.3071,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.998638206883162e-05,
+      "loss": 0.321,
+      "step": 85
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9984733006291306e-05,
+      "loss": 0.2683,
+      "step": 86
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9982989751042046e-05,
+      "loss": 0.3641,
+      "step": 87
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.998115230965593e-05,
+      "loss": 0.2668,
+      "step": 88
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997922068906015e-05,
+      "loss": 0.3709,
+      "step": 89
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9977194896536904e-05,
+      "loss": 0.2223,
+      "step": 90
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9975074939723476e-05,
+      "loss": 0.3262,
+      "step": 91
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.997286082661211e-05,
+      "loss": 0.3202,
+      "step": 92
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9970552565550056e-05,
+      "loss": 0.5339,
+      "step": 93
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9968150165239455e-05,
+      "loss": 0.3457,
+      "step": 94
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9965653634737414e-05,
+      "loss": 0.2291,
+      "step": 95
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9963062983455856e-05,
+      "loss": 0.3465,
+      "step": 96
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9960378221161566e-05,
+      "loss": 0.2793,
+      "step": 97
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9957599357976135e-05,
+      "loss": 0.3772,
+      "step": 98
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995472640437591e-05,
+      "loss": 0.286,
+      "step": 99
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.995175937119194e-05,
+      "loss": 0.236,
+      "step": 100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9948698269609984e-05,
+      "loss": 0.2889,
+      "step": 101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.994554311117042e-05,
+      "loss": 0.3081,
+      "step": 102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9942293907768244e-05,
+      "loss": 0.2765,
+      "step": 103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.993895067165297e-05,
+      "loss": 0.2452,
+      "step": 104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9935513415428636e-05,
+      "loss": 0.2907,
+      "step": 105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9931982152053744e-05,
+      "loss": 0.3427,
+      "step": 106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9928356894841195e-05,
+      "loss": 0.4537,
+      "step": 107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.992463765745824e-05,
+      "loss": 0.2331,
+      "step": 108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.992082445392644e-05,
+      "loss": 0.248,
+      "step": 109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.991691729862161e-05,
+      "loss": 0.3122,
+      "step": 110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.991291620627379e-05,
+      "loss": 0.3075,
+      "step": 111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.990882119196712e-05,
+      "loss": 0.2334,
+      "step": 112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.990463227113985e-05,
+      "loss": 0.3013,
+      "step": 113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.990034945958427e-05,
+      "loss": 0.3308,
+      "step": 114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.98959727734466e-05,
+      "loss": 0.2707,
+      "step": 115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.989150222922702e-05,
+      "loss": 0.1658,
+      "step": 116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.988693784377951e-05,
+      "loss": 0.339,
+      "step": 117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9882279634311855e-05,
+      "loss": 0.2905,
+      "step": 118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.987752761838557e-05,
+      "loss": 0.2974,
+      "step": 119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.987268181391578e-05,
+      "loss": 0.3136,
+      "step": 120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.986774223917123e-05,
+      "loss": 0.299,
+      "step": 121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.986270891277418e-05,
+      "loss": 0.1877,
+      "step": 122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.985758185370032e-05,
+      "loss": 0.4754,
+      "step": 123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9852361081278704e-05,
+      "loss": 0.2564,
+      "step": 124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.984704661519171e-05,
+      "loss": 0.294,
+      "step": 125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9841638475474936e-05,
+      "loss": 0.2843,
+      "step": 126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9836136682517106e-05,
+      "loss": 0.3276,
+      "step": 127
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.983054125706007e-05,
+      "loss": 0.3818,
+      "step": 128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9824852220198615e-05,
+      "loss": 0.2484,
+      "step": 129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9819069593380475e-05,
+      "loss": 0.3457,
+      "step": 130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.981319339840622e-05,
+      "loss": 0.2792,
+      "step": 131
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.980722365742918e-05,
+      "loss": 0.2812,
+      "step": 132
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.980116039295531e-05,
+      "loss": 0.3248,
+      "step": 133
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9795003627843204e-05,
+      "loss": 0.3296,
+      "step": 134
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9788753385303936e-05,
+      "loss": 0.3651,
+      "step": 135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.978240968890097e-05,
+      "loss": 0.2841,
+      "step": 136
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.977597256255012e-05,
+      "loss": 0.321,
+      "step": 137
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.976944203051942e-05,
+      "loss": 0.4024,
+      "step": 138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9762818117429036e-05,
+      "loss": 0.3881,
+      "step": 139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9756100848251194e-05,
+      "loss": 0.3207,
+      "step": 140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.974929024831009e-05,
+      "loss": 0.292,
+      "step": 141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.974238634328173e-05,
+      "loss": 0.2295,
+      "step": 142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.973538915919392e-05,
+      "loss": 0.2864,
+      "step": 143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9728298722426124e-05,
+      "loss": 0.274,
+      "step": 144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.972111505970936e-05,
+      "loss": 0.3188,
+      "step": 145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.971383819812612e-05,
+      "loss": 0.2161,
+      "step": 146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.970646816511023e-05,
+      "loss": 0.3466,
+      "step": 147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.969900498844681e-05,
+      "loss": 0.3352,
+      "step": 148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.96914486962721e-05,
+      "loss": 0.2267,
+      "step": 149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9683799317073425e-05,
+      "loss": 0.329,
+      "step": 150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.967605687968901e-05,
+      "loss": 0.2721,
+      "step": 151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.966822141330793e-05,
+      "loss": 0.3088,
+      "step": 152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.966029294746997e-05,
+      "loss": 0.2308,
+      "step": 153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.965227151206554e-05,
+      "loss": 0.3343,
+      "step": 154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.964415713733553e-05,
+      "loss": 0.3477,
+      "step": 155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9635949853871225e-05,
+      "loss": 0.2654,
+      "step": 156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9627649692614165e-05,
+      "loss": 0.2521,
+      "step": 157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.961925668485604e-05,
+      "loss": 0.3123,
+      "step": 158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.96107708622386e-05,
+      "loss": 0.3097,
+      "step": 159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.960219225675347e-05,
+      "loss": 0.4019,
+      "step": 160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.959352090074209e-05,
+      "loss": 0.3019,
+      "step": 161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.958475682689556e-05,
+      "loss": 0.2934,
+      "step": 162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9575900068254526e-05,
+      "loss": 0.2897,
+      "step": 163
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9566950658209074e-05,
+      "loss": 0.2571,
+      "step": 164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9557908630498573e-05,
+      "loss": 0.362,
+      "step": 165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9548774019211566e-05,
+      "loss": 0.2667,
+      "step": 166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.953954685878562e-05,
+      "loss": 0.2559,
+      "step": 167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.953022718400724e-05,
+      "loss": 0.2787,
+      "step": 168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.952081503001169e-05,
+      "loss": 0.2514,
+      "step": 169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.951131043228291e-05,
+      "loss": 0.3691,
+      "step": 170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.950171342665332e-05,
+      "loss": 0.3312,
+      "step": 171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.949202404930372e-05,
+      "loss": 0.4162,
+      "step": 172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.94822423367632e-05,
+      "loss": 0.3788,
+      "step": 173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.947236832590888e-05,
+      "loss": 0.3016,
+      "step": 174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9462402053965914e-05,
+      "loss": 0.3429,
+      "step": 175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.945234355850725e-05,
+      "loss": 0.2481,
+      "step": 176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9442192877453516e-05,
+      "loss": 0.2757,
+      "step": 177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9431950049072895e-05,
+      "loss": 0.2579,
+      "step": 178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.942161511198094e-05,
+      "loss": 0.3994,
+      "step": 179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9411188105140505e-05,
+      "loss": 0.248,
+      "step": 180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9400669067861486e-05,
+      "loss": 0.3515,
+      "step": 181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.939005803980078e-05,
+      "loss": 0.3575,
+      "step": 182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.937935506096207e-05,
+      "loss": 0.2387,
+      "step": 183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.936856017169572e-05,
+      "loss": 0.3204,
+      "step": 184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9357673412698544e-05,
+      "loss": 0.3025,
+      "step": 185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.934669482501376e-05,
+      "loss": 0.248,
+      "step": 186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.933562445003076e-05,
+      "loss": 0.2191,
+      "step": 187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.932446232948497e-05,
+      "loss": 0.3097,
+      "step": 188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.93132085054577e-05,
+      "loss": 0.4397,
+      "step": 189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9301863020375996e-05,
+      "loss": 0.2615,
+      "step": 190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.929042591701245e-05,
+      "loss": 0.3,
+      "step": 191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9278897238485066e-05,
+      "loss": 0.3201,
+      "step": 192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.926727702825708e-05,
+      "loss": 0.2454,
+      "step": 193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.925556533013682e-05,
+      "loss": 0.328,
+      "step": 194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9243762188277495e-05,
+      "loss": 0.2571,
+      "step": 195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.923186764717709e-05,
+      "loss": 0.3189,
+      "step": 196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9219881751678146e-05,
+      "loss": 0.3703,
+      "step": 197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9207804546967613e-05,
+      "loss": 0.442,
+      "step": 198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9195636078576684e-05,
+      "loss": 0.3707,
+      "step": 199
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.918337639238062e-05,
+      "loss": 0.2851,
+      "step": 200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.917102553459856e-05,
+      "loss": 0.323,
+      "step": 201
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9158583551793367e-05,
+      "loss": 0.2701,
+      "step": 202
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.914605049087146e-05,
+      "loss": 0.3427,
+      "step": 203
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.91334263990826e-05,
+      "loss": 0.329,
+      "step": 204
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.912071132401975e-05,
+      "loss": 0.3434,
+      "step": 205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9107905313618894e-05,
+      "loss": 0.3047,
+      "step": 206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9095008416158805e-05,
+      "loss": 0.2534,
+      "step": 207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.908202068026093e-05,
+      "loss": 0.2302,
+      "step": 208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.906894215488917e-05,
+      "loss": 0.3817,
+      "step": 209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.905577288934971e-05,
+      "loss": 0.2872,
+      "step": 210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.904251293329083e-05,
+      "loss": 0.2572,
+      "step": 211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.902916233670269e-05,
+      "loss": 0.2797,
+      "step": 212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9015721149917184e-05,
+      "loss": 0.2026,
+      "step": 213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.900218942360772e-05,
+      "loss": 0.2158,
+      "step": 214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.898856720878908e-05,
+      "loss": 0.442,
+      "step": 215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.897485455681713e-05,
+      "loss": 0.2516,
+      "step": 216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.896105151938872e-05,
+      "loss": 0.2612,
+      "step": 217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.894715814854145e-05,
+      "loss": 0.377,
+      "step": 218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8933174496653476e-05,
+      "loss": 0.2989,
+      "step": 219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.891910061644332e-05,
+      "loss": 0.307,
+      "step": 220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.890493656096965e-05,
+      "loss": 0.3227,
+      "step": 221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8890682383631115e-05,
+      "loss": 0.3266,
+      "step": 222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8876338138166106e-05,
+      "loss": 0.2701,
+      "step": 223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.886190387865258e-05,
+      "loss": 0.1774,
+      "step": 224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.884737965950786e-05,
+      "loss": 0.3749,
+      "step": 225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.8832765535488404e-05,
+      "loss": 0.3485,
+      "step": 226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.881806156168961e-05,
+      "loss": 0.3136,
+      "step": 227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.880326779354561e-05,
+      "loss": 0.3249,
+      "step": 228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.878838428682908e-05,
+      "loss": 0.2764,
+      "step": 229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.877341109765099e-05,
+      "loss": 0.2733,
+      "step": 230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.875834828246042e-05,
+      "loss": 0.3222,
+      "step": 231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.874319589804436e-05,
+      "loss": 0.2733,
+      "step": 232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.872795400152743e-05,
+      "loss": 0.2473,
+      "step": 233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.871262265037178e-05,
+      "loss": 0.2949,
+      "step": 234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.869720190237673e-05,
+      "loss": 0.1997,
+      "step": 235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.868169181567869e-05,
+      "loss": 0.3084,
+      "step": 236
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.866609244875084e-05,
+      "loss": 0.3271,
+      "step": 237
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.865040386040296e-05,
+      "loss": 0.3074,
+      "step": 238
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.86346261097812e-05,
+      "loss": 0.2595,
+      "step": 239
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.861875925636784e-05,
+      "loss": 0.2023,
+      "step": 240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8602803359981084e-05,
+      "loss": 0.2991,
+      "step": 241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.858675848077485e-05,
+      "loss": 0.2447,
+      "step": 242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.857062467923848e-05,
+      "loss": 0.33,
+      "step": 243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.855440201619659e-05,
+      "loss": 0.3225,
+      "step": 244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.853809055280879e-05,
+      "loss": 0.2931,
+      "step": 245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.852169035056947e-05,
+      "loss": 0.3839,
+      "step": 246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.850520147130756e-05,
+      "loss": 0.3014,
+      "step": 247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.848862397718632e-05,
+      "loss": 0.3673,
+      "step": 248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8471957930703074e-05,
+      "loss": 0.2698,
+      "step": 249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8455203394689e-05,
+      "loss": 0.3481,
+      "step": 250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.843836043230886e-05,
+      "loss": 0.3211,
+      "step": 251
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.842142910706083e-05,
+      "loss": 0.3669,
+      "step": 252
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.840440948277616e-05,
+      "loss": 0.3269,
+      "step": 253
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.838730162361903e-05,
+      "loss": 0.2302,
+      "step": 254
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8370105594086236e-05,
+      "loss": 0.3316,
+      "step": 255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.8352821459007005e-05,
+      "loss": 0.2298,
+      "step": 256
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.833544928354271e-05,
+      "loss": 0.2148,
+      "step": 257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8317989133186624e-05,
+      "loss": 0.2909,
+      "step": 258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8300441073763717e-05,
+      "loss": 0.2634,
+      "step": 259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8282805171430346e-05,
+      "loss": 0.3172,
+      "step": 260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.826508149267406e-05,
+      "loss": 0.2554,
+      "step": 261
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.82472701043133e-05,
+      "loss": 0.3023,
+      "step": 262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.822937107349721e-05,
+      "loss": 0.3491,
+      "step": 263
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8211384467705315e-05,
+      "loss": 0.259,
+      "step": 264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.819331035474731e-05,
+      "loss": 0.3185,
+      "step": 265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.81751488027628e-05,
+      "loss": 0.3491,
+      "step": 266
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8156899880221016e-05,
+      "loss": 0.4103,
+      "step": 267
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.813856365592061e-05,
+      "loss": 0.2375,
+      "step": 268
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.812014019898932e-05,
+      "loss": 0.3101,
+      "step": 269
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.810162957888379e-05,
+      "loss": 0.2408,
+      "step": 270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.808303186538925e-05,
+      "loss": 0.3545,
+      "step": 271
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.806434712861927e-05,
+      "loss": 0.2684,
+      "step": 272
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8045575439015514e-05,
+      "loss": 0.2132,
+      "step": 273
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8026716867347456e-05,
+      "loss": 0.2239,
+      "step": 274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.80077714847121e-05,
+      "loss": 0.2421,
+      "step": 275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.798873936253375e-05,
+      "loss": 0.3354,
+      "step": 276
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.79696205725637e-05,
+      "loss": 0.2413,
+      "step": 277
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.795041518688001e-05,
+      "loss": 0.2617,
+      "step": 278
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.793112327788716e-05,
+      "loss": 0.3021,
+      "step": 279
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.791174491831586e-05,
+      "loss": 0.3067,
+      "step": 280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.789228018122275e-05,
+      "loss": 0.386,
+      "step": 281
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.787272913999008e-05,
+      "loss": 0.3342,
+      "step": 282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7853091868325485e-05,
+      "loss": 0.3078,
+      "step": 283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.783336844026169e-05,
+      "loss": 0.2851,
+      "step": 284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.781355893015623e-05,
+      "loss": 0.2788,
+      "step": 285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.7793663412691164e-05,
+      "loss": 0.3291,
+      "step": 286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.777368196287281e-05,
+      "loss": 0.2228,
+      "step": 287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.775361465603143e-05,
+      "loss": 0.1885,
+      "step": 288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.773346156782101e-05,
+      "loss": 0.3855,
+      "step": 289
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.771322277421889e-05,
+      "loss": 0.2679,
+      "step": 290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.769289835152554e-05,
+      "loss": 0.285,
+      "step": 291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.767248837636425e-05,
+      "loss": 0.2531,
+      "step": 292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7651992925680846e-05,
+      "loss": 0.2426,
+      "step": 293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7631412076743386e-05,
+      "loss": 0.2714,
+      "step": 294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.761074590714189e-05,
+      "loss": 0.2188,
+      "step": 295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.7589994494788056e-05,
+      "loss": 0.2659,
+      "step": 296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.756915791791491e-05,
+      "loss": 0.3369,
+      "step": 297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.754823625507657e-05,
+      "loss": 0.2152,
+      "step": 298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.752722958514794e-05,
+      "loss": 0.3117,
+      "step": 299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7506137987324375e-05,
+      "loss": 0.2674,
+      "step": 300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.748496154112142e-05,
+      "loss": 0.2886,
+      "step": 301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.746370032637452e-05,
+      "loss": 0.2769,
+      "step": 302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.744235442323866e-05,
+      "loss": 0.2641,
+      "step": 303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.742092391218814e-05,
+      "loss": 0.2774,
+      "step": 304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.7399408874016206e-05,
+      "loss": 0.3404,
+      "step": 305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.737780938983478e-05,
+      "loss": 0.2063,
+      "step": 306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.735612554107415e-05,
+      "loss": 0.2887,
+      "step": 307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.733435740948265e-05,
+      "loss": 0.4375,
+      "step": 308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.731250507712637e-05,
+      "loss": 0.2218,
+      "step": 309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7290568626388844e-05,
+      "loss": 0.3174,
+      "step": 310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.726854813997071e-05,
+      "loss": 0.3613,
+      "step": 311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7246443700889453e-05,
+      "loss": 0.2768,
+      "step": 312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.722425539247902e-05,
+      "loss": 0.29,
+      "step": 313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.7201983298389595e-05,
+      "loss": 0.2188,
+      "step": 314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.717962750258719e-05,
+      "loss": 0.2425,
+      "step": 315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.71571880893534e-05,
+      "loss": 0.2412,
+      "step": 316
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.713466514328505e-05,
+      "loss": 0.2124,
+      "step": 317
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7112058749293894e-05,
+      "loss": 0.3265,
+      "step": 318
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.708936899260626e-05,
+      "loss": 0.1941,
+      "step": 319
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7066595958762794e-05,
+      "loss": 0.2355,
+      "step": 320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7043739733618066e-05,
+      "loss": 0.304,
+      "step": 321
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.702080040334027e-05,
+      "loss": 0.3028,
+      "step": 322
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6997778054410956e-05,
+      "loss": 0.2999,
+      "step": 323
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6974672773624606e-05,
+      "loss": 0.3101,
+      "step": 324
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.695148464808837e-05,
+      "loss": 0.379,
+      "step": 325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.692821376522174e-05,
+      "loss": 0.2935,
+      "step": 326
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.690486021275619e-05,
+      "loss": 0.2777,
+      "step": 327
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.688142407873485e-05,
+      "loss": 0.2829,
+      "step": 328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6857905451512205e-05,
+      "loss": 0.2161,
+      "step": 329
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.683430441975373e-05,
+      "loss": 0.286,
+      "step": 330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6810621072435553e-05,
+      "loss": 0.298,
+      "step": 331
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6786855498844165e-05,
+      "loss": 0.2971,
+      "step": 332
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.676300778857601e-05,
+      "loss": 0.4267,
+      "step": 333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6739078031537226e-05,
+      "loss": 0.4661,
+      "step": 334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6715066317943246e-05,
+      "loss": 0.2975,
+      "step": 335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.669097273831847e-05,
+      "loss": 0.3425,
+      "step": 336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.666679738349597e-05,
+      "loss": 0.2916,
+      "step": 337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.664254034461709e-05,
+      "loss": 0.2832,
+      "step": 338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6618201713131116e-05,
+      "loss": 0.3136,
+      "step": 339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.659378158079495e-05,
+      "loss": 0.3325,
+      "step": 340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6569280039672747e-05,
+      "loss": 0.3296,
+      "step": 341
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.654469718213558e-05,
+      "loss": 0.2936,
+      "step": 342
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6520033100861084e-05,
+      "loss": 0.2644,
+      "step": 343
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.649528788883311e-05,
+      "loss": 0.2932,
+      "step": 344
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6470461639341365e-05,
+      "loss": 0.376,
+      "step": 345
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.644555444598107e-05,
+      "loss": 0.3354,
+      "step": 346
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6420566402652623e-05,
+      "loss": 0.2584,
+      "step": 347
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.63954976035612e-05,
+      "loss": 0.2196,
+      "step": 348
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6370348143216436e-05,
+      "loss": 0.3061,
+      "step": 349
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6345118116432085e-05,
+      "loss": 0.3528,
+      "step": 350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.63198076183256e-05,
+      "loss": 0.2717,
+      "step": 351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.629441674431783e-05,
+      "loss": 0.4058,
+      "step": 352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.626894559013263e-05,
+      "loss": 0.3199,
+      "step": 353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6243394251796544e-05,
+      "loss": 0.362,
+      "step": 354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.621776282563838e-05,
+      "loss": 0.2788,
+      "step": 355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6192051408288875e-05,
+      "loss": 0.2861,
+      "step": 356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.616626009668036e-05,
+      "loss": 0.2636,
+      "step": 357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6140388988046345e-05,
+      "loss": 0.2872,
+      "step": 358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.611443817992119e-05,
+      "loss": 0.2773,
+      "step": 359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.6088407770139726e-05,
+      "loss": 0.2935,
+      "step": 360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.606229785683686e-05,
+      "loss": 0.3182,
+      "step": 361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.603610853844724e-05,
+      "loss": 0.2632,
+      "step": 362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.600983991370489e-05,
+      "loss": 0.2177,
+      "step": 363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5983492081642784e-05,
+      "loss": 0.3008,
+      "step": 364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.595706514159255e-05,
+      "loss": 0.2536,
+      "step": 365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.593055919318402e-05,
+      "loss": 0.2911,
+      "step": 366
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.590397433634491e-05,
+      "loss": 0.2796,
+      "step": 367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5877310671300414e-05,
+      "loss": 0.2053,
+      "step": 368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.585056829857281e-05,
+      "loss": 0.3824,
+      "step": 369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.582374731898115e-05,
+      "loss": 0.3492,
+      "step": 370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.579684783364081e-05,
+      "loss": 0.2639,
+      "step": 371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.576986994396313e-05,
+      "loss": 0.2315,
+      "step": 372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5742813751655046e-05,
+      "loss": 0.2623,
+      "step": 373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.571567935871868e-05,
+      "loss": 0.2882,
+      "step": 374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.568846686745098e-05,
+      "loss": 0.1957,
+      "step": 375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5661176380443354e-05,
+      "loss": 0.2981,
+      "step": 376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.56338080005812e-05,
+      "loss": 0.3681,
+      "step": 377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.560636183104361e-05,
+      "loss": 0.2598,
+      "step": 378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.557883797530295e-05,
+      "loss": 0.3944,
+      "step": 379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.555123653712441e-05,
+      "loss": 0.3041,
+      "step": 380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.552355762056575e-05,
+      "loss": 0.2686,
+      "step": 381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.549580132997675e-05,
+      "loss": 0.3159,
+      "step": 382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.546796776999893e-05,
+      "loss": 0.1752,
+      "step": 383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.54400570455651e-05,
+      "loss": 0.2823,
+      "step": 384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5412069261899e-05,
+      "loss": 0.2172,
+      "step": 385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5384004524514875e-05,
+      "loss": 0.3612,
+      "step": 386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.535586293921707e-05,
+      "loss": 0.3427,
+      "step": 387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.532764461209965e-05,
+      "loss": 0.3054,
+      "step": 388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.529934964954603e-05,
+      "loss": 0.2617,
+      "step": 389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.527097815822852e-05,
+      "loss": 0.2361,
+      "step": 390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.524253024510793e-05,
+      "loss": 0.2966,
+      "step": 391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5214006017433206e-05,
+      "loss": 0.207,
+      "step": 392
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.518540558274098e-05,
+      "loss": 0.307,
+      "step": 393
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.51567290488552e-05,
+      "loss": 0.2536,
+      "step": 394
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5127976523886707e-05,
+      "loss": 0.1883,
+      "step": 395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.509914811623282e-05,
+      "loss": 0.7126,
+      "step": 396
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.507024393457693e-05,
+      "loss": 0.2616,
+      "step": 397
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.504126408788814e-05,
+      "loss": 0.234,
+      "step": 398
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.501220868542075e-05,
+      "loss": 0.3386,
+      "step": 399
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.498307783671395e-05,
+      "loss": 0.2801,
+      "step": 400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.4953871651591354e-05,
+      "loss": 0.3349,
+      "step": 401
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.492459024016058e-05,
+      "loss": 0.4669,
+      "step": 402
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.4895233712812866e-05,
+      "loss": 0.1998,
+      "step": 403
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.486580218022263e-05,
+      "loss": 0.4003,
+      "step": 404
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.483629575334707e-05,
+      "loss": 0.3852,
+      "step": 405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.480671454342571e-05,
+      "loss": 0.2936,
+      "step": 406
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.477705866198004e-05,
+      "loss": 0.2771,
+      "step": 407
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.4747328220813054e-05,
+      "loss": 0.2904,
+      "step": 408
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.471752333200882e-05,
+      "loss": 0.2923,
+      "step": 409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.468764410793208e-05,
+      "loss": 0.3194,
+      "step": 410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.465769066122784e-05,
+      "loss": 0.248,
+      "step": 411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.46276631048209e-05,
+      "loss": 0.4047,
+      "step": 412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.459756155191549e-05,
+      "loss": 0.2548,
+      "step": 413
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.4567386115994756e-05,
+      "loss": 0.3067,
+      "step": 414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.4537136910820426e-05,
+      "loss": 0.2111,
+      "step": 415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.450681405043232e-05,
+      "loss": 0.221,
+      "step": 416
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.447641764914796e-05,
+      "loss": 0.2154,
+      "step": 417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4445947821562087e-05,
+      "loss": 0.4225,
+      "step": 418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.441540468254626e-05,
+      "loss": 0.2937,
+      "step": 419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.438478834724847e-05,
+      "loss": 0.2467,
+      "step": 420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.43540989310926e-05,
+      "loss": 0.2943,
+      "step": 421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.432333654977809e-05,
+      "loss": 0.331,
+      "step": 422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.429250131927945e-05,
+      "loss": 0.2773,
+      "step": 423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.426159335584581e-05,
+      "loss": 0.262,
+      "step": 424
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.423061277600053e-05,
+      "loss": 0.3498,
+      "step": 425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.4199559696540716e-05,
+      "loss": 0.4043,
+      "step": 426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4168434234536825e-05,
+      "loss": 0.2111,
+      "step": 427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.413723650733216e-05,
+      "loss": 0.3302,
+      "step": 428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.4105966632542495e-05,
+      "loss": 0.3073,
+      "step": 429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.407462472805559e-05,
+      "loss": 0.2049,
+      "step": 430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.404321091203076e-05,
+      "loss": 0.2015,
+      "step": 431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.401172530289842e-05,
+      "loss": 0.262,
+      "step": 432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.3980168019359645e-05,
+      "loss": 0.3501,
+      "step": 433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.3948539180385744e-05,
+      "loss": 0.2372,
+      "step": 434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.391683890521777e-05,
+      "loss": 0.307,
+      "step": 435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.388506731336609e-05,
+      "loss": 0.2429,
+      "step": 436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.385322452460995e-05,
+      "loss": 0.2586,
+      "step": 437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.382131065899701e-05,
+      "loss": 0.278,
+      "step": 438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.378932583684286e-05,
+      "loss": 0.4167,
+      "step": 439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.375727017873065e-05,
+      "loss": 0.3226,
+      "step": 440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.372514380551052e-05,
+      "loss": 0.2661,
+      "step": 441
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.369294683829926e-05,
+      "loss": 0.3259,
+      "step": 442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.366067939847977e-05,
+      "loss": 0.2947,
+      "step": 443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.362834160770064e-05,
+      "loss": 0.4092,
+      "step": 444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.359593358787569e-05,
+      "loss": 0.2933,
+      "step": 445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.35634554611835e-05,
+      "loss": 0.3961,
+      "step": 446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.353090735006695e-05,
+      "loss": 0.298,
+      "step": 447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.349828937723276e-05,
+      "loss": 0.3062,
+      "step": 448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.346560166565105e-05,
+      "loss": 0.3218,
+      "step": 449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.343284433855481e-05,
+      "loss": 0.3342,
+      "step": 450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.340001751943954e-05,
+      "loss": 0.37,
+      "step": 451
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.336712133206269e-05,
+      "loss": 0.3375,
+      "step": 452
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.333415590044323e-05,
+      "loss": 0.2818,
+      "step": 453
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.330112134886119e-05,
+      "loss": 0.3518,
+      "step": 454
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3268017801857176e-05,
+      "loss": 0.2463,
+      "step": 455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.323484538423192e-05,
+      "loss": 0.2906,
+      "step": 456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.320160422104579e-05,
+      "loss": 0.2432,
+      "step": 457
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.316829443761832e-05,
+      "loss": 0.2806,
+      "step": 458
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3134916159527735e-05,
+      "loss": 0.1853,
+      "step": 459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3101469512610525e-05,
+      "loss": 0.2855,
+      "step": 460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.306795462296088e-05,
+      "loss": 0.3396,
+      "step": 461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.303437161693031e-05,
+      "loss": 0.3534,
+      "step": 462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3000720621127096e-05,
+      "loss": 0.3458,
+      "step": 463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.296700176241584e-05,
+      "loss": 0.2457,
+      "step": 464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.293321516791701e-05,
+      "loss": 0.2472,
+      "step": 465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2899360965006424e-05,
+      "loss": 0.4089,
+      "step": 466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2865439281314785e-05,
+      "loss": 0.3292,
+      "step": 467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.28314502447272e-05,
+      "loss": 0.2968,
+      "step": 468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2797393983382706e-05,
+      "loss": 0.3879,
+      "step": 469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.276327062567376e-05,
+      "loss": 0.2874,
+      "step": 470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.272908030024578e-05,
+      "loss": 0.3083,
+      "step": 471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.269482313599666e-05,
+      "loss": 0.224,
+      "step": 472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.266049926207629e-05,
+      "loss": 0.2349,
+      "step": 473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.262610880788602e-05,
+      "loss": 0.2517,
+      "step": 474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.259165190307823e-05,
+      "loss": 0.3074,
+      "step": 475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.255712867755583e-05,
+      "loss": 0.2581,
+      "step": 476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.252253926147174e-05,
+      "loss": 0.2062,
+      "step": 477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.248788378522841e-05,
+      "loss": 0.205,
+      "step": 478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.245316237947738e-05,
+      "loss": 0.2171,
+      "step": 479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2418375175118694e-05,
+      "loss": 0.3009,
+      "step": 480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2383522303300515e-05,
+      "loss": 0.2514,
+      "step": 481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.23486038954185e-05,
+      "loss": 0.1759,
+      "step": 482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.231362008311545e-05,
+      "loss": 0.1742,
+      "step": 483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.227857099828069e-05,
+      "loss": 0.3008,
+      "step": 484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.224345677304965e-05,
+      "loss": 0.3944,
+      "step": 485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2208277539803334e-05,
+      "loss": 0.2892,
+      "step": 486
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.217303343116782e-05,
+      "loss": 0.3013,
+      "step": 487
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.213772458001378e-05,
+      "loss": 0.3311,
+      "step": 488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.210235111945594e-05,
+      "loss": 0.4286,
+      "step": 489
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.206691318285265e-05,
+      "loss": 0.22,
+      "step": 490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.203141090380528e-05,
+      "loss": 0.2426,
+      "step": 491
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.199584441615782e-05,
+      "loss": 0.3117,
+      "step": 492
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.19602138539963e-05,
+      "loss": 0.4433,
+      "step": 493
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.19245193516483e-05,
+      "loss": 0.4025,
+      "step": 494
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.188876104368251e-05,
+      "loss": 0.2673,
+      "step": 495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.18529390649081e-05,
+      "loss": 0.3724,
+      "step": 496
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.181705355037433e-05,
+      "loss": 0.2437,
+      "step": 497
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.178110463536995e-05,
+      "loss": 0.2298,
+      "step": 498
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.174509245542277e-05,
+      "loss": 0.3817,
+      "step": 499
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.1709017146299087e-05,
+      "loss": 0.2266,
+      "step": 500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.167287884400319e-05,
+      "loss": 0.2456,
+      "step": 501
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.163667768477688e-05,
+      "loss": 0.3525,
+      "step": 502
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.160041380509889e-05,
+      "loss": 0.2361,
+      "step": 503
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.156408734168445e-05,
+      "loss": 0.2911,
+      "step": 504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.152769843148471e-05,
+      "loss": 0.2309,
+      "step": 505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.149124721168625e-05,
+      "loss": 0.2448,
+      "step": 506
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.145473381971054e-05,
+      "loss": 0.2787,
+      "step": 507
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.141815839321347e-05,
+      "loss": 0.3903,
+      "step": 508
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.138152107008478e-05,
+      "loss": 0.2205,
+      "step": 509
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.134482198844758e-05,
+      "loss": 0.3232,
+      "step": 510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.130806128665779e-05,
+      "loss": 0.2219,
+      "step": 511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.1271239103303636e-05,
+      "loss": 0.2765,
+      "step": 512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.1234355577205164e-05,
+      "loss": 0.2903,
+      "step": 513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.119741084741363e-05,
+      "loss": 0.2991,
+      "step": 514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.1160405053211084e-05,
+      "loss": 0.2697,
+      "step": 515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.112333833410975e-05,
+      "loss": 0.1415,
+      "step": 516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.108621082985155e-05,
+      "loss": 0.2174,
+      "step": 517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.104902268040758e-05,
+      "loss": 0.394,
+      "step": 518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.101177402597755e-05,
+      "loss": 0.2551,
+      "step": 519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.097446500698929e-05,
+      "loss": 0.2233,
+      "step": 520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.09370957640982e-05,
+      "loss": 0.2172,
+      "step": 521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.089966643818671e-05,
+      "loss": 0.2568,
+      "step": 522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.0862177170363784e-05,
+      "loss": 0.2724,
+      "step": 523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.0824628101964354e-05,
+      "loss": 0.2839,
+      "step": 524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.078701937454883e-05,
+      "loss": 0.2476,
+      "step": 525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.074935112990249e-05,
+      "loss": 0.2791,
+      "step": 526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.071162351003502e-05,
+      "loss": 0.2558,
+      "step": 527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.0673836657179953e-05,
+      "loss": 0.2586,
+      "step": 528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.0635990713794124e-05,
+      "loss": 0.2541,
+      "step": 529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.059808582255715e-05,
+      "loss": 0.2765,
+      "step": 530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.0560122126370856e-05,
+      "loss": 0.2609,
+      "step": 531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.052209976835879e-05,
+      "loss": 0.236,
+      "step": 532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.048401889186565e-05,
+      "loss": 0.1511,
+      "step": 533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.0445879640456744e-05,
+      "loss": 0.2454,
+      "step": 534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.040768215791745e-05,
+      "loss": 0.2546,
+      "step": 535
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.036942658825267e-05,
+      "loss": 0.2466,
+      "step": 536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.0331113075686344e-05,
+      "loss": 0.2293,
+      "step": 537
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.029274176466079e-05,
+      "loss": 0.2017,
+      "step": 538
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.025431279983627e-05,
+      "loss": 0.3737,
+      "step": 539
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.021582632609039e-05,
+      "loss": 0.1394,
+      "step": 540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.017728248851756e-05,
+      "loss": 0.244,
+      "step": 541
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.013868143242847e-05,
+      "loss": 0.3704,
+      "step": 542
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.01000233033495e-05,
+      "loss": 0.3224,
+      "step": 543
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.00613082470222e-05,
+      "loss": 0.252,
+      "step": 544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.002253640940277e-05,
+      "loss": 0.2751,
+      "step": 545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.998370793666142e-05,
+      "loss": 0.2885,
+      "step": 546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.994482297518192e-05,
+      "loss": 0.2324,
+      "step": 547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9905881671560994e-05,
+      "loss": 0.2592,
+      "step": 548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.986688417260776e-05,
+      "loss": 0.4748,
+      "step": 549
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.982783062534321e-05,
+      "loss": 0.4343,
+      "step": 550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.978872117699962e-05,
+      "loss": 0.2425,
+      "step": 551
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.974955597502004e-05,
+      "loss": 0.3459,
+      "step": 552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.971033516705769e-05,
+      "loss": 0.324,
+      "step": 553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9671058900975446e-05,
+      "loss": 0.1953,
+      "step": 554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.963172732484522e-05,
+      "loss": 0.2776,
+      "step": 555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.95923405869475e-05,
+      "loss": 0.2916,
+      "step": 556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.95528988357707e-05,
+      "loss": 0.2366,
+      "step": 557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9513402220010634e-05,
+      "loss": 0.2539,
+      "step": 558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.947385088856996e-05,
+      "loss": 0.3312,
+      "step": 559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.943424499055763e-05,
+      "loss": 0.2937,
+      "step": 560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9394584675288296e-05,
+      "loss": 0.2306,
+      "step": 561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.935487009228176e-05,
+      "loss": 0.2333,
+      "step": 562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.931510139126243e-05,
+      "loss": 0.2186,
+      "step": 563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9275278722158735e-05,
+      "loss": 0.2813,
+      "step": 564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9235402235102545e-05,
+      "loss": 0.2282,
+      "step": 565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.919547208042866e-05,
+      "loss": 0.3022,
+      "step": 566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.915548840867418e-05,
+      "loss": 0.2804,
+      "step": 567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.911545137057796e-05,
+      "loss": 0.306,
+      "step": 568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.907536111708008e-05,
+      "loss": 0.2364,
+      "step": 569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.903521779932121e-05,
+      "loss": 0.2579,
+      "step": 570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.899502156864209e-05,
+      "loss": 0.3072,
+      "step": 571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.895477257658292e-05,
+      "loss": 0.2572,
+      "step": 572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8914470974882846e-05,
+      "loss": 0.2716,
+      "step": 573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.887411691547933e-05,
+      "loss": 0.2007,
+      "step": 574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8833710550507605e-05,
+      "loss": 0.2523,
+      "step": 575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.87932520323001e-05,
+      "loss": 0.1348,
+      "step": 576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875274151338584e-05,
+      "loss": 0.4596,
+      "step": 577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.871217914648994e-05,
+      "loss": 0.3232,
+      "step": 578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.867156508453293e-05,
+      "loss": 0.332,
+      "step": 579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.863089948063027e-05,
+      "loss": 0.3265,
+      "step": 580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859018248809172e-05,
+      "loss": 0.2594,
+      "step": 581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854941426042076e-05,
+      "loss": 0.233,
+      "step": 582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.850859495131405e-05,
+      "loss": 0.2867,
+      "step": 583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.846772471466081e-05,
+      "loss": 0.2477,
+      "step": 584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.84268037045423e-05,
+      "loss": 0.2583,
+      "step": 585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8385832075231124e-05,
+      "loss": 0.2449,
+      "step": 586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.834480998119078e-05,
+      "loss": 0.2729,
+      "step": 587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.830373757707498e-05,
+      "loss": 0.3224,
+      "step": 588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.826261501772715e-05,
+      "loss": 0.2189,
+      "step": 589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.822144245817976e-05,
+      "loss": 0.3783,
+      "step": 590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.818022005365379e-05,
+      "loss": 0.2048,
+      "step": 591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.813894795955817e-05,
+      "loss": 0.2587,
+      "step": 592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8097626331489106e-05,
+      "loss": 0.3626,
+      "step": 593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.805625532522959e-05,
+      "loss": 0.1753,
+      "step": 594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.801483509674875e-05,
+      "loss": 0.268,
+      "step": 595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.797336580220129e-05,
+      "loss": 0.2268,
+      "step": 596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.793184759792688e-05,
+      "loss": 0.2217,
+      "step": 597
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7890280640449605e-05,
+      "loss": 0.3057,
+      "step": 598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7848665086477334e-05,
+      "loss": 0.1427,
+      "step": 599
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.780700109290115e-05,
+      "loss": 0.3914,
+      "step": 600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.776528881679474e-05,
+      "loss": 0.2081,
+      "step": 601
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.772352841541384e-05,
+      "loss": 0.1958,
+      "step": 602
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.76817200461956e-05,
+      "loss": 0.2173,
+      "step": 603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.763986386675802e-05,
+      "loss": 0.3855,
+      "step": 604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7597960034899335e-05,
+      "loss": 0.3793,
+      "step": 605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7556008708597446e-05,
+      "loss": 0.3456,
+      "step": 606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.751401004600929e-05,
+      "loss": 0.2731,
+      "step": 607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.747196420547029e-05,
+      "loss": 0.2725,
+      "step": 608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.742987134549369e-05,
+      "loss": 0.3186,
+      "step": 609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.738773162477001e-05,
+      "loss": 0.2341,
+      "step": 610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7345545202166476e-05,
+      "loss": 0.4402,
+      "step": 611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.730331223672633e-05,
+      "loss": 0.2449,
+      "step": 612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.726103288766832e-05,
+      "loss": 0.2957,
+      "step": 613
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.721870731438601e-05,
+      "loss": 0.2161,
+      "step": 614
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.717633567644729e-05,
+      "loss": 0.2128,
+      "step": 615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.713391813359366e-05,
+      "loss": 0.2113,
+      "step": 616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7091454845739734e-05,
+      "loss": 0.2892,
+      "step": 617
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.704894597297254e-05,
+      "loss": 0.2417,
+      "step": 618
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.700639167555098e-05,
+      "loss": 0.3562,
+      "step": 619
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.696379211390522e-05,
+      "loss": 0.1681,
+      "step": 620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6921147448636045e-05,
+      "loss": 0.3111,
+      "step": 621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6878457840514316e-05,
+      "loss": 0.3029,
+      "step": 622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6835723450480283e-05,
+      "loss": 0.1842,
+      "step": 623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.679294443964307e-05,
+      "loss": 0.2543,
+      "step": 624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.675012096928e-05,
+      "loss": 0.3299,
+      "step": 625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.670725320083601e-05,
+      "loss": 0.2311,
+      "step": 626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6664341295923035e-05,
+      "loss": 0.2829,
+      "step": 627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.6621385416319425e-05,
+      "loss": 0.2554,
+      "step": 628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.657838572396929e-05,
+      "loss": 0.3379,
+      "step": 629
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.653534238098194e-05,
+      "loss": 0.1747,
+      "step": 630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.649225554963123e-05,
+      "loss": 0.3012,
+      "step": 631
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.644912539235496e-05,
+      "loss": 0.2631,
+      "step": 632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.64059520717543e-05,
+      "loss": 0.3087,
+      "step": 633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.636273575059312e-05,
+      "loss": 0.2936,
+      "step": 634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.631947659179741e-05,
+      "loss": 0.3079,
+      "step": 635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.627617475845466e-05,
+      "loss": 0.1422,
+      "step": 636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.6232830413813235e-05,
+      "loss": 0.2727,
+      "step": 637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.6189443721281786e-05,
+      "loss": 0.2239,
+      "step": 638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.614601484442859e-05,
+      "loss": 0.2528,
+      "step": 639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.6102543946980985e-05,
+      "loss": 0.2446,
+      "step": 640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.605903119282472e-05,
+      "loss": 0.2392,
+      "step": 641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.601547674600333e-05,
+      "loss": 0.2591,
+      "step": 642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.5971880770717554e-05,
+      "loss": 0.28,
+      "step": 643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.592824343132467e-05,
+      "loss": 0.4283,
+      "step": 644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.588456489233792e-05,
+      "loss": 0.2747,
+      "step": 645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.5840845318425855e-05,
+      "loss": 0.2312,
+      "step": 646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.5797084874411735e-05,
+      "loss": 0.2521,
+      "step": 647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.57532837252729e-05,
+      "loss": 0.262,
+      "step": 648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.570944203614014e-05,
+      "loss": 0.1989,
+      "step": 649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.56655599722971e-05,
+      "loss": 0.2692,
+      "step": 650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.562163769917963e-05,
+      "loss": 0.2294,
+      "step": 651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.5577675382375157e-05,
+      "loss": 0.2673,
+      "step": 652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.5533673187622075e-05,
+      "loss": 0.3452,
+      "step": 653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.548963128080915e-05,
+      "loss": 0.2076,
+      "step": 654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.544554982797481e-05,
+      "loss": 0.2218,
+      "step": 655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.540142899530662e-05,
+      "loss": 0.254,
+      "step": 656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.535726894914058e-05,
+      "loss": 0.1941,
+      "step": 657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.531306985596055e-05,
+      "loss": 0.2272,
+      "step": 658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.526883188239755e-05,
+      "loss": 0.3485,
+      "step": 659
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.5224555195229234e-05,
+      "loss": 0.1742,
+      "step": 660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.518023996137918e-05,
+      "loss": 0.2912,
+      "step": 661
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.513588634791628e-05,
+      "loss": 0.1661,
+      "step": 662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.509149452205414e-05,
+      "loss": 0.2866,
+      "step": 663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.504706465115042e-05,
+      "loss": 0.3944,
+      "step": 664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.500259690270618e-05,
+      "loss": 0.2728,
+      "step": 665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.495809144436533e-05,
+      "loss": 0.2458,
+      "step": 666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.491354844391391e-05,
+      "loss": 0.152,
+      "step": 667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.486896806927951e-05,
+      "loss": 0.2026,
+      "step": 668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.48243504885306e-05,
+      "loss": 0.2478,
+      "step": 669
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.477969586987596e-05,
+      "loss": 0.2907,
+      "step": 670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.4735004381663955e-05,
+      "loss": 0.2569,
+      "step": 671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.4690276192381975e-05,
+      "loss": 0.3092,
+      "step": 672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.464551147065578e-05,
+      "loss": 0.1909,
+      "step": 673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.460071038524883e-05,
+      "loss": 0.2399,
+      "step": 674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.455587310506171e-05,
+      "loss": 0.2615,
+      "step": 675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.451099979913143e-05,
+      "loss": 0.3121,
+      "step": 676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.446609063663084e-05,
+      "loss": 0.2779,
+      "step": 677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.442114578686795e-05,
+      "loss": 0.2808,
+      "step": 678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.4376165419285336e-05,
+      "loss": 0.271,
+      "step": 679
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.433114970345944e-05,
+      "loss": 0.2158,
+      "step": 680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4286098809100005e-05,
+      "loss": 0.2295,
+      "step": 681
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.424101290604938e-05,
+      "loss": 0.3262,
+      "step": 682
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.41958921642819e-05,
+      "loss": 0.2281,
+      "step": 683
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.415073675390322e-05,
+      "loss": 0.3043,
+      "step": 684
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.410554684514975e-05,
+      "loss": 0.2987,
+      "step": 685
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.4060322608387896e-05,
+      "loss": 0.2203,
+      "step": 686
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.401506421411354e-05,
+      "loss": 0.252,
+      "step": 687
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3969771832951284e-05,
+      "loss": 0.2814,
+      "step": 688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.39244456356539e-05,
+      "loss": 0.2556,
+      "step": 689
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.387908579310164e-05,
+      "loss": 0.2388,
+      "step": 690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3833692476301574e-05,
+      "loss": 0.3186,
+      "step": 691
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3788265856387e-05,
+      "loss": 0.2689,
+      "step": 692
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.374280610461675e-05,
+      "loss": 0.3176,
+      "step": 693
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3697313392374585e-05,
+      "loss": 0.2756,
+      "step": 694
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.365178789116849e-05,
+      "loss": 0.2494,
+      "step": 695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.3606229772630125e-05,
+      "loss": 0.2898,
+      "step": 696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.356063920851405e-05,
+      "loss": 0.2514,
+      "step": 697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.351501637069719e-05,
+      "loss": 0.3513,
+      "step": 698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.346936143117811e-05,
+      "loss": 0.2898,
+      "step": 699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3423674562076446e-05,
+      "loss": 0.1813,
+      "step": 700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.337795593563217e-05,
+      "loss": 0.1925,
+      "step": 701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.333220572420497e-05,
+      "loss": 0.268,
+      "step": 702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.328642410027366e-05,
+      "loss": 0.2425,
+      "step": 703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3240611236435415e-05,
+      "loss": 0.273,
+      "step": 704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.319476730540525e-05,
+      "loss": 0.2221,
+      "step": 705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.314889248001525e-05,
+      "loss": 0.1541,
+      "step": 706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3102986933214e-05,
+      "loss": 0.2146,
+      "step": 707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.305705083806592e-05,
+      "loss": 0.2487,
+      "step": 708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3011084367750544e-05,
+      "loss": 0.2341,
+      "step": 709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.296508769556198e-05,
+      "loss": 0.1824,
+      "step": 710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.291906099490815e-05,
+      "loss": 0.2224,
+      "step": 711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2873004439310216e-05,
+      "loss": 0.252,
+      "step": 712
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.282691820240188e-05,
+      "loss": 0.3569,
+      "step": 713
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.278080245792875e-05,
+      "loss": 0.2553,
+      "step": 714
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.273465737974767e-05,
+      "loss": 0.397,
+      "step": 715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.268848314182606e-05,
+      "loss": 0.2171,
+      "step": 716
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.264227991824131e-05,
+      "loss": 0.3236,
+      "step": 717
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.259604788318005e-05,
+      "loss": 0.2252,
+      "step": 718
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.254978721093755e-05,
+      "loss": 0.2994,
+      "step": 719
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.250349807591704e-05,
+      "loss": 0.2406,
+      "step": 720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.245718065262904e-05,
+      "loss": 0.2831,
+      "step": 721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2410835115690735e-05,
+      "loss": 0.175,
+      "step": 722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.236446163982528e-05,
+      "loss": 0.2545,
+      "step": 723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.23180603998612e-05,
+      "loss": 0.2398,
+      "step": 724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.227163157073163e-05,
+      "loss": 0.2577,
+      "step": 725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.222517532747377e-05,
+      "loss": 0.2555,
+      "step": 726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2178691845228136e-05,
+      "loss": 0.2365,
+      "step": 727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2132181299237975e-05,
+      "loss": 0.3234,
+      "step": 728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.2085643864848504e-05,
+      "loss": 0.223,
+      "step": 729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.203907971750637e-05,
+      "loss": 0.321,
+      "step": 730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1992489032758897e-05,
+      "loss": 0.2785,
+      "step": 731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.194587198625347e-05,
+      "loss": 0.2555,
+      "step": 732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.189922875373684e-05,
+      "loss": 0.2354,
+      "step": 733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.1852559511054484e-05,
+      "loss": 0.2835,
+      "step": 734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.180586443414996e-05,
+      "loss": 0.2872,
+      "step": 735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.175914369906418e-05,
+      "loss": 0.2141,
+      "step": 736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.171239748193483e-05,
+      "loss": 0.228,
+      "step": 737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.166562595899565e-05,
+      "loss": 0.143,
+      "step": 738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.161882930657576e-05,
+      "loss": 0.2461,
+      "step": 739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.157200770109905e-05,
+      "loss": 0.2437,
+      "step": 740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.152516131908347e-05,
+      "loss": 0.2225,
+      "step": 741
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.147829033714038e-05,
+      "loss": 0.3087,
+      "step": 742
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.143139493197387e-05,
+      "loss": 0.1815,
+      "step": 743
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.138447528038011e-05,
+      "loss": 0.3681,
+      "step": 744
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.133753155924671e-05,
+      "loss": 0.1193,
+      "step": 745
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.129056394555197e-05,
+      "loss": 0.2512,
+      "step": 746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1243572616364324e-05,
+      "loss": 0.4096,
+      "step": 747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1196557748841555e-05,
+      "loss": 0.1937,
+      "step": 748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1149519520230224e-05,
+      "loss": 0.2525,
+      "step": 749
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1102458107864944e-05,
+      "loss": 0.2583,
+      "step": 750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.105537368916775e-05,
+      "loss": 0.2784,
+      "step": 751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.10082664416474e-05,
+      "loss": 0.2542,
+      "step": 752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.096113654289873e-05,
+      "loss": 0.3522,
+      "step": 753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.091398417060193e-05,
+      "loss": 0.3038,
+      "step": 754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.086680950252196e-05,
+      "loss": 0.2119,
+      "step": 755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0819612716507825e-05,
+      "loss": 0.2564,
+      "step": 756
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.077239399049189e-05,
+      "loss": 0.121,
+      "step": 757
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0725153502489285e-05,
+      "loss": 0.3507,
+      "step": 758
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.067789143059713e-05,
+      "loss": 0.1827,
+      "step": 759
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.063060795299394e-05,
+      "loss": 0.2606,
+      "step": 760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.058330324793894e-05,
+      "loss": 0.2924,
+      "step": 761
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.053597749377135e-05,
+      "loss": 0.2594,
+      "step": 762
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0488630868909786e-05,
+      "loss": 0.2064,
+      "step": 763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.044126355185151e-05,
+      "loss": 0.3151,
+      "step": 764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.0393875721171832e-05,
+      "loss": 0.3077,
+      "step": 765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.034646755552336e-05,
+      "loss": 0.2053,
+      "step": 766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.029903923363539e-05,
+      "loss": 0.1773,
+      "step": 767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.0251590934313197e-05,
+      "loss": 0.1823,
+      "step": 768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.0204122836437366e-05,
+      "loss": 0.2068,
+      "step": 769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.0156635118963146e-05,
+      "loss": 0.185,
+      "step": 770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.0109127960919725e-05,
+      "loss": 0.2399,
+      "step": 771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.0061601541409595e-05,
+      "loss": 0.2785,
+      "step": 772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0014056039607875e-05,
+      "loss": 0.3422,
+      "step": 773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.996649163476159e-05,
+      "loss": 0.2575,
+      "step": 774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.991890850618907e-05,
+      "loss": 0.2477,
+      "step": 775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9871306833279205e-05,
+      "loss": 0.2519,
+      "step": 776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.982368679549081e-05,
+      "loss": 0.2906,
+      "step": 777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.977604857235194e-05,
+      "loss": 0.2857,
+      "step": 778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9728392343459205e-05,
+      "loss": 0.2739,
+      "step": 779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9680718288477078e-05,
+      "loss": 0.18,
+      "step": 780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.963302658713726e-05,
+      "loss": 0.361,
+      "step": 781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9585317419237977e-05,
+      "loss": 0.273,
+      "step": 782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.953759096464329e-05,
+      "loss": 0.2964,
+      "step": 783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9489847403282454e-05,
+      "loss": 0.1882,
+      "step": 784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9442086915149198e-05,
+      "loss": 0.2735,
+      "step": 785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9394309680301063e-05,
+      "loss": 0.1811,
+      "step": 786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9346515878858728e-05,
+      "loss": 0.1866,
+      "step": 787
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.929870569100534e-05,
+      "loss": 0.3179,
+      "step": 788
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9250879296985822e-05,
+      "loss": 0.2438,
+      "step": 789
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9203036877106178e-05,
+      "loss": 0.2603,
+      "step": 790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.915517861173284e-05,
+      "loss": 0.2509,
+      "step": 791
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9107304681291987e-05,
+      "loss": 0.1607,
+      "step": 792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9059415266268835e-05,
+      "loss": 0.2151,
+      "step": 793
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.9011510547206983e-05,
+      "loss": 0.1825,
+      "step": 794
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8963590704707744e-05,
+      "loss": 0.2732,
+      "step": 795
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8915655919429424e-05,
+      "loss": 0.2262,
+      "step": 796
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8867706372086684e-05,
+      "loss": 0.2756,
+      "step": 797
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8819742243449815e-05,
+      "loss": 0.244,
+      "step": 798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.877176371434409e-05,
+      "loss": 0.2626,
+      "step": 799
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.872377096564909e-05,
+      "loss": 0.2115,
+      "step": 800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.867576417829797e-05,
+      "loss": 0.2856,
+      "step": 801
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.862774353327684e-05,
+      "loss": 0.1677,
+      "step": 802
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8579709211624034e-05,
+      "loss": 0.3426,
+      "step": 803
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8531661394429464e-05,
+      "loss": 0.3588,
+      "step": 804
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8483600262833903e-05,
+      "loss": 0.2081,
+      "step": 805
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8435525998028334e-05,
+      "loss": 0.1989,
+      "step": 806
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.838743878125324e-05,
+      "loss": 0.1828,
+      "step": 807
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8339338793797942e-05,
+      "loss": 0.1784,
+      "step": 808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.829122621699991e-05,
+      "loss": 0.2827,
+      "step": 809
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8243101232244064e-05,
+      "loss": 0.2907,
+      "step": 810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8194964020962106e-05,
+      "loss": 0.1445,
+      "step": 811
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8146814764631835e-05,
+      "loss": 0.2718,
+      "step": 812
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8098653644776462e-05,
+      "loss": 0.2541,
+      "step": 813
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8050480842963923e-05,
+      "loss": 0.2368,
+      "step": 814
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8002296540806176e-05,
+      "loss": 0.388,
+      "step": 815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.795410091995858e-05,
+      "loss": 0.2741,
+      "step": 816
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7905894162119118e-05,
+      "loss": 0.2027,
+      "step": 817
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.785767644902778e-05,
+      "loss": 0.2919,
+      "step": 818
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.780944796246587e-05,
+      "loss": 0.2176,
+      "step": 819
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7761208884255274e-05,
+      "loss": 0.2243,
+      "step": 820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7712959396257847e-05,
+      "loss": 0.2093,
+      "step": 821
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7664699680374668e-05,
+      "loss": 0.2598,
+      "step": 822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7616429918545372e-05,
+      "loss": 0.2837,
+      "step": 823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7568150292747487e-05,
+      "loss": 0.3447,
+      "step": 824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7519860984995703e-05,
+      "loss": 0.1362,
+      "step": 825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.747156217734122e-05,
+      "loss": 0.2897,
+      "step": 826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7423254051871067e-05,
+      "loss": 0.3078,
+      "step": 827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.737493679070739e-05,
+      "loss": 0.3228,
+      "step": 828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.732661057600678e-05,
+      "loss": 0.2625,
+      "step": 829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7278275589959567e-05,
+      "loss": 0.3225,
+      "step": 830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.722993201478916e-05,
+      "loss": 0.347,
+      "step": 831
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7181580032751375e-05,
+      "loss": 0.1953,
+      "step": 832
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.713321982613367e-05,
+      "loss": 0.2488,
+      "step": 833
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7084851577254554e-05,
+      "loss": 0.1825,
+      "step": 834
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7036475468462847e-05,
+      "loss": 0.2255,
+      "step": 835
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6988091682136973e-05,
+      "loss": 0.2152,
+      "step": 836
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6939700400684337e-05,
+      "loss": 0.3146,
+      "step": 837
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6891301806540568e-05,
+      "loss": 0.3407,
+      "step": 838
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6842896082168893e-05,
+      "loss": 0.2713,
+      "step": 839
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6794483410059406e-05,
+      "loss": 0.1729,
+      "step": 840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.674606397272839e-05,
+      "loss": 0.2376,
+      "step": 841
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6697637952717646e-05,
+      "loss": 0.2077,
+      "step": 842
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.664920553259378e-05,
+      "loss": 0.2652,
+      "step": 843
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.2752176523208618,
+      "eval_runtime": 119.1657,
+      "eval_samples_per_second": 2.518,
+      "eval_steps_per_second": 0.629,
+      "step": 843
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6600766894947532e-05,
+      "loss": 0.2246,
+      "step": 844
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.655232222239307e-05,
+      "loss": 0.1462,
+      "step": 845
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.650387169756734e-05,
+      "loss": 0.2393,
+      "step": 846
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6455415503129326e-05,
+      "loss": 0.1314,
+      "step": 847
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.64069538217594e-05,
+      "loss": 0.192,
+      "step": 848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.635848683615862e-05,
+      "loss": 0.0874,
+      "step": 849
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6310014729048023e-05,
+      "loss": 0.0998,
+      "step": 850
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6261537683167976e-05,
+      "loss": 0.1193,
+      "step": 851
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.621305588127746e-05,
+      "loss": 0.1411,
+      "step": 852
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.616456950615338e-05,
+      "loss": 0.1109,
+      "step": 853
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6116078740589883e-05,
+      "loss": 0.143,
+      "step": 854
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.606758376739768e-05,
+      "loss": 0.1552,
+      "step": 855
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6019084769403322e-05,
+      "loss": 0.1401,
+      "step": 856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5970581929448555e-05,
+      "loss": 0.1238,
+      "step": 857
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.592207543038961e-05,
+      "loss": 0.157,
+      "step": 858
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.587356545509648e-05,
+      "loss": 0.1354,
+      "step": 859
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5825052186452307e-05,
+      "loss": 0.0772,
+      "step": 860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5776535807352624e-05,
+      "loss": 0.1249,
+      "step": 861
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5728016500704698e-05,
+      "loss": 0.1722,
+      "step": 862
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5679494449426822e-05,
+      "loss": 0.1109,
+      "step": 863
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.563096983644765e-05,
+      "loss": 0.1392,
+      "step": 864
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5582442844705496e-05,
+      "loss": 0.2738,
+      "step": 865
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5533913657147617e-05,
+      "loss": 0.1032,
+      "step": 866
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.548538245672959e-05,
+      "loss": 0.1362,
+      "step": 867
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.543684942641454e-05,
+      "loss": 0.1145,
+      "step": 868
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5388314749172504e-05,
+      "loss": 0.1221,
+      "step": 869
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5339778607979735e-05,
+      "loss": 0.1397,
+      "step": 870
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5291241185818003e-05,
+      "loss": 0.1286,
+      "step": 871
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5242702665673917e-05,
+      "loss": 0.1395,
+      "step": 872
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5194163230538193e-05,
+      "loss": 0.1255,
+      "step": 873
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5145623063405032e-05,
+      "loss": 0.1144,
+      "step": 874
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5097082347271374e-05,
+      "loss": 0.1663,
+      "step": 875
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5048541265136232e-05,
+      "loss": 0.0835,
+      "step": 876
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.5e-05,
+      "loss": 0.1649,
+      "step": 877
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.4951458734863774e-05,
+      "loss": 0.1575,
+      "step": 878
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.490291765272863e-05,
+      "loss": 0.1065,
+      "step": 879
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.4854376936594977e-05,
+      "loss": 0.1426,
+      "step": 880
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.4805836769461813e-05,
+      "loss": 0.1179,
+      "step": 881
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.475729733432609e-05,
+      "loss": 0.0955,
+      "step": 882
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.4708758814181992e-05,
+      "loss": 0.1272,
+      "step": 883
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.4660221392020274e-05,
+      "loss": 0.1693,
+      "step": 884
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.461168525082751e-05,
+      "loss": 0.1303,
+      "step": 885
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.456315057358547e-05,
+      "loss": 0.1074,
+      "step": 886
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.451461754327042e-05,
+      "loss": 0.1135,
+      "step": 887
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.4466086342852382e-05,
+      "loss": 0.1152,
+      "step": 888
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.441755715529452e-05,
+      "loss": 0.1625,
+      "step": 889
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.4369030163552357e-05,
+      "loss": 0.1822,
+      "step": 890
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.4320505550573183e-05,
+      "loss": 0.1428,
+      "step": 891
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.4271983499295305e-05,
+      "loss": 0.1132,
+      "step": 892
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.422346419264738e-05,
+      "loss": 0.1193,
+      "step": 893
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.4174947813547702e-05,
+      "loss": 0.1895,
+      "step": 894
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.4126434544903525e-05,
+      "loss": 0.1162,
+      "step": 895
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.4077924569610396e-05,
+      "loss": 0.1293,
+      "step": 896
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.402941807055144e-05,
+      "loss": 0.0926,
+      "step": 897
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.3980915230596684e-05,
+      "loss": 0.1238,
+      "step": 898
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.3932416232602327e-05,
+      "loss": 0.1547,
+      "step": 899
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.3883921259410123e-05,
+      "loss": 0.1359,
+      "step": 900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.3835430493846626e-05,
+      "loss": 0.1336,
+      "step": 901
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.378694411872254e-05,
+      "loss": 0.1205,
+      "step": 902
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.3738462316832033e-05,
+      "loss": 0.1257,
+      "step": 903
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.3689985270951987e-05,
+      "loss": 0.149,
+      "step": 904
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.364151316384139e-05,
+      "loss": 0.0938,
+      "step": 905
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.3593046178240606e-05,
+      "loss": 0.15,
+      "step": 906
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.3544584496870677e-05,
+      "loss": 0.2233,
+      "step": 907
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.3496128302432667e-05,
+      "loss": 0.1576,
+      "step": 908
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.3447677777606935e-05,
+      "loss": 0.0962,
+      "step": 909
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.3399233105052477e-05,
+      "loss": 0.1173,
+      "step": 910
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.335079446740622e-05,
+      "loss": 0.0992,
+      "step": 911
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.330236204728236e-05,
+      "loss": 0.135,
+      "step": 912
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.3253936027271618e-05,
+      "loss": 0.18,
+      "step": 913
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.32055165899406e-05,
+      "loss": 0.1581,
+      "step": 914
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.315710391783111e-05,
+      "loss": 0.1085,
+      "step": 915
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.3108698193459434e-05,
+      "loss": 0.1132,
+      "step": 916
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.3060299599315676e-05,
+      "loss": 0.1664,
+      "step": 917
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.3011908317863033e-05,
+      "loss": 0.2236,
+      "step": 918
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.2963524531537163e-05,
+      "loss": 0.1851,
+      "step": 919
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.2915148422745445e-05,
+      "loss": 0.1282,
+      "step": 920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.286678017386634e-05,
+      "loss": 0.1503,
+      "step": 921
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.2818419967248637e-05,
+      "loss": 0.1979,
+      "step": 922
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.2770067985210842e-05,
+      "loss": 0.1662,
+      "step": 923
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.272172441004044e-05,
+      "loss": 0.0973,
+      "step": 924
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.267338942399323e-05,
+      "loss": 0.1029,
+      "step": 925
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.262506320929262e-05,
+      "loss": 0.1139,
+      "step": 926
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.257674594812894e-05,
+      "loss": 0.1147,
+      "step": 927
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.2528437822658782e-05,
+      "loss": 0.1123,
+      "step": 928
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.2480139015004303e-05,
+      "loss": 0.1194,
+      "step": 929
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.243184970725252e-05,
+      "loss": 0.1294,
+      "step": 930
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.238357008145463e-05,
+      "loss": 0.1303,
+      "step": 931
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.233530031962534e-05,
+      "loss": 0.1026,
+      "step": 932
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.2287040603742155e-05,
+      "loss": 0.1535,
+      "step": 933
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.2238791115744725e-05,
+      "loss": 0.085,
+      "step": 934
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.2190552037534143e-05,
+      "loss": 0.1252,
+      "step": 935
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.2142323550972224e-05,
+      "loss": 0.1267,
+      "step": 936
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.2094105837880884e-05,
+      "loss": 0.1736,
+      "step": 937
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.2045899080041428e-05,
+      "loss": 0.0939,
+      "step": 938
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.1997703459193823e-05,
+      "loss": 0.1203,
+      "step": 939
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.1949519157036093e-05,
+      "loss": 0.1051,
+      "step": 940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.1901346355223547e-05,
+      "loss": 0.1457,
+      "step": 941
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.1853185235368167e-05,
+      "loss": 0.0554,
+      "step": 942
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.1805035979037897e-05,
+      "loss": 0.1718,
+      "step": 943
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.1756898767755938e-05,
+      "loss": 0.1159,
+      "step": 944
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.17087737830001e-05,
+      "loss": 0.1209,
+      "step": 945
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.1660661206202064e-05,
+      "loss": 0.1403,
+      "step": 946
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.1612561218746764e-05,
+      "loss": 0.1215,
+      "step": 947
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.1564474001971668e-05,
+      "loss": 0.1591,
+      "step": 948
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.1516399737166103e-05,
+      "loss": 0.1329,
+      "step": 949
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.1468338605570542e-05,
+      "loss": 0.1541,
+      "step": 950
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.142029078837597e-05,
+      "loss": 0.1101,
+      "step": 951
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.1372256466723165e-05,
+      "loss": 0.1102,
+      "step": 952
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.132423582170203e-05,
+      "loss": 0.1187,
+      "step": 953
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.1276229034350924e-05,
+      "loss": 0.0774,
+      "step": 954
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.1228236285655915e-05,
+      "loss": 0.1375,
+      "step": 955
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.118025775655019e-05,
+      "loss": 0.1574,
+      "step": 956
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.1132293627913325e-05,
+      "loss": 0.2015,
+      "step": 957
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.1084344080570588e-05,
+      "loss": 0.1211,
+      "step": 958
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.1036409295292266e-05,
+      "loss": 0.1638,
+      "step": 959
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.0988489452793023e-05,
+      "loss": 0.1607,
+      "step": 960
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.094058473373117e-05,
+      "loss": 0.1598,
+      "step": 961
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.089269531870802e-05,
+      "loss": 0.1297,
+      "step": 962
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.0844821388267162e-05,
+      "loss": 0.0894,
+      "step": 963
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.079696312289383e-05,
+      "loss": 0.0989,
+      "step": 964
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.0749120703014184e-05,
+      "loss": 0.117,
+      "step": 965
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.070129430899466e-05,
+      "loss": 0.1257,
+      "step": 966
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0653484121141274e-05,
+      "loss": 0.0908,
+      "step": 967
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.060569031969895e-05,
+      "loss": 0.0963,
+      "step": 968
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.055791308485081e-05,
+      "loss": 0.1518,
+      "step": 969
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0510152596717548e-05,
+      "loss": 0.1295,
+      "step": 970
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0462409035356707e-05,
+      "loss": 0.1009,
+      "step": 971
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0414682580762035e-05,
+      "loss": 0.1177,
+      "step": 972
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0366973412862747e-05,
+      "loss": 0.1528,
+      "step": 973
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.031928171152293e-05,
+      "loss": 0.1452,
+      "step": 974
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.02716076565408e-05,
+      "loss": 0.1188,
+      "step": 975
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0223951427648057e-05,
+      "loss": 0.1005,
+      "step": 976
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0176313204509194e-05,
+      "loss": 0.1281,
+      "step": 977
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0128693166720804e-05,
+      "loss": 0.1171,
+      "step": 978
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0081091493810937e-05,
+      "loss": 0.0953,
+      "step": 979
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0033508365238414e-05,
+      "loss": 0.1419,
+      "step": 980
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9985943960392134e-05,
+      "loss": 0.1677,
+      "step": 981
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9938398458590407e-05,
+      "loss": 0.1232,
+      "step": 982
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9890872039080284e-05,
+      "loss": 0.1195,
+      "step": 983
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.984336488103686e-05,
+      "loss": 0.1141,
+      "step": 984
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9795877163562633e-05,
+      "loss": 0.1568,
+      "step": 985
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9748409065686816e-05,
+      "loss": 0.2223,
+      "step": 986
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9700960766364616e-05,
+      "loss": 0.1446,
+      "step": 987
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.965353244447664e-05,
+      "loss": 0.1296,
+      "step": 988
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9606124278828173e-05,
+      "loss": 0.1106,
+      "step": 989
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9558736448148485e-05,
+      "loss": 0.1635,
+      "step": 990
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9511369131090223e-05,
+      "loss": 0.0826,
+      "step": 991
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9464022506228657e-05,
+      "loss": 0.1426,
+      "step": 992
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.941669675206107e-05,
+      "loss": 0.0822,
+      "step": 993
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.936939204700606e-05,
+      "loss": 0.1312,
+      "step": 994
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.932210856940288e-05,
+      "loss": 0.1644,
+      "step": 995
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9274846497510727e-05,
+      "loss": 0.1393,
+      "step": 996
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9227606009508113e-05,
+      "loss": 0.1476,
+      "step": 997
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9180387283492185e-05,
+      "loss": 0.096,
+      "step": 998
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9133190497478043e-05,
+      "loss": 0.1272,
+      "step": 999
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.908601582939808e-05,
+      "loss": 0.0936,
+      "step": 1000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.903886345710128e-05,
+      "loss": 0.0965,
+      "step": 1001
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.89917335583526e-05,
+      "loss": 0.1546,
+      "step": 1002
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8944626310832248e-05,
+      "loss": 0.0956,
+      "step": 1003
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8897541892135055e-05,
+      "loss": 0.0933,
+      "step": 1004
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.885048047976979e-05,
+      "loss": 0.1077,
+      "step": 1005
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8803442251158454e-05,
+      "loss": 0.1396,
+      "step": 1006
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8756427383635682e-05,
+      "loss": 0.1161,
+      "step": 1007
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8709436054448026e-05,
+      "loss": 0.1937,
+      "step": 1008
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.86624684407533e-05,
+      "loss": 0.1633,
+      "step": 1009
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8615524719619894e-05,
+      "loss": 0.1792,
+      "step": 1010
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8568605068026142e-05,
+      "loss": 0.1649,
+      "step": 1011
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8521709662859626e-05,
+      "loss": 0.0818,
+      "step": 1012
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8474838680916527e-05,
+      "loss": 0.1711,
+      "step": 1013
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.842799229890095e-05,
+      "loss": 0.0896,
+      "step": 1014
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8381170693424245e-05,
+      "loss": 0.1483,
+      "step": 1015
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8334374041004354e-05,
+      "loss": 0.1491,
+      "step": 1016
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8287602518065167e-05,
+      "loss": 0.1216,
+      "step": 1017
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8240856300935827e-05,
+      "loss": 0.1505,
+      "step": 1018
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8194135565850052e-05,
+      "loss": 0.1163,
+      "step": 1019
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8147440488945518e-05,
+      "loss": 0.1217,
+      "step": 1020
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.810077124626317e-05,
+      "loss": 0.1485,
+      "step": 1021
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.8054128013746535e-05,
+      "loss": 0.1189,
+      "step": 1022
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.800751096724111e-05,
+      "loss": 0.1427,
+      "step": 1023
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7960920282493634e-05,
+      "loss": 0.1004,
+      "step": 1024
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.79143561351515e-05,
+      "loss": 0.1543,
+      "step": 1025
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7867818700762034e-05,
+      "loss": 0.1175,
+      "step": 1026
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7821308154771866e-05,
+      "loss": 0.0988,
+      "step": 1027
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7774824672526244e-05,
+      "loss": 0.1532,
+      "step": 1028
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7728368429268377e-05,
+      "loss": 0.1442,
+      "step": 1029
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768193960013881e-05,
+      "loss": 0.12,
+      "step": 1030
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7635538360174714e-05,
+      "loss": 0.1821,
+      "step": 1031
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7589164884309274e-05,
+      "loss": 0.0993,
+      "step": 1032
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7542819347370966e-05,
+      "loss": 0.1641,
+      "step": 1033
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7496501924082968e-05,
+      "loss": 0.1128,
+      "step": 1034
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7450212789062447e-05,
+      "loss": 0.1039,
+      "step": 1035
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.740395211681995e-05,
+      "loss": 0.0791,
+      "step": 1036
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.73577200817587e-05,
+      "loss": 0.1634,
+      "step": 1037
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7311516858173943e-05,
+      "loss": 0.1786,
+      "step": 1038
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.726534262025234e-05,
+      "loss": 0.1224,
+      "step": 1039
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7219197542071255e-05,
+      "loss": 0.1128,
+      "step": 1040
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7173081797598117e-05,
+      "loss": 0.1182,
+      "step": 1041
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.712699556068979e-05,
+      "loss": 0.1226,
+      "step": 1042
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7080939005091855e-05,
+      "loss": 0.1392,
+      "step": 1043
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.703491230443803e-05,
+      "loss": 0.1443,
+      "step": 1044
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6988915632249458e-05,
+      "loss": 0.1276,
+      "step": 1045
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.694294916193409e-05,
+      "loss": 0.1953,
+      "step": 1046
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6897013066786e-05,
+      "loss": 0.1054,
+      "step": 1047
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6851107519984756e-05,
+      "loss": 0.1029,
+      "step": 1048
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.680523269459476e-05,
+      "loss": 0.1613,
+      "step": 1049
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6759388763564584e-05,
+      "loss": 0.1227,
+      "step": 1050
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6713575899726356e-05,
+      "loss": 0.125,
+      "step": 1051
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6667794275795036e-05,
+      "loss": 0.1452,
+      "step": 1052
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.662204406436784e-05,
+      "loss": 0.1336,
+      "step": 1053
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6576325437923556e-05,
+      "loss": 0.2167,
+      "step": 1054
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6530638568821894e-05,
+      "loss": 0.1426,
+      "step": 1055
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6484983629302822e-05,
+      "loss": 0.1325,
+      "step": 1056
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6439360791485954e-05,
+      "loss": 0.1278,
+      "step": 1057
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6393770227369874e-05,
+      "loss": 0.0895,
+      "step": 1058
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6348212108831502e-05,
+      "loss": 0.1309,
+      "step": 1059
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6302686607625428e-05,
+      "loss": 0.1519,
+      "step": 1060
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6257193895383253e-05,
+      "loss": 0.1153,
+      "step": 1061
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.621173414361301e-05,
+      "loss": 0.1797,
+      "step": 1062
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.616630752369843e-05,
+      "loss": 0.0967,
+      "step": 1063
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.612091420689837e-05,
+      "loss": 0.1474,
+      "step": 1064
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6075554364346103e-05,
+      "loss": 0.2184,
+      "step": 1065
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6030228167048722e-05,
+      "loss": 0.1102,
+      "step": 1066
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.5984935785886468e-05,
+      "loss": 0.1438,
+      "step": 1067
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.59396773916121e-05,
+      "loss": 0.0791,
+      "step": 1068
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5894453154850264e-05,
+      "loss": 0.1601,
+      "step": 1069
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5849263246096786e-05,
+      "loss": 0.1176,
+      "step": 1070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5804107835718114e-05,
+      "loss": 0.1305,
+      "step": 1071
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.575898709395063e-05,
+      "loss": 0.1591,
+      "step": 1072
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5713901190899998e-05,
+      "loss": 0.1057,
+      "step": 1073
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.566885029654057e-05,
+      "loss": 0.1391,
+      "step": 1074
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5623834580714676e-05,
+      "loss": 0.1242,
+      "step": 1075
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5578854213132048e-05,
+      "loss": 0.1166,
+      "step": 1076
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5533909363369164e-05,
+      "loss": 0.1637,
+      "step": 1077
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5489000200868568e-05,
+      "loss": 0.1469,
+      "step": 1078
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.54441268949383e-05,
+      "loss": 0.0992,
+      "step": 1079
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5399289614751173e-05,
+      "loss": 0.1564,
+      "step": 1080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.535448852934423e-05,
+      "loss": 0.1034,
+      "step": 1081
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5309723807618027e-05,
+      "loss": 0.151,
+      "step": 1082
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5264995618336057e-05,
+      "loss": 0.0875,
+      "step": 1083
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5220304130124052e-05,
+      "loss": 0.1088,
+      "step": 1084
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5175649511469405e-05,
+      "loss": 0.1347,
+      "step": 1085
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5131031930720502e-05,
+      "loss": 0.2023,
+      "step": 1086
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5086451556086093e-05,
+      "loss": 0.1685,
+      "step": 1087
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5041908555634678e-05,
+      "loss": 0.1629,
+      "step": 1088
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4997403097293822e-05,
+      "loss": 0.1254,
+      "step": 1089
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4952935348849587e-05,
+      "loss": 0.1185,
+      "step": 1090
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.490850547794586e-05,
+      "loss": 0.0834,
+      "step": 1091
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4864113652083728e-05,
+      "loss": 0.1567,
+      "step": 1092
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4819760038620829e-05,
+      "loss": 0.0952,
+      "step": 1093
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4775444804770772e-05,
+      "loss": 0.1094,
+      "step": 1094
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.473116811760245e-05,
+      "loss": 0.1304,
+      "step": 1095
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4686930144039462e-05,
+      "loss": 0.1338,
+      "step": 1096
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.464273105085942e-05,
+      "loss": 0.0972,
+      "step": 1097
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.459857100469339e-05,
+      "loss": 0.1665,
+      "step": 1098
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4554450172025191e-05,
+      "loss": 0.1335,
+      "step": 1099
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4510368719190859e-05,
+      "loss": 0.1649,
+      "step": 1100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4466326812377928e-05,
+      "loss": 0.1451,
+      "step": 1101
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4422324617624854e-05,
+      "loss": 0.1221,
+      "step": 1102
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.437836230082038e-05,
+      "loss": 0.122,
+      "step": 1103
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4334440027702894e-05,
+      "loss": 0.1342,
+      "step": 1104
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4290557963859857e-05,
+      "loss": 0.0989,
+      "step": 1105
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4246716274727107e-05,
+      "loss": 0.08,
+      "step": 1106
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4202915125588273e-05,
+      "loss": 0.1212,
+      "step": 1107
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4159154681574155e-05,
+      "loss": 0.1019,
+      "step": 1108
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4115435107662082e-05,
+      "loss": 0.0637,
+      "step": 1109
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4071756568675331e-05,
+      "loss": 0.1287,
+      "step": 1110
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.402811922928245e-05,
+      "loss": 0.1413,
+      "step": 1111
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3984523253996672e-05,
+      "loss": 0.2187,
+      "step": 1112
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3940968807175287e-05,
+      "loss": 0.1599,
+      "step": 1113
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3897456053019013e-05,
+      "loss": 0.0722,
+      "step": 1114
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3853985155571411e-05,
+      "loss": 0.1208,
+      "step": 1115
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3810556278718223e-05,
+      "loss": 0.17,
+      "step": 1116
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.376716958618677e-05,
+      "loss": 0.1113,
+      "step": 1117
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.372382524154534e-05,
+      "loss": 0.1129,
+      "step": 1118
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3680523408202589e-05,
+      "loss": 0.097,
+      "step": 1119
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3637264249406882e-05,
+      "loss": 0.1424,
+      "step": 1120
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3594047928245701e-05,
+      "loss": 0.0904,
+      "step": 1121
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3550874607645043e-05,
+      "loss": 0.0723,
+      "step": 1122
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3507744450368773e-05,
+      "loss": 0.0766,
+      "step": 1123
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3464657619018062e-05,
+      "loss": 0.0977,
+      "step": 1124
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.342161427603071e-05,
+      "loss": 0.1093,
+      "step": 1125
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3378614583680583e-05,
+      "loss": 0.1345,
+      "step": 1126
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.333565870407697e-05,
+      "loss": 0.0753,
+      "step": 1127
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3292746799163991e-05,
+      "loss": 0.1454,
+      "step": 1128
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.324987903072001e-05,
+      "loss": 0.1266,
+      "step": 1129
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3207055560356931e-05,
+      "loss": 0.1338,
+      "step": 1130
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3164276549519722e-05,
+      "loss": 0.1396,
+      "step": 1131
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3121542159485697e-05,
+      "loss": 0.1833,
+      "step": 1132
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3078852551363954e-05,
+      "loss": 0.194,
+      "step": 1133
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3036207886094795e-05,
+      "loss": 0.2313,
+      "step": 1134
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2993608324449025e-05,
+      "loss": 0.0586,
+      "step": 1135
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.295105402702747e-05,
+      "loss": 0.1231,
+      "step": 1136
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2908545154260268e-05,
+      "loss": 0.1453,
+      "step": 1137
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2866081866406337e-05,
+      "loss": 0.1503,
+      "step": 1138
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2823664323552717e-05,
+      "loss": 0.1043,
+      "step": 1139
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2781292685613994e-05,
+      "loss": 0.1002,
+      "step": 1140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.273896711233169e-05,
+      "loss": 0.1011,
+      "step": 1141
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2696687763273663e-05,
+      "loss": 0.1465,
+      "step": 1142
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2654454797833531e-05,
+      "loss": 0.1675,
+      "step": 1143
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.261226837522999e-05,
+      "loss": 0.1005,
+      "step": 1144
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2570128654506324e-05,
+      "loss": 0.168,
+      "step": 1145
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2528035794529724e-05,
+      "loss": 0.1272,
+      "step": 1146
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2485989953990703e-05,
+      "loss": 0.122,
+      "step": 1147
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2443991291402563e-05,
+      "loss": 0.1225,
+      "step": 1148
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2402039965100667e-05,
+      "loss": 0.126,
+      "step": 1149
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2360136133241986e-05,
+      "loss": 0.113,
+      "step": 1150
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2318279953804408e-05,
+      "loss": 0.1045,
+      "step": 1151
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.227647158458616e-05,
+      "loss": 0.1646,
+      "step": 1152
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.223471118320527e-05,
+      "loss": 0.1719,
+      "step": 1153
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2192998907098855e-05,
+      "loss": 0.1184,
+      "step": 1154
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2151334913522668e-05,
+      "loss": 0.1288,
+      "step": 1155
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.210971935955039e-05,
+      "loss": 0.125,
+      "step": 1156
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2068152402073128e-05,
+      "loss": 0.12,
+      "step": 1157
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2026634197798725e-05,
+      "loss": 0.0967,
+      "step": 1158
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1985164903251252e-05,
+      "loss": 0.1315,
+      "step": 1159
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1943744674770416e-05,
+      "loss": 0.1556,
+      "step": 1160
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1902373668510888e-05,
+      "loss": 0.0964,
+      "step": 1161
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.186105204044184e-05,
+      "loss": 0.1493,
+      "step": 1162
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1819779946346204e-05,
+      "loss": 0.1061,
+      "step": 1163
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1778557541820246e-05,
+      "loss": 0.0629,
+      "step": 1164
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1737384982272855e-05,
+      "loss": 0.1333,
+      "step": 1165
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1696262422925026e-05,
+      "loss": 0.1346,
+      "step": 1166
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1655190018809236e-05,
+      "loss": 0.1959,
+      "step": 1167
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1614167924768882e-05,
+      "loss": 0.1594,
+      "step": 1168
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1573196295457708e-05,
+      "loss": 0.1052,
+      "step": 1169
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1532275285339189e-05,
+      "loss": 0.0575,
+      "step": 1170
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.149140504868596e-05,
+      "loss": 0.1455,
+      "step": 1171
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1450585739579256e-05,
+      "loss": 0.1342,
+      "step": 1172
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1409817511908289e-05,
+      "loss": 0.1609,
+      "step": 1173
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1369100519369735e-05,
+      "loss": 0.1078,
+      "step": 1174
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1328434915467068e-05,
+      "loss": 0.1862,
+      "step": 1175
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1287820853510073e-05,
+      "loss": 0.0874,
+      "step": 1176
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1247258486614169e-05,
+      "loss": 0.1342,
+      "step": 1177
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.120674796769991e-05,
+      "loss": 0.0909,
+      "step": 1178
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1166289449492401e-05,
+      "loss": 0.1023,
+      "step": 1179
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1125883084520674e-05,
+      "loss": 0.1757,
+      "step": 1180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1085529025117161e-05,
+      "loss": 0.1451,
+      "step": 1181
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1045227423417082e-05,
+      "loss": 0.1336,
+      "step": 1182
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.100497843135792e-05,
+      "loss": 0.1091,
+      "step": 1183
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0964782200678795e-05,
+      "loss": 0.1478,
+      "step": 1184
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0924638882919925e-05,
+      "loss": 0.1093,
+      "step": 1185
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0884548629422042e-05,
+      "loss": 0.1458,
+      "step": 1186
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0844511591325823e-05,
+      "loss": 0.1238,
+      "step": 1187
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0804527919571341e-05,
+      "loss": 0.0978,
+      "step": 1188
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0764597764897455e-05,
+      "loss": 0.1192,
+      "step": 1189
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0724721277841276e-05,
+      "loss": 0.0812,
+      "step": 1190
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.068489860873758e-05,
+      "loss": 0.0981,
+      "step": 1191
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0645129907718243e-05,
+      "loss": 0.1202,
+      "step": 1192
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0605415324711713e-05,
+      "loss": 0.1237,
+      "step": 1193
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0565755009442378e-05,
+      "loss": 0.1078,
+      "step": 1194
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0526149111430047e-05,
+      "loss": 0.177,
+      "step": 1195
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0486597779989379e-05,
+      "loss": 0.1333,
+      "step": 1196
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0447101164229304e-05,
+      "loss": 0.1389,
+      "step": 1197
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0407659413052501e-05,
+      "loss": 0.1018,
+      "step": 1198
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.036827267515478e-05,
+      "loss": 0.119,
+      "step": 1199
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0328941099024566e-05,
+      "loss": 0.1026,
+      "step": 1200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0289664832942306e-05,
+      "loss": 0.113,
+      "step": 1201
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.025044402497996e-05,
+      "loss": 0.118,
+      "step": 1202
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0211278823000381e-05,
+      "loss": 0.0616,
+      "step": 1203
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0172169374656799e-05,
+      "loss": 0.1393,
+      "step": 1204
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0133115827392244e-05,
+      "loss": 0.0941,
+      "step": 1205
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0094118328439004e-05,
+      "loss": 0.1501,
+      "step": 1206
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0055177024818078e-05,
+      "loss": 0.1502,
+      "step": 1207
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0016292063338583e-05,
+      "loss": 0.0988,
+      "step": 1208
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.977463590597242e-06,
+      "loss": 0.1966,
+      "step": 1209
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.938691752977803e-06,
+      "loss": 0.1601,
+      "step": 1210
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.899976696650504e-06,
+      "loss": 0.0721,
+      "step": 1211
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.861318567571534e-06,
+      "loss": 0.0712,
+      "step": 1212
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.82271751148244e-06,
+      "loss": 0.1241,
+      "step": 1213
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.784173673909616e-06,
+      "loss": 0.1512,
+      "step": 1214
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.745687200163736e-06,
+      "loss": 0.1194,
+      "step": 1215
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.707258235339212e-06,
+      "loss": 0.1073,
+      "step": 1216
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.668886924313668e-06,
+      "loss": 0.1113,
+      "step": 1217
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.630573411747329e-06,
+      "loss": 0.1102,
+      "step": 1218
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.592317842082563e-06,
+      "loss": 0.1276,
+      "step": 1219
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.554120359543259e-06,
+      "loss": 0.1438,
+      "step": 1220
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.51598110813435e-06,
+      "loss": 0.1074,
+      "step": 1221
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.47790023164122e-06,
+      "loss": 0.0751,
+      "step": 1222
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.43987787362915e-06,
+      "loss": 0.1376,
+      "step": 1223
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.40191417744286e-06,
+      "loss": 0.0956,
+      "step": 1224
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.364009286205874e-06,
+      "loss": 0.1432,
+      "step": 1225
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.326163342820046e-06,
+      "loss": 0.1,
+      "step": 1226
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.288376489964982e-06,
+      "loss": 0.1296,
+      "step": 1227
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.250648870097518e-06,
+      "loss": 0.0851,
+      "step": 1228
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.212980625451176e-06,
+      "loss": 0.2346,
+      "step": 1229
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.175371898035636e-06,
+      "loss": 0.1305,
+      "step": 1230
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.137822829636225e-06,
+      "loss": 0.165,
+      "step": 1231
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.100333561813296e-06,
+      "loss": 0.1135,
+      "step": 1232
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.062904235901808e-06,
+      "loss": 0.1517,
+      "step": 1233
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.025534993010715e-06,
+      "loss": 0.1258,
+      "step": 1234
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.988225974022446e-06,
+      "loss": 0.2852,
+      "step": 1235
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.95097731959243e-06,
+      "loss": 0.1114,
+      "step": 1236
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.913789170148454e-06,
+      "loss": 0.1386,
+      "step": 1237
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.87666166589026e-06,
+      "loss": 0.1488,
+      "step": 1238
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.839594946788915e-06,
+      "loss": 0.2591,
+      "step": 1239
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.802589152586376e-06,
+      "loss": 0.11,
+      "step": 1240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.76564442279485e-06,
+      "loss": 0.0713,
+      "step": 1241
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.728760896696362e-06,
+      "loss": 0.1083,
+      "step": 1242
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.691938713342215e-06,
+      "loss": 0.1353,
+      "step": 1243
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.655178011552414e-06,
+      "loss": 0.0968,
+      "step": 1244
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.618478929915221e-06,
+      "loss": 0.0819,
+      "step": 1245
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.581841606786531e-06,
+      "loss": 0.1388,
+      "step": 1246
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.545266180289465e-06,
+      "loss": 0.1,
+      "step": 1247
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.508752788313761e-06,
+      "loss": 0.1425,
+      "step": 1248
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.472301568515287e-06,
+      "loss": 0.1154,
+      "step": 1249
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.435912658315556e-06,
+      "loss": 0.1349,
+      "step": 1250
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.39958619490111e-06,
+      "loss": 0.1487,
+      "step": 1251
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.363322315223129e-06,
+      "loss": 0.1043,
+      "step": 1252
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.327121155996814e-06,
+      "loss": 0.1116,
+      "step": 1253
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.290982853700924e-06,
+      "loss": 0.086,
+      "step": 1254
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.25490754457724e-06,
+      "loss": 0.1255,
+      "step": 1255
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.218895364630052e-06,
+      "loss": 0.1263,
+      "step": 1256
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.182946449625678e-06,
+      "loss": 0.1045,
+      "step": 1257
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.147060935091894e-06,
+      "loss": 0.153,
+      "step": 1258
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.111238956317496e-06,
+      "loss": 0.0833,
+      "step": 1259
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.0754806483517e-06,
+      "loss": 0.1391,
+      "step": 1260
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.039786146003705e-06,
+      "loss": 0.1349,
+      "step": 1261
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.004155583842183e-06,
+      "loss": 0.1256,
+      "step": 1262
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.968589096194715e-06,
+      "loss": 0.1078,
+      "step": 1263
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.933086817147359e-06,
+      "loss": 0.0909,
+      "step": 1264
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.897648880544056e-06,
+      "loss": 0.0835,
+      "step": 1265
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.862275419986226e-06,
+      "loss": 0.0972,
+      "step": 1266
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.826966568832184e-06,
+      "loss": 0.0991,
+      "step": 1267
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.791722460196673e-06,
+      "loss": 0.1872,
+      "step": 1268
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.756543226950357e-06,
+      "loss": 0.1335,
+      "step": 1269
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.72142900171931e-06,
+      "loss": 0.0709,
+      "step": 1270
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.686379916884557e-06,
+      "loss": 0.0947,
+      "step": 1271
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.651396104581501e-06,
+      "loss": 0.0985,
+      "step": 1272
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.616477696699498e-06,
+      "loss": 0.1115,
+      "step": 1273
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.581624824881306e-06,
+      "loss": 0.1039,
+      "step": 1274
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.5468376205226236e-06,
+      "loss": 0.1254,
+      "step": 1275
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.512116214771589e-06,
+      "loss": 0.1149,
+      "step": 1276
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.4774607385282695e-06,
+      "loss": 0.1292,
+      "step": 1277
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.442871322444175e-06,
+      "loss": 0.1437,
+      "step": 1278
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.408348096921772e-06,
+      "loss": 0.1445,
+      "step": 1279
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.373891192113979e-06,
+      "loss": 0.1095,
+      "step": 1280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.339500737923713e-06,
+      "loss": 0.1278,
+      "step": 1281
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.305176864003338e-06,
+      "loss": 0.1386,
+      "step": 1282
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.2709196997542275e-06,
+      "loss": 0.121,
+      "step": 1283
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.236729374326248e-06,
+      "loss": 0.0748,
+      "step": 1284
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.2026060166173025e-06,
+      "loss": 0.1057,
+      "step": 1285
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.168549755272805e-06,
+      "loss": 0.1348,
+      "step": 1286
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.134560718685221e-06,
+      "loss": 0.0988,
+      "step": 1287
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.100639034993583e-06,
+      "loss": 0.1222,
+      "step": 1288
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.06678483208299e-06,
+      "loss": 0.2182,
+      "step": 1289
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.032998237584163e-06,
+      "loss": 0.0978,
+      "step": 1290
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.999279378872911e-06,
+      "loss": 0.1155,
+      "step": 1291
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.965628383069695e-06,
+      "loss": 0.1458,
+      "step": 1292
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.932045377039123e-06,
+      "loss": 0.1299,
+      "step": 1293
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.898530487389476e-06,
+      "loss": 0.0944,
+      "step": 1294
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.865083840472264e-06,
+      "loss": 0.1597,
+      "step": 1295
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.831705562381691e-06,
+      "loss": 0.1299,
+      "step": 1296
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.79839577895422e-06,
+      "loss": 0.1078,
+      "step": 1297
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.765154615768088e-06,
+      "loss": 0.0956,
+      "step": 1298
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.731982198142825e-06,
+      "loss": 0.0884,
+      "step": 1299
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.698878651138826e-06,
+      "loss": 0.1004,
+      "step": 1300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.665844099556778e-06,
+      "loss": 0.1263,
+      "step": 1301
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.632878667937318e-06,
+      "loss": 0.1453,
+      "step": 1302
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.59998248056046e-06,
+      "loss": 0.1626,
+      "step": 1303
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.56715566144519e-06,
+      "loss": 0.11,
+      "step": 1304
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.534398334348968e-06,
+      "loss": 0.1286,
+      "step": 1305
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.501710622767243e-06,
+      "loss": 0.0673,
+      "step": 1306
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.469092649933056e-06,
+      "loss": 0.1513,
+      "step": 1307
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.436544538816497e-06,
+      "loss": 0.0717,
+      "step": 1308
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.404066412124307e-06,
+      "loss": 0.1094,
+      "step": 1309
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.371658392299359e-06,
+      "loss": 0.1099,
+      "step": 1310
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.339320601520235e-06,
+      "loss": 0.1405,
+      "step": 1311
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.307053161700746e-06,
+      "loss": 0.1109,
+      "step": 1312
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.274856194489478e-06,
+      "loss": 0.1052,
+      "step": 1313
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.242729821269363e-06,
+      "loss": 0.0979,
+      "step": 1314
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.210674163157137e-06,
+      "loss": 0.1077,
+      "step": 1315
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.178689341002997e-06,
+      "loss": 0.1181,
+      "step": 1316
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.146775475390051e-06,
+      "loss": 0.0948,
+      "step": 1317
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.114932686633909e-06,
+      "loss": 0.0884,
+      "step": 1318
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.083161094782242e-06,
+      "loss": 0.1241,
+      "step": 1319
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.0514608196142605e-06,
+      "loss": 0.1073,
+      "step": 1320
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.0198319806403585e-06,
+      "loss": 0.0649,
+      "step": 1321
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.988274697101584e-06,
+      "loss": 0.1385,
+      "step": 1322
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.956789087969241e-06,
+      "loss": 0.1089,
+      "step": 1323
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.925375271944417e-06,
+      "loss": 0.1236,
+      "step": 1324
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.894033367457505e-06,
+      "loss": 0.1025,
+      "step": 1325
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.862763492667844e-06,
+      "loss": 0.1173,
+      "step": 1326
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.831565765463176e-06,
+      "loss": 0.1165,
+      "step": 1327
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.80044030345929e-06,
+      "loss": 0.0823,
+      "step": 1328
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.769387223999473e-06,
+      "loss": 0.1191,
+      "step": 1329
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.738406644154193e-06,
+      "loss": 0.0739,
+      "step": 1330
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.707498680720555e-06,
+      "loss": 0.1119,
+      "step": 1331
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.6766634502219e-06,
+      "loss": 0.1317,
+      "step": 1332
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.645901068907402e-06,
+      "loss": 0.1156,
+      "step": 1333
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.615211652751534e-06,
+      "loss": 0.1069,
+      "step": 1334
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.584595317453742e-06,
+      "loss": 0.1262,
+      "step": 1335
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.554052178437927e-06,
+      "loss": 0.134,
+      "step": 1336
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.5235823508520496e-06,
+      "loss": 0.1134,
+      "step": 1337
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.493185949567683e-06,
+      "loss": 0.1353,
+      "step": 1338
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.462863089179576e-06,
+      "loss": 0.1129,
+      "step": 1339
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.43261388400525e-06,
+      "loss": 0.0818,
+      "step": 1340
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.4024384480845126e-06,
+      "loss": 0.0989,
+      "step": 1341
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.372336895179103e-06,
+      "loss": 0.11,
+      "step": 1342
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.34230933877217e-06,
+      "loss": 0.1333,
+      "step": 1343
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.312355892067924e-06,
+      "loss": 0.1065,
+      "step": 1344
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.282476667991193e-06,
+      "loss": 0.1304,
+      "step": 1345
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.2526717791869486e-06,
+      "loss": 0.116,
+      "step": 1346
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.222941338019966e-06,
+      "loss": 0.1408,
+      "step": 1347
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.193285456574293e-06,
+      "loss": 0.1197,
+      "step": 1348
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.163704246652939e-06,
+      "loss": 0.0932,
+      "step": 1349
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.134197819777373e-06,
+      "loss": 0.1074,
+      "step": 1350
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.1047662871871376e-06,
+      "loss": 0.0862,
+      "step": 1351
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.075409759839425e-06,
+      "loss": 0.1344,
+      "step": 1352
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.046128348408647e-06,
+      "loss": 0.104,
+      "step": 1353
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.01692216328605e-06,
+      "loss": 0.0985,
+      "step": 1354
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.987791314579254e-06,
+      "loss": 0.1116,
+      "step": 1355
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.95873591211187e-06,
+      "loss": 0.1118,
+      "step": 1356
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.929756065423074e-06,
+      "loss": 0.1349,
+      "step": 1357
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.900851883767191e-06,
+      "loss": 0.1698,
+      "step": 1358
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.872023476113302e-06,
+      "loss": 0.0819,
+      "step": 1359
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.8432709511448046e-06,
+      "loss": 0.0712,
+      "step": 1360
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.814594417259024e-06,
+      "loss": 0.0786,
+      "step": 1361
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.785993982566802e-06,
+      "loss": 0.1385,
+      "step": 1362
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.757469754892069e-06,
+      "loss": 0.1181,
+      "step": 1363
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.729021841771483e-06,
+      "loss": 0.1203,
+      "step": 1364
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.700650350453969e-06,
+      "loss": 0.096,
+      "step": 1365
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.672355387900354e-06,
+      "loss": 0.1278,
+      "step": 1366
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.644137060782938e-06,
+      "loss": 0.1027,
+      "step": 1367
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.61599547548513e-06,
+      "loss": 0.1045,
+      "step": 1368
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.5879307381009986e-06,
+      "loss": 0.0892,
+      "step": 1369
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.559942954434901e-06,
+      "loss": 0.0985,
+      "step": 1370
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.532032230001079e-06,
+      "loss": 0.0946,
+      "step": 1371
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.504198670023255e-06,
+      "loss": 0.1091,
+      "step": 1372
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4764423794342566e-06,
+      "loss": 0.092,
+      "step": 1373
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.4487634628755884e-06,
+      "loss": 0.0996,
+      "step": 1374
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.421162024697062e-06,
+      "loss": 0.0878,
+      "step": 1375
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.39363816895639e-06,
+      "loss": 0.0931,
+      "step": 1376
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.3661919994188e-06,
+      "loss": 0.2033,
+      "step": 1377
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.338823619556648e-06,
+      "loss": 0.1361,
+      "step": 1378
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.311533132549017e-06,
+      "loss": 0.1271,
+      "step": 1379
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.284320641281328e-06,
+      "loss": 0.164,
+      "step": 1380
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.257186248344966e-06,
+      "loss": 0.1327,
+      "step": 1381
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.2301300560368715e-06,
+      "loss": 0.0989,
+      "step": 1382
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.203152166359192e-06,
+      "loss": 0.1076,
+      "step": 1383
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.176252681018852e-06,
+      "loss": 0.1741,
+      "step": 1384
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.149431701427195e-06,
+      "loss": 0.1466,
+      "step": 1385
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.122689328699595e-06,
+      "loss": 0.0996,
+      "step": 1386
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.09602566365509e-06,
+      "loss": 0.0916,
+      "step": 1387
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.069440806815985e-06,
+      "loss": 0.1181,
+      "step": 1388
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.042934858407454e-06,
+      "loss": 0.114,
+      "step": 1389
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.016507918357218e-06,
+      "loss": 0.1012,
+      "step": 1390
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.990160086295116e-06,
+      "loss": 0.0982,
+      "step": 1391
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.963891461552762e-06,
+      "loss": 0.1343,
+      "step": 1392
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.93770214316315e-06,
+      "loss": 0.124,
+      "step": 1393
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.911592229860281e-06,
+      "loss": 0.1191,
+      "step": 1394
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.885561820078809e-06,
+      "loss": 0.1061,
+      "step": 1395
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.859611011953654e-06,
+      "loss": 0.0841,
+      "step": 1396
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.833739903319644e-06,
+      "loss": 0.0811,
+      "step": 1397
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.80794859171113e-06,
+      "loss": 0.146,
+      "step": 1398
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.7822371743616305e-06,
+      "loss": 0.1197,
+      "step": 1399
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.7566057482034622e-06,
+      "loss": 0.1273,
+      "step": 1400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.7310544098673684e-06,
+      "loss": 0.1497,
+      "step": 1401
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.705583255682185e-06,
+      "loss": 0.0806,
+      "step": 1402
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.6801923816744078e-06,
+      "loss": 0.087,
+      "step": 1403
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.6548818835679203e-06,
+      "loss": 0.1089,
+      "step": 1404
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.6296518567835614e-06,
+      "loss": 0.0876,
+      "step": 1405
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.604502396438805e-06,
+      "loss": 0.1278,
+      "step": 1406
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.5794335973473896e-06,
+      "loss": 0.0956,
+      "step": 1407
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.5544455540189313e-06,
+      "loss": 0.0795,
+      "step": 1408
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.5295383606586436e-06,
+      "loss": 0.1412,
+      "step": 1409
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.5047121111668914e-06,
+      "loss": 0.1463,
+      "step": 1410
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.4799668991389227e-06,
+      "loss": 0.1356,
+      "step": 1411
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.4553028178644227e-06,
+      "loss": 0.133,
+      "step": 1412
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.4307199603272554e-06,
+      "loss": 0.0912,
+      "step": 1413
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.406218419205054e-06,
+      "loss": 0.1031,
+      "step": 1414
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.381798286868881e-06,
+      "loss": 0.1309,
+      "step": 1415
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3574596553829134e-06,
+      "loss": 0.1876,
+      "step": 1416
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3332026165040247e-06,
+      "loss": 0.061,
+      "step": 1417
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.309027261681527e-06,
+      "loss": 0.1391,
+      "step": 1418
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2849336820567623e-06,
+      "loss": 0.0946,
+      "step": 1419
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.260921968462774e-06,
+      "loss": 0.1705,
+      "step": 1420
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.236992211423995e-06,
+      "loss": 0.1328,
+      "step": 1421
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2131445011558403e-06,
+      "loss": 0.0996,
+      "step": 1422
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.189378927564446e-06,
+      "loss": 0.1193,
+      "step": 1423
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.165695580246278e-06,
+      "loss": 0.0792,
+      "step": 1424
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1420945484878004e-06,
+      "loss": 0.1219,
+      "step": 1425
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.1185759212651542e-06,
+      "loss": 0.0859,
+      "step": 1426
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.0951397872438144e-06,
+      "loss": 0.1128,
+      "step": 1427
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.071786234778262e-06,
+      "loss": 0.119,
+      "step": 1428
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.0485153519116264e-06,
+      "loss": 0.1078,
+      "step": 1429
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.025327226375402e-06,
+      "loss": 0.125,
+      "step": 1430
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.0022219455890455e-06,
+      "loss": 0.0977,
+      "step": 1431
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9791995966597264e-06,
+      "loss": 0.1145,
+      "step": 1432
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.956260266381941e-06,
+      "loss": 0.1194,
+      "step": 1433
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.9334040412372034e-06,
+      "loss": 0.1032,
+      "step": 1434
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.910631007393741e-06,
+      "loss": 0.1075,
+      "step": 1435
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.887941250706108e-06,
+      "loss": 0.086,
+      "step": 1436
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.865334856714949e-06,
+      "loss": 0.096,
+      "step": 1437
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.8428119106466033e-06,
+      "loss": 0.0993,
+      "step": 1438
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.8203724974128153e-06,
+      "loss": 0.1105,
+      "step": 1439
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.7980167016104148e-06,
+      "loss": 0.0958,
+      "step": 1440
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.7757446075209804e-06,
+      "loss": 0.0537,
+      "step": 1441
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.7535562991105545e-06,
+      "loss": 0.1117,
+      "step": 1442
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.731451860029291e-06,
+      "loss": 0.0974,
+      "step": 1443
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.7094313736111627e-06,
+      "loss": 0.11,
+      "step": 1444
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.6874949228736345e-06,
+      "loss": 0.0977,
+      "step": 1445
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.6656425905173534e-06,
+      "loss": 0.0965,
+      "step": 1446
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.6438744589258575e-06,
+      "loss": 0.2249,
+      "step": 1447
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.622190610165226e-06,
+      "loss": 0.0798,
+      "step": 1448
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.6005911259838004e-06,
+      "loss": 0.0977,
+      "step": 1449
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.57907608781186e-06,
+      "loss": 0.0972,
+      "step": 1450
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.5576455767613416e-06,
+      "loss": 0.0788,
+      "step": 1451
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.536299673625489e-06,
+      "loss": 0.0978,
+      "step": 1452
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.5150384588785854e-06,
+      "loss": 0.0989,
+      "step": 1453
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4938620126756377e-06,
+      "loss": 0.0892,
+      "step": 1454
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.472770414852066e-06,
+      "loss": 0.1043,
+      "step": 1455
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4517637449234325e-06,
+      "loss": 0.128,
+      "step": 1456
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.430842082085097e-06,
+      "loss": 0.0625,
+      "step": 1457
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4100055052119515e-06,
+      "loss": 0.0961,
+      "step": 1458
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3892540928581118e-06,
+      "loss": 0.0753,
+      "step": 1459
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3685879232566184e-06,
+      "loss": 0.0984,
+      "step": 1460
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.348007074319164e-06,
+      "loss": 0.1287,
+      "step": 1461
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.327511623635756e-06,
+      "loss": 0.1022,
+      "step": 1462
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.3071016484744663e-06,
+      "loss": 0.0877,
+      "step": 1463
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.286777225781117e-06,
+      "loss": 0.0959,
+      "step": 1464
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.266538432178991e-06,
+      "loss": 0.1011,
+      "step": 1465
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.2463853439685683e-06,
+      "loss": 0.1171,
+      "step": 1466
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.226318037127198e-06,
+      "loss": 0.1105,
+      "step": 1467
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.206336587308841e-06,
+      "loss": 0.1006,
+      "step": 1468
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1864410698437724e-06,
+      "loss": 0.1348,
+      "step": 1469
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1666315597383107e-06,
+      "loss": 0.1102,
+      "step": 1470
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.146908131674516e-06,
+      "loss": 0.1541,
+      "step": 1471
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1272708600099223e-06,
+      "loss": 0.1348,
+      "step": 1472
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.107719818777254e-06,
+      "loss": 0.0756,
+      "step": 1473
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0882550816841375e-06,
+      "loss": 0.0993,
+      "step": 1474
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0688767221128475e-06,
+      "loss": 0.1071,
+      "step": 1475
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0495848131200005e-06,
+      "loss": 0.1125,
+      "step": 1476
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.030379427436299e-06,
+      "loss": 0.1436,
+      "step": 1477
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.011260637466253e-06,
+      "loss": 0.1354,
+      "step": 1478
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9922285152879e-06,
+      "loss": 0.108,
+      "step": 1479
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9732831326525453e-06,
+      "loss": 0.1493,
+      "step": 1480
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9544245609844857e-06,
+      "loss": 0.1273,
+      "step": 1481
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.935652871380733e-06,
+      "loss": 0.1096,
+      "step": 1482
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.916968134610758e-06,
+      "loss": 0.0969,
+      "step": 1483
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8983704211162107e-06,
+      "loss": 0.0998,
+      "step": 1484
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8798598010106861e-06,
+      "loss": 0.1205,
+      "step": 1485
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8614363440793954e-06,
+      "loss": 0.106,
+      "step": 1486
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8431001197789816e-06,
+      "loss": 0.1222,
+      "step": 1487
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8248511972372012e-06,
+      "loss": 0.1139,
+      "step": 1488
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8066896452526877e-06,
+      "loss": 0.124,
+      "step": 1489
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7886155322946907e-06,
+      "loss": 0.1522,
+      "step": 1490
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7706289265027932e-06,
+      "loss": 0.1053,
+      "step": 1491
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7527298956867e-06,
+      "loss": 0.1356,
+      "step": 1492
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7349185073259444e-06,
+      "loss": 0.1165,
+      "step": 1493
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.7171948285696544e-06,
+      "loss": 0.1294,
+      "step": 1494
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.699558926236286e-06,
+      "loss": 0.1029,
+      "step": 1495
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.6820108668133771e-06,
+      "loss": 0.1121,
+      "step": 1496
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.664550716457297e-06,
+      "loss": 0.0673,
+      "step": 1497
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.647178540992994e-06,
+      "loss": 0.1169,
+      "step": 1498
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.6298944059137677e-06,
+      "loss": 0.1314,
+      "step": 1499
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.612698376380975e-06,
+      "loss": 0.088,
+      "step": 1500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.595590517223844e-06,
+      "loss": 0.1354,
+      "step": 1501
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.5785708929391741e-06,
+      "loss": 0.1067,
+      "step": 1502
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.5616395676911339e-06,
+      "loss": 0.0763,
+      "step": 1503
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.544796605311008e-06,
+      "loss": 0.1398,
+      "step": 1504
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.5280420692969283e-06,
+      "loss": 0.1444,
+      "step": 1505
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.5113760228136847e-06,
+      "loss": 0.1295,
+      "step": 1506
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4947985286924454e-06,
+      "loss": 0.0716,
+      "step": 1507
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4783096494305342e-06,
+      "loss": 0.1242,
+      "step": 1508
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4619094471912177e-06,
+      "loss": 0.0726,
+      "step": 1509
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.445597983803415e-06,
+      "loss": 0.1215,
+      "step": 1510
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.429375320761525e-06,
+      "loss": 0.1018,
+      "step": 1511
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.413241519225153e-06,
+      "loss": 0.1488,
+      "step": 1512
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3971966400189162e-06,
+      "loss": 0.098,
+      "step": 1513
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3812407436321645e-06,
+      "loss": 0.1123,
+      "step": 1514
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3653738902188073e-06,
+      "loss": 0.093,
+      "step": 1515
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3495961395970452e-06,
+      "loss": 0.1102,
+      "step": 1516
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3339075512491639e-06,
+      "loss": 0.0882,
+      "step": 1517
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.318308184321318e-06,
+      "loss": 0.107,
+      "step": 1518
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3027980976232702e-06,
+      "loss": 0.0766,
+      "step": 1519
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2873773496282298e-06,
+      "loss": 0.1422,
+      "step": 1520
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2720459984725707e-06,
+      "loss": 0.1008,
+      "step": 1521
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2568041019556498e-06,
+      "loss": 0.1396,
+      "step": 1522
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2416517175395826e-06,
+      "loss": 0.0963,
+      "step": 1523
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2265889023490157e-06,
+      "loss": 0.0879,
+      "step": 1524
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2116157131709244e-06,
+      "loss": 0.1432,
+      "step": 1525
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1967322064543928e-06,
+      "loss": 0.0763,
+      "step": 1526
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1819384383103981e-06,
+      "loss": 0.105,
+      "step": 1527
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1672344645116019e-06,
+      "loss": 0.0904,
+      "step": 1528
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1526203404921387e-06,
+      "loss": 0.129,
+      "step": 1529
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1380961213474177e-06,
+      "loss": 0.0667,
+      "step": 1530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1236618618338967e-06,
+      "loss": 0.1257,
+      "step": 1531
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1093176163688905e-06,
+      "loss": 0.1067,
+      "step": 1532
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.09506343903035e-06,
+      "loss": 0.0863,
+      "step": 1533
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0808993835566828e-06,
+      "loss": 0.1366,
+      "step": 1534
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0668255033465247e-06,
+      "loss": 0.1116,
+      "step": 1535
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0528418514585525e-06,
+      "loss": 0.115,
+      "step": 1536
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0389484806112876e-06,
+      "loss": 0.1097,
+      "step": 1537
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.025145443182876e-06,
+      "loss": 0.1206,
+      "step": 1538
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.0114327912109256e-06,
+      "loss": 0.1492,
+      "step": 1539
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.978105763922773e-07,
+      "loss": 0.1415,
+      "step": 1540
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.84278850082823e-07,
+      "loss": 0.0782,
+      "step": 1541
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.708376632973164e-07,
+      "loss": 0.0994,
+      "step": 1542
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.574870667091701e-07,
+      "loss": 0.1182,
+      "step": 1543
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.442271106502837e-07,
+      "loss": 0.093,
+      "step": 1544
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.310578451108304e-07,
+      "loss": 0.0951,
+      "step": 1545
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.179793197390763e-07,
+      "loss": 0.1345,
+      "step": 1546
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.049915838412027e-07,
+      "loss": 0.1178,
+      "step": 1547
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.920946863811119e-07,
+      "loss": 0.1118,
+      "step": 1548
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.792886759802471e-07,
+      "loss": 0.1547,
+      "step": 1549
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.66573600917403e-07,
+      "loss": 0.151,
+      "step": 1550
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.539495091285459e-07,
+      "loss": 0.1292,
+      "step": 1551
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.414164482066333e-07,
+      "loss": 0.1194,
+      "step": 1552
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.28974465401447e-07,
+      "loss": 0.0824,
+      "step": 1553
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.16623607619385e-07,
+      "loss": 0.1535,
+      "step": 1554
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.043639214233207e-07,
+      "loss": 0.1059,
+      "step": 1555
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.921954530323933e-07,
+      "loss": 0.1577,
+      "step": 1556
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.801182483218595e-07,
+      "loss": 0.0878,
+      "step": 1557
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.681323528229145e-07,
+      "loss": 0.1004,
+      "step": 1558
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.5623781172251e-07,
+      "loss": 0.0641,
+      "step": 1559
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.444346698631894e-07,
+      "loss": 0.1381,
+      "step": 1560
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.327229717429245e-07,
+      "loss": 0.1161,
+      "step": 1561
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.211027615149385e-07,
+      "loss": 0.1388,
+      "step": 1562
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.095740829875547e-07,
+      "loss": 0.118,
+      "step": 1563
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.981369796240089e-07,
+      "loss": 0.1204,
+      "step": 1564
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.867914945423049e-07,
+      "loss": 0.1001,
+      "step": 1565
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.755376705150363e-07,
+      "loss": 0.1359,
+      "step": 1566
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.643755499692428e-07,
+      "loss": 0.1169,
+      "step": 1567
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.533051749862407e-07,
+      "loss": 0.0915,
+      "step": 1568
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.423265873014589e-07,
+      "loss": 0.1115,
+      "step": 1569
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.314398283042893e-07,
+      "loss": 0.1241,
+      "step": 1570
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.20644939037926e-07,
+      "loss": 0.115,
+      "step": 1571
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.099419601992201e-07,
+      "loss": 0.1246,
+      "step": 1572
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.993309321385171e-07,
+      "loss": 0.0975,
+      "step": 1573
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.888118948595006e-07,
+      "loss": 0.1134,
+      "step": 1574
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.783848880190567e-07,
+      "loss": 0.0923,
+      "step": 1575
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.680499509271076e-07,
+      "loss": 0.1309,
+      "step": 1576
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.578071225464803e-07,
+      "loss": 0.113,
+      "step": 1577
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.47656441492747e-07,
+      "loss": 0.14,
+      "step": 1578
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.375979460340824e-07,
+      "loss": 0.093,
+      "step": 1579
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.276316740911197e-07,
+      "loss": 0.1098,
+      "step": 1580
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.177576632368092e-07,
+      "loss": 0.1014,
+      "step": 1581
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.079759506962795e-07,
+      "loss": 0.1092,
+      "step": 1582
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.982865733466874e-07,
+      "loss": 0.1228,
+      "step": 1583
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.886895677170933e-07,
+      "loss": 0.0758,
+      "step": 1584
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.791849699883083e-07,
+      "loss": 0.0999,
+      "step": 1585
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.697728159927639e-07,
+      "loss": 0.0957,
+      "step": 1586
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.60453141214387e-07,
+      "loss": 0.0792,
+      "step": 1587
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.512259807884417e-07,
+      "loss": 0.0915,
+      "step": 1588
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.420913695014295e-07,
+      "loss": 0.1734,
+      "step": 1589
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.330493417909254e-07,
+      "loss": 0.1091,
+      "step": 1590
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.240999317454725e-07,
+      "loss": 0.1025,
+      "step": 1591
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.152431731044487e-07,
+      "loss": 0.15,
+      "step": 1592
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0647909925791426e-07,
+      "loss": 0.1162,
+      "step": 1593
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.97807743246531e-07,
+      "loss": 0.1279,
+      "step": 1594
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.892291377613988e-07,
+      "loss": 0.1216,
+      "step": 1595
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8074331514395824e-07,
+      "loss": 0.1044,
+      "step": 1596
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7235030738584084e-07,
+      "loss": 0.0641,
+      "step": 1597
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.6405014612878044e-07,
+      "loss": 0.0878,
+      "step": 1598
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.558428626644739e-07,
+      "loss": 0.1061,
+      "step": 1599
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.47728487934465e-07,
+      "loss": 0.0992,
+      "step": 1600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.3970705253003607e-07,
+      "loss": 0.1816,
+      "step": 1601
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.317785866920775e-07,
+      "loss": 0.1118,
+      "step": 1602
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.239431203109933e-07,
+      "loss": 0.1187,
+      "step": 1603
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1620068292657625e-07,
+      "loss": 0.081,
+      "step": 1604
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.08551303727897e-07,
+      "loss": 0.1085,
+      "step": 1605
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.009950115531984e-07,
+      "loss": 0.0814,
+      "step": 1606
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9353183488977364e-07,
+      "loss": 0.1123,
+      "step": 1607
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.861618018738882e-07,
+      "loss": 0.0972,
+      "step": 1608
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.788849402906413e-07,
+      "loss": 0.0961,
+      "step": 1609
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.71701277573877e-07,
+      "loss": 0.1137,
+      "step": 1610
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.6461084080608157e-07,
+      "loss": 0.128,
+      "step": 1611
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.576136567182752e-07,
+      "loss": 0.104,
+      "step": 1612
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.507097516899176e-07,
+      "loss": 0.1349,
+      "step": 1613
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.4389915174880264e-07,
+      "loss": 0.1089,
+      "step": 1614
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.371818825709693e-07,
+      "loss": 0.1062,
+      "step": 1615
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3055796948058816e-07,
+      "loss": 0.0802,
+      "step": 1616
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.240274374498863e-07,
+      "loss": 0.1262,
+      "step": 1617
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.175903110990335e-07,
+      "loss": 0.0956,
+      "step": 1618
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.112466146960701e-07,
+      "loss": 0.1017,
+      "step": 1619
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.04996372156796e-07,
+      "loss": 0.0907,
+      "step": 1620
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9883960704469018e-07,
+      "loss": 0.2302,
+      "step": 1621
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9277634257082734e-07,
+      "loss": 0.1546,
+      "step": 1622
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8680660159377806e-07,
+      "loss": 0.0741,
+      "step": 1623
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8093040661952555e-07,
+      "loss": 0.0831,
+      "step": 1624
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7514777980139062e-07,
+      "loss": 0.3893,
+      "step": 1625
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.694587429399347e-07,
+      "loss": 0.1135,
+      "step": 1626
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.6386331748289018e-07,
+      "loss": 0.072,
+      "step": 1627
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5836152452506913e-07,
+      "loss": 0.136,
+      "step": 1628
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.529533848082909e-07,
+      "loss": 0.1248,
+      "step": 1629
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.47638918721299e-07,
+      "loss": 0.2179,
+      "step": 1630
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4241814629968597e-07,
+      "loss": 0.079,
+      "step": 1631
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3729108722581873e-07,
+      "loss": 0.1404,
+      "step": 1632
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3225776082876895e-07,
+      "loss": 0.1148,
+      "step": 1633
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2731818608422707e-07,
+      "loss": 0.2161,
+      "step": 1634
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.224723816144385e-07,
+      "loss": 0.1393,
+      "step": 1635
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1772036568814537e-07,
+      "loss": 0.1139,
+      "step": 1636
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1306215622049477e-07,
+      "loss": 0.1484,
+      "step": 1637
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0849777077298617e-07,
+      "loss": 0.1434,
+      "step": 1638
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0402722655339924e-07,
+      "loss": 0.1226,
+      "step": 1639
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.965054041573552e-08,
+      "loss": 0.1252,
+      "step": 1640
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.536772886014899e-08,
+      "loss": 0.0951,
+      "step": 1641
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.117880803288237e-08,
+      "loss": 0.1476,
+      "step": 1642
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.708379372621144e-08,
+      "loss": 0.1062,
+      "step": 1643
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.308270137838692e-08,
+      "loss": 0.092,
+      "step": 1644
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.917554607356492e-08,
+      "loss": 0.1016,
+      "step": 1645
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.536234254176544e-08,
+      "loss": 0.0641,
+      "step": 1646
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.164310515880846e-08,
+      "loss": 0.1292,
+      "step": 1647
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.801784794625287e-08,
+      "loss": 0.0773,
+      "step": 1648
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.448658457136326e-08,
+      "loss": 0.0808,
+      "step": 1649
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.10493283470348e-08,
+      "loss": 0.0779,
+      "step": 1650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7706092231760156e-08,
+      "loss": 0.0776,
+      "step": 1651
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.445688882957933e-08,
+      "loss": 0.1363,
+      "step": 1652
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.1301730390018734e-08,
+      "loss": 0.0925,
+      "step": 1653
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.8240628808063346e-08,
+      "loss": 0.0877,
+      "step": 1654
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.527359562409572e-08,
+      "loss": 0.0768,
+      "step": 1655
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.24006420238654e-08,
+      "loss": 0.0894,
+      "step": 1656
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.962177883843343e-08,
+      "loss": 0.1266,
+      "step": 1657
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.6937016544147386e-08,
+      "loss": 0.1209,
+      "step": 1658
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.434636526258861e-08,
+      "loss": 0.1436,
+      "step": 1659
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.184983476054171e-08,
+      "loss": 0.0835,
+      "step": 1660
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.944743444995013e-08,
+      "loss": 0.1548,
+      "step": 1661
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.7139173387891182e-08,
+      "loss": 0.1046,
+      "step": 1662
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.4925060276528857e-08,
+      "loss": 0.1075,
+      "step": 1663
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.2805103463099942e-08,
+      "loss": 0.1032,
+      "step": 1664
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.077931093985852e-08,
+      "loss": 0.1386,
+      "step": 1665
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8847690344067637e-08,
+      "loss": 0.0772,
+      "step": 1666
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7010248957954887e-08,
+      "loss": 0.1207,
+      "step": 1667
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.5266993708695776e-08,
+      "loss": 0.1153,
+      "step": 1668
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3617931168380394e-08,
+      "loss": 0.1396,
+      "step": 1669
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2063067553991225e-08,
+      "loss": 0.101,
+      "step": 1670
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0602408727383717e-08,
+      "loss": 0.1365,
+      "step": 1671
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.235960195252969e-09,
+      "loss": 0.0849,
+      "step": 1672
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.963727109125407e-09,
+      "loss": 0.1093,
+      "step": 1673
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.785714265333809e-09,
+      "loss": 0.1422,
+      "step": 1674
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.7019261050006434e-09,
+      "loss": 0.1219,
+      "step": 1675
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.712366714018646e-09,
+      "loss": 0.1076,
+      "step": 1676
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.817039823034163e-09,
+      "loss": 0.0883,
+      "step": 1677
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0159488074416044e-09,
+      "loss": 0.1105,
+      "step": 1678
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3090966873640096e-09,
+      "loss": 0.0981,
+      "step": 1679
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6964861276363985e-09,
+      "loss": 0.0843,
+      "step": 1680
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1781194378113203e-09,
+      "loss": 0.1171,
+      "step": 1681
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.539985721366494e-10,
+      "loss": 0.0812,
+      "step": 1682
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.2412512955281037e-10,
+      "loss": 0.1296,
+      "step": 1683
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.885003536816754e-10,
+      "loss": 0.0793,
+      "step": 1684
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.712513283489095e-11,
+      "loss": 0.1052,
+      "step": 1685
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.1093,
+      "step": 1686
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.24669183790683746,
+      "eval_runtime": 119.2309,
+      "eval_samples_per_second": 2.516,
+      "eval_steps_per_second": 0.629,
+      "step": 1686
+    },
+    {
+      "epoch": 2.0,
+      "step": 1686,
+      "total_flos": 8.934629359151481e+18,
+      "train_loss": 0.20717893251850936,
+      "train_runtime": 104058.0629,
+      "train_samples_per_second": 0.519,
+      "train_steps_per_second": 0.016
+    }
+  ],
+  "max_steps": 1686,
+  "num_train_epochs": 2,
+  "total_flos": 8.934629359151481e+18,
+  "trial_name": null,
+  "trial_params": null
+}