diff --git "a/checkpoint-1134/trainer_state.json" "b/checkpoint-1134/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1134/trainer_state.json"
@@ -0,0 +1,6847 @@
+{
+  "best_metric": 0.40209120512008667,
+  "best_model_checkpoint": "./disaster-tweet-distilbert-classification/checkpoint-378",
+  "epoch": 3.0,
+  "global_step": 1134,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.7503,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.7789,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 0.7344,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.7709,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.6884,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-06,
+      "loss": 0.7087,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 0.6655,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.6978,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9e-06,
+      "loss": 0.7435,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 0.719,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.7129,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2e-05,
+      "loss": 0.7249,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.7436,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.6886,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5e-05,
+      "loss": 0.702,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.7105,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.6709,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8e-05,
+      "loss": 0.6767,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9e-05,
+      "loss": 0.6784,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 0.7016,
+      "step": 20
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1e-05,
+      "loss": 0.6308,
+      "step": 21
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.645,
+      "step": 22
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.6845,
+      "step": 23
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.4e-05,
+      "loss": 0.6891,
+      "step": 24
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5914,
+      "step": 25
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.6916,
+      "step": 26
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.5884,
+      "step": 27
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.6587,
+      "step": 28
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9e-05,
+      "loss": 0.6008,
+      "step": 29
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3e-05,
+      "loss": 0.6717,
+      "step": 30
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.1e-05,
+      "loss": 0.5795,
+      "step": 31
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.6358,
+      "step": 32
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.3e-05,
+      "loss": 0.7508,
+      "step": 33
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.6601,
+      "step": 34
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.5e-05,
+      "loss": 0.6573,
+      "step": 35
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.6e-05,
+      "loss": 0.5695,
+      "step": 36
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.7e-05,
+      "loss": 0.5535,
+      "step": 37
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.8e-05,
+      "loss": 0.5813,
+      "step": 38
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.5224,
+      "step": 39
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4e-05,
+      "loss": 0.4757,
+      "step": 40
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.1e-05,
+      "loss": 0.5529,
+      "step": 41
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.2e-05,
+      "loss": 0.4964,
+      "step": 42
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.3e-05,
+      "loss": 0.4565,
+      "step": 43
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.5848,
+      "step": 44
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.5e-05,
+      "loss": 0.7333,
+      "step": 45
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.5224,
+      "step": 46
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7e-05,
+      "loss": 0.4826,
+      "step": 47
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8e-05,
+      "loss": 0.4328,
+      "step": 48
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9e-05,
+      "loss": 0.2546,
+      "step": 49
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5e-05,
+      "loss": 0.2636,
+      "step": 50
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.998659517426274e-05,
+      "loss": 0.4116,
+      "step": 51
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.997319034852547e-05,
+      "loss": 0.3191,
+      "step": 52
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.995978552278821e-05,
+      "loss": 0.2822,
+      "step": 53
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.994638069705094e-05,
+      "loss": 0.8821,
+      "step": 54
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.993297587131368e-05,
+      "loss": 0.483,
+      "step": 55
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9919571045576406e-05,
+      "loss": 0.7164,
+      "step": 56
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.990616621983915e-05,
+      "loss": 0.4161,
+      "step": 57
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.989276139410188e-05,
+      "loss": 0.2668,
+      "step": 58
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.987935656836462e-05,
+      "loss": 0.5255,
+      "step": 59
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.986595174262735e-05,
+      "loss": 0.3784,
+      "step": 60
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.985254691689008e-05,
+      "loss": 0.5065,
+      "step": 61
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.983914209115282e-05,
+      "loss": 0.1988,
+      "step": 62
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.982573726541555e-05,
+      "loss": 0.4362,
+      "step": 63
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.981233243967829e-05,
+      "loss": 0.6619,
+      "step": 64
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.979892761394102e-05,
+      "loss": 0.3217,
+      "step": 65
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.978552278820375e-05,
+      "loss": 0.2967,
+      "step": 66
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.977211796246649e-05,
+      "loss": 0.2429,
+      "step": 67
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.975871313672922e-05,
+      "loss": 0.6642,
+      "step": 68
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.974530831099196e-05,
+      "loss": 0.56,
+      "step": 69
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.973190348525469e-05,
+      "loss": 1.2979,
+      "step": 70
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9718498659517427e-05,
+      "loss": 0.5287,
+      "step": 71
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.970509383378016e-05,
+      "loss": 0.4684,
+      "step": 72
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.96916890080429e-05,
+      "loss": 0.283,
+      "step": 73
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.967828418230563e-05,
+      "loss": 0.6818,
+      "step": 74
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.966487935656837e-05,
+      "loss": 0.6141,
+      "step": 75
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.96514745308311e-05,
+      "loss": 0.5046,
+      "step": 76
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.963806970509384e-05,
+      "loss": 0.5266,
+      "step": 77
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.962466487935657e-05,
+      "loss": 0.5944,
+      "step": 78
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.961126005361931e-05,
+      "loss": 0.5631,
+      "step": 79
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.959785522788204e-05,
+      "loss": 0.4791,
+      "step": 80
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.958445040214477e-05,
+      "loss": 0.5645,
+      "step": 81
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.957104557640751e-05,
+      "loss": 0.4349,
+      "step": 82
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.955764075067024e-05,
+      "loss": 0.3865,
+      "step": 83
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.954423592493298e-05,
+      "loss": 0.486,
+      "step": 84
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.953083109919571e-05,
+      "loss": 0.2179,
+      "step": 85
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.951742627345845e-05,
+      "loss": 0.3896,
+      "step": 86
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.950402144772118e-05,
+      "loss": 0.4247,
+      "step": 87
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.949061662198392e-05,
+      "loss": 0.4906,
+      "step": 88
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.947721179624665e-05,
+      "loss": 0.4483,
+      "step": 89
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.946380697050939e-05,
+      "loss": 0.557,
+      "step": 90
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9450402144772116e-05,
+      "loss": 0.7521,
+      "step": 91
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.943699731903486e-05,
+      "loss": 0.3103,
+      "step": 92
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9423592493297586e-05,
+      "loss": 0.757,
+      "step": 93
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.941018766756033e-05,
+      "loss": 0.8248,
+      "step": 94
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9396782841823056e-05,
+      "loss": 0.4591,
+      "step": 95
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.938337801608579e-05,
+      "loss": 0.3912,
+      "step": 96
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9369973190348526e-05,
+      "loss": 0.5289,
+      "step": 97
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.935656836461126e-05,
+      "loss": 0.3264,
+      "step": 98
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9343163538874e-05,
+      "loss": 0.2947,
+      "step": 99
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.932975871313673e-05,
+      "loss": 0.2647,
+      "step": 100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.931635388739946e-05,
+      "loss": 0.3691,
+      "step": 101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.93029490616622e-05,
+      "loss": 0.4796,
+      "step": 102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.928954423592493e-05,
+      "loss": 0.4827,
+      "step": 103
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.927613941018767e-05,
+      "loss": 0.2672,
+      "step": 104
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.92627345844504e-05,
+      "loss": 0.7456,
+      "step": 105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9249329758713136e-05,
+      "loss": 0.5206,
+      "step": 106
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.923592493297587e-05,
+      "loss": 0.3576,
+      "step": 107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9222520107238606e-05,
+      "loss": 0.2596,
+      "step": 108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.920911528150134e-05,
+      "loss": 0.4115,
+      "step": 109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9195710455764076e-05,
+      "loss": 0.3481,
+      "step": 110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.918230563002681e-05,
+      "loss": 0.4387,
+      "step": 111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.916890080428955e-05,
+      "loss": 0.5023,
+      "step": 112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.915549597855228e-05,
+      "loss": 0.5916,
+      "step": 113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.914209115281502e-05,
+      "loss": 0.5467,
+      "step": 114
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.912868632707775e-05,
+      "loss": 0.5631,
+      "step": 115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.911528150134049e-05,
+      "loss": 0.5512,
+      "step": 116
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.910187667560322e-05,
+      "loss": 0.5546,
+      "step": 117
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.908847184986595e-05,
+      "loss": 0.4209,
+      "step": 118
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.907506702412869e-05,
+      "loss": 0.6064,
+      "step": 119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.906166219839142e-05,
+      "loss": 0.5301,
+      "step": 120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.904825737265416e-05,
+      "loss": 0.436,
+      "step": 121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.903485254691689e-05,
+      "loss": 0.726,
+      "step": 122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9021447721179626e-05,
+      "loss": 0.5288,
+      "step": 123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.900804289544236e-05,
+      "loss": 0.4887,
+      "step": 124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.8994638069705097e-05,
+      "loss": 0.3752,
+      "step": 125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.898123324396783e-05,
+      "loss": 0.4895,
+      "step": 126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.896782841823057e-05,
+      "loss": 0.5046,
+      "step": 127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8954423592493295e-05,
+      "loss": 0.3953,
+      "step": 128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.894101876675604e-05,
+      "loss": 0.2015,
+      "step": 129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8927613941018765e-05,
+      "loss": 0.5165,
+      "step": 130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.891420911528151e-05,
+      "loss": 0.4237,
+      "step": 131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8900804289544236e-05,
+      "loss": 0.239,
+      "step": 132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.888739946380697e-05,
+      "loss": 0.5515,
+      "step": 133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8873994638069706e-05,
+      "loss": 0.303,
+      "step": 134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.886058981233244e-05,
+      "loss": 0.2867,
+      "step": 135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8847184986595176e-05,
+      "loss": 0.6756,
+      "step": 136
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.883378016085791e-05,
+      "loss": 0.4996,
+      "step": 137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8820375335120646e-05,
+      "loss": 0.2798,
+      "step": 138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.880697050938338e-05,
+      "loss": 0.8877,
+      "step": 139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.879356568364612e-05,
+      "loss": 0.5022,
+      "step": 140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.878016085790885e-05,
+      "loss": 0.2177,
+      "step": 141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.876675603217159e-05,
+      "loss": 0.8708,
+      "step": 142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8753351206434315e-05,
+      "loss": 0.719,
+      "step": 143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.873994638069706e-05,
+      "loss": 0.562,
+      "step": 144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8726541554959786e-05,
+      "loss": 0.492,
+      "step": 145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.871313672922253e-05,
+      "loss": 0.4637,
+      "step": 146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8699731903485256e-05,
+      "loss": 0.4132,
+      "step": 147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.868632707774799e-05,
+      "loss": 0.2889,
+      "step": 148
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8672922252010726e-05,
+      "loss": 0.2213,
+      "step": 149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.865951742627346e-05,
+      "loss": 0.9268,
+      "step": 150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8646112600536196e-05,
+      "loss": 0.2852,
+      "step": 151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.863270777479893e-05,
+      "loss": 0.4599,
+      "step": 152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.861930294906166e-05,
+      "loss": 0.1913,
+      "step": 153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.86058981233244e-05,
+      "loss": 0.4488,
+      "step": 154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.859249329758713e-05,
+      "loss": 0.9022,
+      "step": 155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.857908847184987e-05,
+      "loss": 0.5221,
+      "step": 156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.85656836461126e-05,
+      "loss": 0.2394,
+      "step": 157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8552278820375336e-05,
+      "loss": 0.3332,
+      "step": 158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.853887399463807e-05,
+      "loss": 0.4015,
+      "step": 159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8525469168900806e-05,
+      "loss": 0.4461,
+      "step": 160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.851206434316354e-05,
+      "loss": 0.337,
+      "step": 161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8498659517426276e-05,
+      "loss": 0.4908,
+      "step": 162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.848525469168901e-05,
+      "loss": 0.526,
+      "step": 163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8471849865951746e-05,
+      "loss": 0.5262,
+      "step": 164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.845844504021448e-05,
+      "loss": 0.6818,
+      "step": 165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8445040214477217e-05,
+      "loss": 0.3154,
+      "step": 166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.843163538873995e-05,
+      "loss": 0.5963,
+      "step": 167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.841823056300268e-05,
+      "loss": 0.4451,
+      "step": 168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.840482573726542e-05,
+      "loss": 0.5969,
+      "step": 169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.839142091152815e-05,
+      "loss": 0.438,
+      "step": 170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.837801608579089e-05,
+      "loss": 0.4827,
+      "step": 171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.836461126005362e-05,
+      "loss": 0.2029,
+      "step": 172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.8351206434316356e-05,
+      "loss": 0.5195,
+      "step": 173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.833780160857909e-05,
+      "loss": 0.517,
+      "step": 174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.8324396782841826e-05,
+      "loss": 0.5532,
+      "step": 175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.831099195710456e-05,
+      "loss": 0.4198,
+      "step": 176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8297587131367296e-05,
+      "loss": 0.8386,
+      "step": 177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8284182305630025e-05,
+      "loss": 0.575,
+      "step": 178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8270777479892766e-05,
+      "loss": 0.6156,
+      "step": 179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8257372654155495e-05,
+      "loss": 0.7044,
+      "step": 180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.824396782841824e-05,
+      "loss": 0.5712,
+      "step": 181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8230563002680965e-05,
+      "loss": 0.34,
+      "step": 182
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.82171581769437e-05,
+      "loss": 0.5773,
+      "step": 183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8203753351206435e-05,
+      "loss": 0.546,
+      "step": 184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.819034852546917e-05,
+      "loss": 0.3955,
+      "step": 185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8176943699731906e-05,
+      "loss": 0.5921,
+      "step": 186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.816353887399464e-05,
+      "loss": 0.3108,
+      "step": 187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8150134048257376e-05,
+      "loss": 0.5469,
+      "step": 188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.813672922252011e-05,
+      "loss": 0.64,
+      "step": 189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8123324396782846e-05,
+      "loss": 0.5153,
+      "step": 190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.810991957104558e-05,
+      "loss": 0.4719,
+      "step": 191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8096514745308316e-05,
+      "loss": 0.52,
+      "step": 192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8083109919571045e-05,
+      "loss": 0.5114,
+      "step": 193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.806970509383379e-05,
+      "loss": 0.5469,
+      "step": 194
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8056300268096515e-05,
+      "loss": 0.3435,
+      "step": 195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.804289544235926e-05,
+      "loss": 0.6469,
+      "step": 196
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8029490616621985e-05,
+      "loss": 0.6595,
+      "step": 197
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.801608579088472e-05,
+      "loss": 0.5503,
+      "step": 198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8002680965147456e-05,
+      "loss": 0.3799,
+      "step": 199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.798927613941019e-05,
+      "loss": 0.417,
+      "step": 200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7975871313672926e-05,
+      "loss": 0.5281,
+      "step": 201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.796246648793566e-05,
+      "loss": 0.3439,
+      "step": 202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.794906166219839e-05,
+      "loss": 0.5777,
+      "step": 203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.793565683646113e-05,
+      "loss": 0.5286,
+      "step": 204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.792225201072386e-05,
+      "loss": 0.4302,
+      "step": 205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.79088471849866e-05,
+      "loss": 0.5413,
+      "step": 206
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.789544235924933e-05,
+      "loss": 0.3087,
+      "step": 207
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7882037533512065e-05,
+      "loss": 0.4385,
+      "step": 208
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.78686327077748e-05,
+      "loss": 0.7137,
+      "step": 209
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7855227882037535e-05,
+      "loss": 0.6278,
+      "step": 210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.784182305630027e-05,
+      "loss": 0.2832,
+      "step": 211
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7828418230563005e-05,
+      "loss": 0.5899,
+      "step": 212
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7815013404825734e-05,
+      "loss": 0.4421,
+      "step": 213
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7801608579088476e-05,
+      "loss": 0.3411,
+      "step": 214
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7788203753351204e-05,
+      "loss": 0.4236,
+      "step": 215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7774798927613946e-05,
+      "loss": 0.4444,
+      "step": 216
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7761394101876674e-05,
+      "loss": 0.4765,
+      "step": 217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.774798927613941e-05,
+      "loss": 0.2453,
+      "step": 218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7734584450402145e-05,
+      "loss": 0.2765,
+      "step": 219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.772117962466488e-05,
+      "loss": 0.2075,
+      "step": 220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7707774798927615e-05,
+      "loss": 0.5905,
+      "step": 221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.769436997319035e-05,
+      "loss": 0.3457,
+      "step": 222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7680965147453085e-05,
+      "loss": 0.5986,
+      "step": 223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.766756032171582e-05,
+      "loss": 0.3881,
+      "step": 224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7654155495978555e-05,
+      "loss": 0.5655,
+      "step": 225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.764075067024129e-05,
+      "loss": 0.6156,
+      "step": 226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7627345844504026e-05,
+      "loss": 0.4784,
+      "step": 227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7613941018766754e-05,
+      "loss": 0.5195,
+      "step": 228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7600536193029496e-05,
+      "loss": 0.66,
+      "step": 229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7587131367292224e-05,
+      "loss": 0.4104,
+      "step": 230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7573726541554966e-05,
+      "loss": 0.3478,
+      "step": 231
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7560321715817695e-05,
+      "loss": 0.2037,
+      "step": 232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.754691689008043e-05,
+      "loss": 0.75,
+      "step": 233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7533512064343165e-05,
+      "loss": 0.4237,
+      "step": 234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.75201072386059e-05,
+      "loss": 0.2372,
+      "step": 235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7506702412868635e-05,
+      "loss": 0.7874,
+      "step": 236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.749329758713137e-05,
+      "loss": 0.5751,
+      "step": 237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.7479892761394105e-05,
+      "loss": 0.5801,
+      "step": 238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.746648793565684e-05,
+      "loss": 0.4983,
+      "step": 239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.745308310991957e-05,
+      "loss": 0.4215,
+      "step": 240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.743967828418231e-05,
+      "loss": 0.3655,
+      "step": 241
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.742627345844504e-05,
+      "loss": 0.523,
+      "step": 242
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.741286863270778e-05,
+      "loss": 0.5952,
+      "step": 243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.739946380697051e-05,
+      "loss": 0.4226,
+      "step": 244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.7386058981233244e-05,
+      "loss": 0.316,
+      "step": 245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.737265415549598e-05,
+      "loss": 0.5096,
+      "step": 246
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.7359249329758715e-05,
+      "loss": 0.435,
+      "step": 247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.734584450402145e-05,
+      "loss": 0.7516,
+      "step": 248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.7332439678284185e-05,
+      "loss": 0.531,
+      "step": 249
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.731903485254692e-05,
+      "loss": 0.4096,
+      "step": 250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.7305630026809655e-05,
+      "loss": 0.387,
+      "step": 251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.729222520107239e-05,
+      "loss": 0.5468,
+      "step": 252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7278820375335125e-05,
+      "loss": 0.4613,
+      "step": 253
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.726541554959786e-05,
+      "loss": 0.8437,
+      "step": 254
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.725201072386059e-05,
+      "loss": 0.522,
+      "step": 255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.723860589812333e-05,
+      "loss": 0.3922,
+      "step": 256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.722520107238606e-05,
+      "loss": 0.5114,
+      "step": 257
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.72117962466488e-05,
+      "loss": 0.6148,
+      "step": 258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719839142091153e-05,
+      "loss": 0.4578,
+      "step": 259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7184986595174265e-05,
+      "loss": 0.6286,
+      "step": 260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7171581769437e-05,
+      "loss": 0.5883,
+      "step": 261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7158176943699735e-05,
+      "loss": 0.5634,
+      "step": 262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.714477211796247e-05,
+      "loss": 0.4085,
+      "step": 263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7131367292225205e-05,
+      "loss": 0.2988,
+      "step": 264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7117962466487934e-05,
+      "loss": 0.6353,
+      "step": 265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7104557640750675e-05,
+      "loss": 0.4598,
+      "step": 266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7091152815013404e-05,
+      "loss": 0.5072,
+      "step": 267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7077747989276146e-05,
+      "loss": 0.49,
+      "step": 268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7064343163538874e-05,
+      "loss": 0.7225,
+      "step": 269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.705093833780161e-05,
+      "loss": 0.5332,
+      "step": 270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7037533512064344e-05,
+      "loss": 0.6064,
+      "step": 271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.702412868632708e-05,
+      "loss": 0.3518,
+      "step": 272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7010723860589815e-05,
+      "loss": 0.3673,
+      "step": 273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.699731903485255e-05,
+      "loss": 0.4688,
+      "step": 274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6983914209115285e-05,
+      "loss": 0.5389,
+      "step": 275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.697050938337802e-05,
+      "loss": 0.374,
+      "step": 276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.6957104557640755e-05,
+      "loss": 0.475,
+      "step": 277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.694369973190349e-05,
+      "loss": 0.5397,
+      "step": 278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6930294906166225e-05,
+      "loss": 0.3821,
+      "step": 279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6916890080428954e-05,
+      "loss": 0.3372,
+      "step": 280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.6903485254691696e-05,
+      "loss": 0.6652,
+      "step": 281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6890080428954424e-05,
+      "loss": 0.2894,
+      "step": 282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6876675603217166e-05,
+      "loss": 0.5639,
+      "step": 283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.6863270777479894e-05,
+      "loss": 0.353,
+      "step": 284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.684986595174263e-05,
+      "loss": 0.2932,
+      "step": 285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6836461126005364e-05,
+      "loss": 0.467,
+      "step": 286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.68230563002681e-05,
+      "loss": 0.4732,
+      "step": 287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.6809651474530835e-05,
+      "loss": 0.1808,
+      "step": 288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.679624664879357e-05,
+      "loss": 0.6031,
+      "step": 289
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.67828418230563e-05,
+      "loss": 0.2555,
+      "step": 290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.676943699731904e-05,
+      "loss": 0.4041,
+      "step": 291
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.675603217158177e-05,
+      "loss": 0.7822,
+      "step": 292
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.674262734584451e-05,
+      "loss": 0.138,
+      "step": 293
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.672922252010724e-05,
+      "loss": 0.2746,
+      "step": 294
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6715817694369974e-05,
+      "loss": 0.5835,
+      "step": 295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.670241286863271e-05,
+      "loss": 0.2367,
+      "step": 296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6689008042895444e-05,
+      "loss": 0.3247,
+      "step": 297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.667560321715818e-05,
+      "loss": 0.306,
+      "step": 298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6662198391420914e-05,
+      "loss": 0.2825,
+      "step": 299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.664879356568364e-05,
+      "loss": 0.6102,
+      "step": 300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6635388739946385e-05,
+      "loss": 0.3613,
+      "step": 301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.662198391420911e-05,
+      "loss": 0.6327,
+      "step": 302
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6608579088471855e-05,
+      "loss": 0.4059,
+      "step": 303
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.659517426273458e-05,
+      "loss": 0.4027,
+      "step": 304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.658176943699732e-05,
+      "loss": 0.9133,
+      "step": 305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6568364611260054e-05,
+      "loss": 0.1869,
+      "step": 306
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.655495978552279e-05,
+      "loss": 0.3987,
+      "step": 307
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.6541554959785524e-05,
+      "loss": 0.6114,
+      "step": 308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.652815013404826e-05,
+      "loss": 0.4406,
+      "step": 309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.6514745308310994e-05,
+      "loss": 0.2954,
+      "step": 310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.650134048257373e-05,
+      "loss": 0.5067,
+      "step": 311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.6487935656836464e-05,
+      "loss": 0.3985,
+      "step": 312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.64745308310992e-05,
+      "loss": 0.3756,
+      "step": 313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.6461126005361935e-05,
+      "loss": 0.2618,
+      "step": 314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.644772117962466e-05,
+      "loss": 0.4992,
+      "step": 315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.6434316353887405e-05,
+      "loss": 0.8224,
+      "step": 316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.642091152815013e-05,
+      "loss": 0.3425,
+      "step": 317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.6407506702412875e-05,
+      "loss": 0.4062,
+      "step": 318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.6394101876675603e-05,
+      "loss": 0.4748,
+      "step": 319
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.638069705093834e-05,
+      "loss": 0.6857,
+      "step": 320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.6367292225201074e-05,
+      "loss": 0.5368,
+      "step": 321
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.635388739946381e-05,
+      "loss": 0.5571,
+      "step": 322
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.6340482573726544e-05,
+      "loss": 0.3045,
+      "step": 323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.632707774798928e-05,
+      "loss": 0.3189,
+      "step": 324
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.631367292225201e-05,
+      "loss": 0.3033,
+      "step": 325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.630026809651475e-05,
+      "loss": 0.4659,
+      "step": 326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.628686327077748e-05,
+      "loss": 0.4058,
+      "step": 327
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.627345844504022e-05,
+      "loss": 0.6745,
+      "step": 328
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.626005361930295e-05,
+      "loss": 0.3259,
+      "step": 329
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.624664879356568e-05,
+      "loss": 0.5126,
+      "step": 330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.623324396782842e-05,
+      "loss": 0.2759,
+      "step": 331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.621983914209115e-05,
+      "loss": 0.2512,
+      "step": 332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.620643431635389e-05,
+      "loss": 0.3046,
+      "step": 333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.6193029490616624e-05,
+      "loss": 0.3931,
+      "step": 334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.617962466487936e-05,
+      "loss": 0.4838,
+      "step": 335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.6166219839142094e-05,
+      "loss": 0.2925,
+      "step": 336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.615281501340483e-05,
+      "loss": 0.4481,
+      "step": 337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.6139410187667564e-05,
+      "loss": 0.4528,
+      "step": 338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.61260053619303e-05,
+      "loss": 0.2934,
+      "step": 339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.611260053619303e-05,
+      "loss": 0.609,
+      "step": 340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.609919571045577e-05,
+      "loss": 0.8988,
+      "step": 341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.60857908847185e-05,
+      "loss": 1.1222,
+      "step": 342
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.607238605898124e-05,
+      "loss": 0.3265,
+      "step": 343
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.605898123324397e-05,
+      "loss": 0.4722,
+      "step": 344
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.60455764075067e-05,
+      "loss": 0.2791,
+      "step": 345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.603217158176944e-05,
+      "loss": 0.4183,
+      "step": 346
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6018766756032174e-05,
+      "loss": 0.5323,
+      "step": 347
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.600536193029491e-05,
+      "loss": 0.6108,
+      "step": 348
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.5991957104557644e-05,
+      "loss": 0.2875,
+      "step": 349
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.597855227882037e-05,
+      "loss": 0.6642,
+      "step": 350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5965147453083114e-05,
+      "loss": 0.5244,
+      "step": 351
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.595174262734584e-05,
+      "loss": 0.5562,
+      "step": 352
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.5938337801608584e-05,
+      "loss": 0.3972,
+      "step": 353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.592493297587131e-05,
+      "loss": 0.4532,
+      "step": 354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.591152815013405e-05,
+      "loss": 0.3368,
+      "step": 355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.589812332439678e-05,
+      "loss": 0.171,
+      "step": 356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.588471849865952e-05,
+      "loss": 0.4036,
+      "step": 357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.587131367292225e-05,
+      "loss": 0.4305,
+      "step": 358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.585790884718499e-05,
+      "loss": 0.2643,
+      "step": 359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.5844504021447723e-05,
+      "loss": 0.3782,
+      "step": 360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.583109919571046e-05,
+      "loss": 0.1673,
+      "step": 361
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5817694369973194e-05,
+      "loss": 0.3969,
+      "step": 362
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.580428954423593e-05,
+      "loss": 0.3249,
+      "step": 363
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5790884718498664e-05,
+      "loss": 0.1656,
+      "step": 364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.57774798927614e-05,
+      "loss": 0.4551,
+      "step": 365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5764075067024134e-05,
+      "loss": 0.6075,
+      "step": 366
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.575067024128686e-05,
+      "loss": 0.4699,
+      "step": 367
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5737265415549605e-05,
+      "loss": 0.5752,
+      "step": 368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.572386058981233e-05,
+      "loss": 0.3114,
+      "step": 369
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5710455764075075e-05,
+      "loss": 0.7407,
+      "step": 370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.56970509383378e-05,
+      "loss": 0.6427,
+      "step": 371
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.568364611260054e-05,
+      "loss": 0.5021,
+      "step": 372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5670241286863273e-05,
+      "loss": 0.4209,
+      "step": 373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.565683646112601e-05,
+      "loss": 0.5957,
+      "step": 374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5643431635388744e-05,
+      "loss": 0.495,
+      "step": 375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.563002680965148e-05,
+      "loss": 0.3101,
+      "step": 376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.561662198391421e-05,
+      "loss": 0.472,
+      "step": 377
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.560321715817695e-05,
+      "loss": 0.4607,
+      "step": 378
+    },
+    {
+      "epoch": 1.0,
+      "eval_f1": 0.7914963205233032,
+      "eval_loss": 0.40209120512008667,
+      "eval_runtime": 2.1585,
+      "eval_samples_per_second": 700.949,
+      "eval_steps_per_second": 44.012,
+      "step": 378
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.558981233243968e-05,
+      "loss": 0.2422,
+      "step": 379
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.557640750670242e-05,
+      "loss": 0.2074,
+      "step": 380
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.556300268096515e-05,
+      "loss": 0.4469,
+      "step": 381
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.554959785522788e-05,
+      "loss": 0.3121,
+      "step": 382
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.553619302949062e-05,
+      "loss": 0.3634,
+      "step": 383
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.552278820375335e-05,
+      "loss": 0.2139,
+      "step": 384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.550938337801609e-05,
+      "loss": 0.5273,
+      "step": 385
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.549597855227882e-05,
+      "loss": 0.1228,
+      "step": 386
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.548257372654156e-05,
+      "loss": 0.2459,
+      "step": 387
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.5469168900804294e-05,
+      "loss": 0.443,
+      "step": 388
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.545576407506703e-05,
+      "loss": 0.364,
+      "step": 389
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.5442359249329764e-05,
+      "loss": 0.4071,
+      "step": 390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.54289544235925e-05,
+      "loss": 0.3876,
+      "step": 391
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.541554959785523e-05,
+      "loss": 0.5239,
+      "step": 392
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.540214477211797e-05,
+      "loss": 0.4548,
+      "step": 393
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.53887399463807e-05,
+      "loss": 0.3588,
+      "step": 394
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.537533512064344e-05,
+      "loss": 0.2693,
+      "step": 395
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.536193029490617e-05,
+      "loss": 0.3773,
+      "step": 396
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.53485254691689e-05,
+      "loss": 0.3982,
+      "step": 397
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.533512064343164e-05,
+      "loss": 0.2074,
+      "step": 398
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.532171581769437e-05,
+      "loss": 0.3322,
+      "step": 399
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.530831099195711e-05,
+      "loss": 0.118,
+      "step": 400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5294906166219844e-05,
+      "loss": 0.6216,
+      "step": 401
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.528150134048257e-05,
+      "loss": 0.4028,
+      "step": 402
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5268096514745314e-05,
+      "loss": 0.3179,
+      "step": 403
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.525469168900804e-05,
+      "loss": 0.2815,
+      "step": 404
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5241286863270784e-05,
+      "loss": 0.2366,
+      "step": 405
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.522788203753351e-05,
+      "loss": 0.14,
+      "step": 406
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.521447721179625e-05,
+      "loss": 0.1255,
+      "step": 407
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.520107238605898e-05,
+      "loss": 0.1482,
+      "step": 408
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.518766756032172e-05,
+      "loss": 0.665,
+      "step": 409
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.517426273458445e-05,
+      "loss": 0.2535,
+      "step": 410
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.516085790884719e-05,
+      "loss": 0.1716,
+      "step": 411
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5147453083109916e-05,
+      "loss": 0.0765,
+      "step": 412
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.513404825737266e-05,
+      "loss": 0.1238,
+      "step": 413
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.512064343163539e-05,
+      "loss": 0.3877,
+      "step": 414
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.510723860589813e-05,
+      "loss": 0.206,
+      "step": 415
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.509383378016086e-05,
+      "loss": 0.1822,
+      "step": 416
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.508042895442359e-05,
+      "loss": 0.2294,
+      "step": 417
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.506702412868633e-05,
+      "loss": 0.7083,
+      "step": 418
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.505361930294906e-05,
+      "loss": 0.325,
+      "step": 419
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.50402144772118e-05,
+      "loss": 0.2378,
+      "step": 420
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.502680965147453e-05,
+      "loss": 0.2494,
+      "step": 421
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.501340482573727e-05,
+      "loss": 0.3812,
+      "step": 422
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5e-05,
+      "loss": 0.1491,
+      "step": 423
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.498659517426274e-05,
+      "loss": 0.0833,
+      "step": 424
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.497319034852547e-05,
+      "loss": 0.4637,
+      "step": 425
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.495978552278821e-05,
+      "loss": 0.2594,
+      "step": 426
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.4946380697050937e-05,
+      "loss": 0.4136,
+      "step": 427
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.493297587131368e-05,
+      "loss": 0.4962,
+      "step": 428
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.491957104557641e-05,
+      "loss": 0.0842,
+      "step": 429
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.490616621983915e-05,
+      "loss": 0.4761,
+      "step": 430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.489276139410188e-05,
+      "loss": 0.4351,
+      "step": 431
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.487935656836461e-05,
+      "loss": 0.4226,
+      "step": 432
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.486595174262735e-05,
+      "loss": 0.5714,
+      "step": 433
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.485254691689008e-05,
+      "loss": 0.2034,
+      "step": 434
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.483914209115282e-05,
+      "loss": 0.3354,
+      "step": 435
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.482573726541555e-05,
+      "loss": 0.7503,
+      "step": 436
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.481233243967828e-05,
+      "loss": 0.1542,
+      "step": 437
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.479892761394102e-05,
+      "loss": 0.3067,
+      "step": 438
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.478552278820375e-05,
+      "loss": 0.4134,
+      "step": 439
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.477211796246649e-05,
+      "loss": 0.1458,
+      "step": 440
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.475871313672922e-05,
+      "loss": 0.2814,
+      "step": 441
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.474530831099196e-05,
+      "loss": 0.1751,
+      "step": 442
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.473190348525469e-05,
+      "loss": 0.4144,
+      "step": 443
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.471849865951743e-05,
+      "loss": 0.3275,
+      "step": 444
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.470509383378016e-05,
+      "loss": 0.3083,
+      "step": 445
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.46916890080429e-05,
+      "loss": 0.4185,
+      "step": 446
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.467828418230563e-05,
+      "loss": 0.3181,
+      "step": 447
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.466487935656837e-05,
+      "loss": 0.438,
+      "step": 448
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.46514745308311e-05,
+      "loss": 0.4509,
+      "step": 449
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.463806970509384e-05,
+      "loss": 0.4597,
+      "step": 450
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.462466487935657e-05,
+      "loss": 0.3123,
+      "step": 451
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.46112600536193e-05,
+      "loss": 0.376,
+      "step": 452
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.459785522788204e-05,
+      "loss": 0.2392,
+      "step": 453
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.458445040214477e-05,
+      "loss": 0.4451,
+      "step": 454
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4571045576407513e-05,
+      "loss": 0.2436,
+      "step": 455
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.455764075067024e-05,
+      "loss": 0.2944,
+      "step": 456
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.454423592493298e-05,
+      "loss": 0.323,
+      "step": 457
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.453083109919571e-05,
+      "loss": 0.1966,
+      "step": 458
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.451742627345845e-05,
+      "loss": 0.273,
+      "step": 459
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.450402144772118e-05,
+      "loss": 0.3498,
+      "step": 460
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.449061662198392e-05,
+      "loss": 0.458,
+      "step": 461
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4477211796246646e-05,
+      "loss": 0.1607,
+      "step": 462
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.446380697050939e-05,
+      "loss": 0.5712,
+      "step": 463
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4450402144772116e-05,
+      "loss": 0.0908,
+      "step": 464
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.443699731903486e-05,
+      "loss": 0.2625,
+      "step": 465
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4423592493297586e-05,
+      "loss": 0.368,
+      "step": 466
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.441018766756032e-05,
+      "loss": 0.5096,
+      "step": 467
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4396782841823057e-05,
+      "loss": 0.209,
+      "step": 468
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.438337801608579e-05,
+      "loss": 0.5235,
+      "step": 469
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.436997319034853e-05,
+      "loss": 0.5116,
+      "step": 470
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.435656836461126e-05,
+      "loss": 0.49,
+      "step": 471
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.4343163538874e-05,
+      "loss": 0.4409,
+      "step": 472
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.432975871313673e-05,
+      "loss": 0.1324,
+      "step": 473
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.431635388739947e-05,
+      "loss": 0.1257,
+      "step": 474
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.43029490616622e-05,
+      "loss": 0.3844,
+      "step": 475
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.428954423592494e-05,
+      "loss": 0.2942,
+      "step": 476
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.4276139410187666e-05,
+      "loss": 0.316,
+      "step": 477
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.426273458445041e-05,
+      "loss": 0.3724,
+      "step": 478
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.4249329758713136e-05,
+      "loss": 0.4498,
+      "step": 479
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.423592493297588e-05,
+      "loss": 0.2677,
+      "step": 480
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.4222520107238607e-05,
+      "loss": 0.3331,
+      "step": 481
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.420911528150134e-05,
+      "loss": 0.4375,
+      "step": 482
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.419571045576408e-05,
+      "loss": 0.3097,
+      "step": 483
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.418230563002681e-05,
+      "loss": 0.4438,
+      "step": 484
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.416890080428955e-05,
+      "loss": 0.3648,
+      "step": 485
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.415549597855228e-05,
+      "loss": 0.3023,
+      "step": 486
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.414209115281502e-05,
+      "loss": 0.2963,
+      "step": 487
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.412868632707775e-05,
+      "loss": 0.1749,
+      "step": 488
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.411528150134048e-05,
+      "loss": 0.2006,
+      "step": 489
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.410187667560322e-05,
+      "loss": 0.0923,
+      "step": 490
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.408847184986595e-05,
+      "loss": 0.2665,
+      "step": 491
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.407506702412869e-05,
+      "loss": 0.3378,
+      "step": 492
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.406166219839142e-05,
+      "loss": 0.2543,
+      "step": 493
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.4048257372654156e-05,
+      "loss": 0.2521,
+      "step": 494
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.403485254691689e-05,
+      "loss": 0.1616,
+      "step": 495
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.402144772117963e-05,
+      "loss": 0.1445,
+      "step": 496
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.400804289544236e-05,
+      "loss": 0.5352,
+      "step": 497
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.39946380697051e-05,
+      "loss": 0.0654,
+      "step": 498
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.398123324396783e-05,
+      "loss": 0.4354,
+      "step": 499
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.396782841823057e-05,
+      "loss": 0.3893,
+      "step": 500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.39544235924933e-05,
+      "loss": 0.2779,
+      "step": 501
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.394101876675604e-05,
+      "loss": 0.3702,
+      "step": 502
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.392761394101877e-05,
+      "loss": 0.3899,
+      "step": 503
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.39142091152815e-05,
+      "loss": 0.3091,
+      "step": 504
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.390080428954424e-05,
+      "loss": 0.2143,
+      "step": 505
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.388739946380697e-05,
+      "loss": 0.2551,
+      "step": 506
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.387399463806971e-05,
+      "loss": 0.2227,
+      "step": 507
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.386058981233244e-05,
+      "loss": 0.1383,
+      "step": 508
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.384718498659518e-05,
+      "loss": 0.3463,
+      "step": 509
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.383378016085791e-05,
+      "loss": 0.1104,
+      "step": 510
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.382037533512065e-05,
+      "loss": 0.2439,
+      "step": 511
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.380697050938338e-05,
+      "loss": 0.0725,
+      "step": 512
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.379356568364612e-05,
+      "loss": 0.4928,
+      "step": 513
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3780160857908846e-05,
+      "loss": 0.0903,
+      "step": 514
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.376675603217159e-05,
+      "loss": 0.1808,
+      "step": 515
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3753351206434316e-05,
+      "loss": 0.617,
+      "step": 516
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.373994638069706e-05,
+      "loss": 0.0333,
+      "step": 517
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3726541554959786e-05,
+      "loss": 0.1113,
+      "step": 518
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.371313672922252e-05,
+      "loss": 0.4604,
+      "step": 519
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3699731903485256e-05,
+      "loss": 0.4267,
+      "step": 520
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.368632707774799e-05,
+      "loss": 0.2621,
+      "step": 521
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.3672922252010727e-05,
+      "loss": 0.3236,
+      "step": 522
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.365951742627346e-05,
+      "loss": 0.0559,
+      "step": 523
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.364611260053619e-05,
+      "loss": 0.0801,
+      "step": 524
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.363270777479893e-05,
+      "loss": 0.0518,
+      "step": 525
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.361930294906166e-05,
+      "loss": 0.0618,
+      "step": 526
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.36058981233244e-05,
+      "loss": 0.3043,
+      "step": 527
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.359249329758713e-05,
+      "loss": 0.4693,
+      "step": 528
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3579088471849866e-05,
+      "loss": 0.2978,
+      "step": 529
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.35656836461126e-05,
+      "loss": 0.0817,
+      "step": 530
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.3552278820375336e-05,
+      "loss": 0.0784,
+      "step": 531
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.353887399463807e-05,
+      "loss": 0.5389,
+      "step": 532
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.3525469168900806e-05,
+      "loss": 0.2337,
+      "step": 533
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.351206434316354e-05,
+      "loss": 0.3189,
+      "step": 534
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.3498659517426276e-05,
+      "loss": 0.2503,
+      "step": 535
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.348525469168901e-05,
+      "loss": 0.333,
+      "step": 536
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.347184986595175e-05,
+      "loss": 0.2311,
+      "step": 537
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.345844504021448e-05,
+      "loss": 0.5196,
+      "step": 538
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.344504021447721e-05,
+      "loss": 0.5466,
+      "step": 539
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.343163538873995e-05,
+      "loss": 0.4296,
+      "step": 540
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.341823056300268e-05,
+      "loss": 0.6559,
+      "step": 541
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.340482573726542e-05,
+      "loss": 0.6134,
+      "step": 542
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.339142091152815e-05,
+      "loss": 0.5614,
+      "step": 543
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.3378016085790886e-05,
+      "loss": 0.3143,
+      "step": 544
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.336461126005362e-05,
+      "loss": 0.564,
+      "step": 545
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.3351206434316356e-05,
+      "loss": 0.2141,
+      "step": 546
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.333780160857909e-05,
+      "loss": 0.5886,
+      "step": 547
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.3324396782841826e-05,
+      "loss": 0.2129,
+      "step": 548
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.3310991957104555e-05,
+      "loss": 0.2491,
+      "step": 549
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.32975871313673e-05,
+      "loss": 0.4023,
+      "step": 550
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.3284182305630025e-05,
+      "loss": 0.2159,
+      "step": 551
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.327077747989277e-05,
+      "loss": 0.2975,
+      "step": 552
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.3257372654155495e-05,
+      "loss": 0.165,
+      "step": 553
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.324396782841823e-05,
+      "loss": 0.063,
+      "step": 554
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.3230563002680966e-05,
+      "loss": 0.4209,
+      "step": 555
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.32171581769437e-05,
+      "loss": 0.2704,
+      "step": 556
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.3203753351206436e-05,
+      "loss": 0.6455,
+      "step": 557
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.319034852546917e-05,
+      "loss": 0.3354,
+      "step": 558
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.3176943699731906e-05,
+      "loss": 0.4069,
+      "step": 559
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.316353887399464e-05,
+      "loss": 0.2334,
+      "step": 560
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.3150134048257376e-05,
+      "loss": 0.4192,
+      "step": 561
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.313672922252011e-05,
+      "loss": 0.2942,
+      "step": 562
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.3123324396782847e-05,
+      "loss": 0.2503,
+      "step": 563
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.3109919571045575e-05,
+      "loss": 0.15,
+      "step": 564
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.309651474530832e-05,
+      "loss": 0.6895,
+      "step": 565
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.3083109919571045e-05,
+      "loss": 0.2462,
+      "step": 566
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.306970509383379e-05,
+      "loss": 0.142,
+      "step": 567
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.3056300268096515e-05,
+      "loss": 0.4838,
+      "step": 568
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.304289544235925e-05,
+      "loss": 0.4714,
+      "step": 569
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.3029490616621986e-05,
+      "loss": 0.4916,
+      "step": 570
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.301608579088472e-05,
+      "loss": 0.2001,
+      "step": 571
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.3002680965147456e-05,
+      "loss": 0.3897,
+      "step": 572
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.298927613941019e-05,
+      "loss": 0.1723,
+      "step": 573
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.297587131367292e-05,
+      "loss": 0.4095,
+      "step": 574
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.296246648793566e-05,
+      "loss": 0.2615,
+      "step": 575
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.294906166219839e-05,
+      "loss": 0.221,
+      "step": 576
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.293565683646113e-05,
+      "loss": 0.1373,
+      "step": 577
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.292225201072386e-05,
+      "loss": 0.313,
+      "step": 578
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.2908847184986595e-05,
+      "loss": 0.4608,
+      "step": 579
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.289544235924933e-05,
+      "loss": 0.2894,
+      "step": 580
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2882037533512065e-05,
+      "loss": 0.5123,
+      "step": 581
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.28686327077748e-05,
+      "loss": 0.3287,
+      "step": 582
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.2855227882037536e-05,
+      "loss": 0.4634,
+      "step": 583
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.284182305630027e-05,
+      "loss": 0.0753,
+      "step": 584
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2828418230563006e-05,
+      "loss": 0.5287,
+      "step": 585
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.281501340482574e-05,
+      "loss": 0.5483,
+      "step": 586
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2801608579088476e-05,
+      "loss": 0.1076,
+      "step": 587
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.278820375335121e-05,
+      "loss": 0.232,
+      "step": 588
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.277479892761394e-05,
+      "loss": 0.4706,
+      "step": 589
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.276139410187668e-05,
+      "loss": 0.5791,
+      "step": 590
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.274798927613941e-05,
+      "loss": 0.4995,
+      "step": 591
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.273458445040215e-05,
+      "loss": 0.3419,
+      "step": 592
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.272117962466488e-05,
+      "loss": 0.2728,
+      "step": 593
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.2707774798927615e-05,
+      "loss": 0.3598,
+      "step": 594
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.269436997319035e-05,
+      "loss": 0.2669,
+      "step": 595
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2680965147453086e-05,
+      "loss": 0.4091,
+      "step": 596
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.266756032171582e-05,
+      "loss": 0.7653,
+      "step": 597
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2654155495978556e-05,
+      "loss": 0.4323,
+      "step": 598
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2640750670241284e-05,
+      "loss": 0.2147,
+      "step": 599
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2627345844504026e-05,
+      "loss": 0.2229,
+      "step": 600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2613941018766754e-05,
+      "loss": 0.1477,
+      "step": 601
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2600536193029496e-05,
+      "loss": 0.4813,
+      "step": 602
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2587131367292225e-05,
+      "loss": 0.2719,
+      "step": 603
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.257372654155496e-05,
+      "loss": 0.1577,
+      "step": 604
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.2560321715817695e-05,
+      "loss": 0.5718,
+      "step": 605
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.254691689008043e-05,
+      "loss": 0.1481,
+      "step": 606
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2533512064343165e-05,
+      "loss": 0.3548,
+      "step": 607
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.25201072386059e-05,
+      "loss": 0.1907,
+      "step": 608
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.250670241286863e-05,
+      "loss": 0.4215,
+      "step": 609
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.249329758713137e-05,
+      "loss": 0.0681,
+      "step": 610
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.24798927613941e-05,
+      "loss": 0.514,
+      "step": 611
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.246648793565684e-05,
+      "loss": 0.1242,
+      "step": 612
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.245308310991957e-05,
+      "loss": 0.217,
+      "step": 613
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.243967828418231e-05,
+      "loss": 0.1641,
+      "step": 614
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.242627345844504e-05,
+      "loss": 0.7095,
+      "step": 615
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.2412868632707775e-05,
+      "loss": 0.359,
+      "step": 616
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.239946380697051e-05,
+      "loss": 0.94,
+      "step": 617
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.2386058981233245e-05,
+      "loss": 0.5194,
+      "step": 618
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.237265415549598e-05,
+      "loss": 0.3398,
+      "step": 619
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.2359249329758715e-05,
+      "loss": 0.2305,
+      "step": 620
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.234584450402145e-05,
+      "loss": 0.6045,
+      "step": 621
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.2332439678284185e-05,
+      "loss": 0.3196,
+      "step": 622
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.231903485254692e-05,
+      "loss": 0.0753,
+      "step": 623
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.2305630026809656e-05,
+      "loss": 0.2732,
+      "step": 624
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.229222520107239e-05,
+      "loss": 0.4372,
+      "step": 625
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.227882037533512e-05,
+      "loss": 0.1299,
+      "step": 626
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.226541554959786e-05,
+      "loss": 0.2928,
+      "step": 627
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.225201072386059e-05,
+      "loss": 0.2028,
+      "step": 628
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.223860589812333e-05,
+      "loss": 0.2725,
+      "step": 629
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.222520107238606e-05,
+      "loss": 0.0851,
+      "step": 630
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.2211796246648795e-05,
+      "loss": 0.1471,
+      "step": 631
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.219839142091153e-05,
+      "loss": 0.1986,
+      "step": 632
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.2184986595174265e-05,
+      "loss": 0.5156,
+      "step": 633
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.2171581769437e-05,
+      "loss": 0.5286,
+      "step": 634
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.2158176943699735e-05,
+      "loss": 0.3635,
+      "step": 635
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.2144772117962464e-05,
+      "loss": 0.1407,
+      "step": 636
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.2131367292225206e-05,
+      "loss": 0.1042,
+      "step": 637
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.2117962466487934e-05,
+      "loss": 0.1553,
+      "step": 638
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.2104557640750676e-05,
+      "loss": 0.1665,
+      "step": 639
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.2091152815013404e-05,
+      "loss": 0.3706,
+      "step": 640
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.207774798927614e-05,
+      "loss": 0.6195,
+      "step": 641
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.2064343163538874e-05,
+      "loss": 0.1341,
+      "step": 642
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.205093833780161e-05,
+      "loss": 0.5384,
+      "step": 643
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.2037533512064345e-05,
+      "loss": 0.2802,
+      "step": 644
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.202412868632708e-05,
+      "loss": 0.3812,
+      "step": 645
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.2010723860589815e-05,
+      "loss": 0.2433,
+      "step": 646
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.199731903485255e-05,
+      "loss": 0.0642,
+      "step": 647
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.1983914209115285e-05,
+      "loss": 0.0547,
+      "step": 648
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.197050938337802e-05,
+      "loss": 0.4388,
+      "step": 649
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1957104557640756e-05,
+      "loss": 0.8228,
+      "step": 650
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1943699731903484e-05,
+      "loss": 0.6453,
+      "step": 651
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1930294906166226e-05,
+      "loss": 0.3367,
+      "step": 652
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1916890080428954e-05,
+      "loss": 0.2139,
+      "step": 653
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1903485254691696e-05,
+      "loss": 0.2144,
+      "step": 654
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1890080428954424e-05,
+      "loss": 0.3894,
+      "step": 655
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.187667560321716e-05,
+      "loss": 0.0891,
+      "step": 656
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1863270777479895e-05,
+      "loss": 0.1287,
+      "step": 657
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.184986595174263e-05,
+      "loss": 0.2826,
+      "step": 658
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1836461126005365e-05,
+      "loss": 0.474,
+      "step": 659
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.18230563002681e-05,
+      "loss": 0.4228,
+      "step": 660
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.180965147453083e-05,
+      "loss": 0.4952,
+      "step": 661
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.179624664879357e-05,
+      "loss": 0.173,
+      "step": 662
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.17828418230563e-05,
+      "loss": 0.363,
+      "step": 663
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.176943699731904e-05,
+      "loss": 0.4404,
+      "step": 664
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.175603217158177e-05,
+      "loss": 0.486,
+      "step": 665
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1742627345844504e-05,
+      "loss": 0.4463,
+      "step": 666
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.172922252010724e-05,
+      "loss": 0.2409,
+      "step": 667
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1715817694369974e-05,
+      "loss": 0.5291,
+      "step": 668
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.170241286863271e-05,
+      "loss": 0.069,
+      "step": 669
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1689008042895445e-05,
+      "loss": 0.4162,
+      "step": 670
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.167560321715818e-05,
+      "loss": 0.6171,
+      "step": 671
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1662198391420915e-05,
+      "loss": 0.3097,
+      "step": 672
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.164879356568365e-05,
+      "loss": 0.5109,
+      "step": 673
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1635388739946385e-05,
+      "loss": 0.2169,
+      "step": 674
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.162198391420912e-05,
+      "loss": 0.2406,
+      "step": 675
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.160857908847185e-05,
+      "loss": 0.1853,
+      "step": 676
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.159517426273459e-05,
+      "loss": 0.5743,
+      "step": 677
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.158176943699732e-05,
+      "loss": 0.5432,
+      "step": 678
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.156836461126006e-05,
+      "loss": 0.2033,
+      "step": 679
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.155495978552279e-05,
+      "loss": 0.3848,
+      "step": 680
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.1541554959785524e-05,
+      "loss": 0.1721,
+      "step": 681
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.152815013404826e-05,
+      "loss": 0.3793,
+      "step": 682
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1514745308310994e-05,
+      "loss": 0.3848,
+      "step": 683
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.150134048257373e-05,
+      "loss": 0.186,
+      "step": 684
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1487935656836465e-05,
+      "loss": 0.2692,
+      "step": 685
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.147453083109919e-05,
+      "loss": 0.3839,
+      "step": 686
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1461126005361935e-05,
+      "loss": 0.3037,
+      "step": 687
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.144772117962466e-05,
+      "loss": 0.6401,
+      "step": 688
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1434316353887405e-05,
+      "loss": 0.1173,
+      "step": 689
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.1420911528150134e-05,
+      "loss": 0.3217,
+      "step": 690
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.140750670241287e-05,
+      "loss": 0.2358,
+      "step": 691
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.1394101876675604e-05,
+      "loss": 0.7696,
+      "step": 692
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.138069705093834e-05,
+      "loss": 0.2288,
+      "step": 693
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.1367292225201074e-05,
+      "loss": 0.2575,
+      "step": 694
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.135388739946381e-05,
+      "loss": 0.1201,
+      "step": 695
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.1340482573726544e-05,
+      "loss": 0.2034,
+      "step": 696
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.132707774798928e-05,
+      "loss": 0.1142,
+      "step": 697
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.1313672922252015e-05,
+      "loss": 0.5671,
+      "step": 698
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.130026809651475e-05,
+      "loss": 0.3132,
+      "step": 699
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.1286863270777485e-05,
+      "loss": 0.4266,
+      "step": 700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.127345844504021e-05,
+      "loss": 0.1354,
+      "step": 701
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.1260053619302955e-05,
+      "loss": 0.2867,
+      "step": 702
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.1246648793565684e-05,
+      "loss": 0.1839,
+      "step": 703
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.1233243967828425e-05,
+      "loss": 0.4741,
+      "step": 704
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.1219839142091154e-05,
+      "loss": 0.2909,
+      "step": 705
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.120643431635389e-05,
+      "loss": 0.2705,
+      "step": 706
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.1193029490616624e-05,
+      "loss": 0.1354,
+      "step": 707
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.117962466487936e-05,
+      "loss": 0.4801,
+      "step": 708
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.1166219839142094e-05,
+      "loss": 0.189,
+      "step": 709
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.115281501340483e-05,
+      "loss": 0.3204,
+      "step": 710
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.113941018766756e-05,
+      "loss": 0.4358,
+      "step": 711
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.11260053619303e-05,
+      "loss": 0.9474,
+      "step": 712
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.111260053619303e-05,
+      "loss": 0.2102,
+      "step": 713
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.109919571045577e-05,
+      "loss": 0.3927,
+      "step": 714
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.10857908847185e-05,
+      "loss": 0.139,
+      "step": 715
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.1072386058981233e-05,
+      "loss": 0.3575,
+      "step": 716
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.105898123324397e-05,
+      "loss": 0.7534,
+      "step": 717
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.1045576407506704e-05,
+      "loss": 0.1134,
+      "step": 718
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.103217158176944e-05,
+      "loss": 0.2136,
+      "step": 719
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.1018766756032174e-05,
+      "loss": 0.4344,
+      "step": 720
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.10053619302949e-05,
+      "loss": 0.0695,
+      "step": 721
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.0991957104557644e-05,
+      "loss": 0.2286,
+      "step": 722
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.097855227882037e-05,
+      "loss": 0.1189,
+      "step": 723
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.0965147453083115e-05,
+      "loss": 0.2882,
+      "step": 724
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.095174262734584e-05,
+      "loss": 0.2623,
+      "step": 725
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.093833780160858e-05,
+      "loss": 0.2473,
+      "step": 726
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.092493297587131e-05,
+      "loss": 0.4846,
+      "step": 727
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.091152815013405e-05,
+      "loss": 0.1689,
+      "step": 728
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0898123324396783e-05,
+      "loss": 0.3481,
+      "step": 729
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.088471849865952e-05,
+      "loss": 0.3447,
+      "step": 730
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0871313672922254e-05,
+      "loss": 0.2959,
+      "step": 731
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.085790884718499e-05,
+      "loss": 0.3387,
+      "step": 732
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0844504021447724e-05,
+      "loss": 0.3742,
+      "step": 733
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.083109919571046e-05,
+      "loss": 0.3245,
+      "step": 734
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0817694369973194e-05,
+      "loss": 0.4891,
+      "step": 735
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.080428954423593e-05,
+      "loss": 0.1444,
+      "step": 736
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.0790884718498664e-05,
+      "loss": 0.3678,
+      "step": 737
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.077747989276139e-05,
+      "loss": 0.3772,
+      "step": 738
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0764075067024135e-05,
+      "loss": 0.43,
+      "step": 739
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.075067024128686e-05,
+      "loss": 0.2463,
+      "step": 740
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0737265415549605e-05,
+      "loss": 0.2277,
+      "step": 741
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.072386058981233e-05,
+      "loss": 0.2153,
+      "step": 742
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.071045576407507e-05,
+      "loss": 0.1052,
+      "step": 743
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0697050938337804e-05,
+      "loss": 0.5657,
+      "step": 744
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.068364611260054e-05,
+      "loss": 0.2664,
+      "step": 745
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0670241286863274e-05,
+      "loss": 0.1369,
+      "step": 746
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.065683646112601e-05,
+      "loss": 0.4972,
+      "step": 747
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.064343163538874e-05,
+      "loss": 0.4131,
+      "step": 748
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.063002680965148e-05,
+      "loss": 0.3824,
+      "step": 749
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.061662198391421e-05,
+      "loss": 0.323,
+      "step": 750
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.060321715817695e-05,
+      "loss": 0.3698,
+      "step": 751
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.058981233243968e-05,
+      "loss": 0.1708,
+      "step": 752
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.057640750670241e-05,
+      "loss": 0.2941,
+      "step": 753
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.056300268096515e-05,
+      "loss": 0.3224,
+      "step": 754
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.054959785522788e-05,
+      "loss": 0.0851,
+      "step": 755
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.053619302949062e-05,
+      "loss": 0.4694,
+      "step": 756
+    },
+    {
+      "epoch": 2.0,
+      "eval_f1": 0.7882736156351792,
+      "eval_loss": 0.4484867751598358,
+      "eval_runtime": 1.8734,
+      "eval_samples_per_second": 807.638,
+      "eval_steps_per_second": 50.711,
+      "step": 756
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0522788203753354e-05,
+      "loss": 0.4739,
+      "step": 757
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.050938337801609e-05,
+      "loss": 0.4117,
+      "step": 758
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0495978552278824e-05,
+      "loss": 0.2169,
+      "step": 759
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.048257372654156e-05,
+      "loss": 0.1848,
+      "step": 760
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.0469168900804294e-05,
+      "loss": 0.5066,
+      "step": 761
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.045576407506703e-05,
+      "loss": 0.1784,
+      "step": 762
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.044235924932976e-05,
+      "loss": 0.3869,
+      "step": 763
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.04289544235925e-05,
+      "loss": 0.1132,
+      "step": 764
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.041554959785523e-05,
+      "loss": 0.2724,
+      "step": 765
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.040214477211797e-05,
+      "loss": 0.0983,
+      "step": 766
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.03887399463807e-05,
+      "loss": 0.1831,
+      "step": 767
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.037533512064343e-05,
+      "loss": 0.1954,
+      "step": 768
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.036193029490617e-05,
+      "loss": 0.4738,
+      "step": 769
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.0348525469168903e-05,
+      "loss": 0.3375,
+      "step": 770
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.033512064343164e-05,
+      "loss": 0.3991,
+      "step": 771
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.0321715817694374e-05,
+      "loss": 0.0696,
+      "step": 772
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.03083109919571e-05,
+      "loss": 0.0982,
+      "step": 773
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.0294906166219844e-05,
+      "loss": 0.2236,
+      "step": 774
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.028150134048257e-05,
+      "loss": 0.4225,
+      "step": 775
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.0268096514745314e-05,
+      "loss": 0.0583,
+      "step": 776
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.025469168900804e-05,
+      "loss": 0.1643,
+      "step": 777
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.024128686327078e-05,
+      "loss": 0.0351,
+      "step": 778
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.022788203753351e-05,
+      "loss": 0.4496,
+      "step": 779
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.021447721179625e-05,
+      "loss": 0.0372,
+      "step": 780
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.020107238605898e-05,
+      "loss": 0.4198,
+      "step": 781
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.018766756032172e-05,
+      "loss": 0.1968,
+      "step": 782
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.017426273458445e-05,
+      "loss": 0.515,
+      "step": 783
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.016085790884719e-05,
+      "loss": 0.408,
+      "step": 784
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.0147453083109924e-05,
+      "loss": 0.3693,
+      "step": 785
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.013404825737266e-05,
+      "loss": 0.0561,
+      "step": 786
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.0120643431635394e-05,
+      "loss": 0.4011,
+      "step": 787
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.010723860589812e-05,
+      "loss": 0.0872,
+      "step": 788
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.0093833780160864e-05,
+      "loss": 0.0768,
+      "step": 789
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.008042895442359e-05,
+      "loss": 0.0184,
+      "step": 790
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.0067024128686334e-05,
+      "loss": 0.3287,
+      "step": 791
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.005361930294906e-05,
+      "loss": 0.0262,
+      "step": 792
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.00402144772118e-05,
+      "loss": 0.0248,
+      "step": 793
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.002680965147453e-05,
+      "loss": 0.3853,
+      "step": 794
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.001340482573727e-05,
+      "loss": 0.3512,
+      "step": 795
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4e-05,
+      "loss": 0.4188,
+      "step": 796
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.998659517426274e-05,
+      "loss": 0.1834,
+      "step": 797
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.997319034852547e-05,
+      "loss": 0.2074,
+      "step": 798
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.995978552278821e-05,
+      "loss": 0.7317,
+      "step": 799
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.994638069705094e-05,
+      "loss": 0.3534,
+      "step": 800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.993297587131368e-05,
+      "loss": 0.3184,
+      "step": 801
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.991957104557641e-05,
+      "loss": 0.1088,
+      "step": 802
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.990616621983914e-05,
+      "loss": 0.0429,
+      "step": 803
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.989276139410188e-05,
+      "loss": 0.4518,
+      "step": 804
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.987935656836461e-05,
+      "loss": 0.1746,
+      "step": 805
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.986595174262735e-05,
+      "loss": 0.1881,
+      "step": 806
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.985254691689008e-05,
+      "loss": 0.4111,
+      "step": 807
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.983914209115281e-05,
+      "loss": 0.059,
+      "step": 808
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.982573726541555e-05,
+      "loss": 0.0495,
+      "step": 809
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.981233243967828e-05,
+      "loss": 0.1134,
+      "step": 810
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9798927613941023e-05,
+      "loss": 0.5469,
+      "step": 811
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.978552278820375e-05,
+      "loss": 0.0581,
+      "step": 812
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.977211796246649e-05,
+      "loss": 0.1254,
+      "step": 813
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.975871313672922e-05,
+      "loss": 0.0679,
+      "step": 814
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.974530831099196e-05,
+      "loss": 0.0463,
+      "step": 815
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.973190348525469e-05,
+      "loss": 0.33,
+      "step": 816
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.971849865951743e-05,
+      "loss": 0.2931,
+      "step": 817
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.970509383378016e-05,
+      "loss": 0.1034,
+      "step": 818
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.96916890080429e-05,
+      "loss": 0.0379,
+      "step": 819
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.967828418230563e-05,
+      "loss": 0.0456,
+      "step": 820
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.966487935656837e-05,
+      "loss": 0.4862,
+      "step": 821
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.96514745308311e-05,
+      "loss": 0.0512,
+      "step": 822
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.963806970509383e-05,
+      "loss": 0.0879,
+      "step": 823
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.962466487935657e-05,
+      "loss": 0.3664,
+      "step": 824
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.96112600536193e-05,
+      "loss": 0.0975,
+      "step": 825
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9597855227882044e-05,
+      "loss": 0.162,
+      "step": 826
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.958445040214477e-05,
+      "loss": 0.1076,
+      "step": 827
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.957104557640751e-05,
+      "loss": 0.1185,
+      "step": 828
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.955764075067024e-05,
+      "loss": 0.0868,
+      "step": 829
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.954423592493298e-05,
+      "loss": 0.0246,
+      "step": 830
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.953083109919571e-05,
+      "loss": 0.2233,
+      "step": 831
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.951742627345845e-05,
+      "loss": 0.2113,
+      "step": 832
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.9504021447721176e-05,
+      "loss": 0.0587,
+      "step": 833
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.949061662198392e-05,
+      "loss": 0.0521,
+      "step": 834
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9477211796246646e-05,
+      "loss": 0.3845,
+      "step": 835
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.946380697050939e-05,
+      "loss": 0.1096,
+      "step": 836
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9450402144772117e-05,
+      "loss": 0.1488,
+      "step": 837
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.943699731903485e-05,
+      "loss": 0.1937,
+      "step": 838
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.942359249329759e-05,
+      "loss": 0.1309,
+      "step": 839
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.941018766756032e-05,
+      "loss": 0.3271,
+      "step": 840
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.939678284182306e-05,
+      "loss": 0.3318,
+      "step": 841
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.938337801608579e-05,
+      "loss": 0.3516,
+      "step": 842
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.936997319034853e-05,
+      "loss": 0.1641,
+      "step": 843
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.935656836461126e-05,
+      "loss": 0.064,
+      "step": 844
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.9343163538874e-05,
+      "loss": 0.1971,
+      "step": 845
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.932975871313673e-05,
+      "loss": 0.1166,
+      "step": 846
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.931635388739947e-05,
+      "loss": 0.0384,
+      "step": 847
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.9302949061662196e-05,
+      "loss": 0.0462,
+      "step": 848
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.928954423592494e-05,
+      "loss": 0.1073,
+      "step": 849
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.9276139410187666e-05,
+      "loss": 0.0227,
+      "step": 850
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.926273458445041e-05,
+      "loss": 0.0683,
+      "step": 851
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.924932975871314e-05,
+      "loss": 0.7962,
+      "step": 852
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.923592493297587e-05,
+      "loss": 0.0635,
+      "step": 853
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.922252010723861e-05,
+      "loss": 0.0454,
+      "step": 854
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.920911528150134e-05,
+      "loss": 0.5389,
+      "step": 855
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.919571045576408e-05,
+      "loss": 0.1181,
+      "step": 856
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.918230563002681e-05,
+      "loss": 0.1372,
+      "step": 857
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.916890080428954e-05,
+      "loss": 0.3192,
+      "step": 858
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.915549597855228e-05,
+      "loss": 0.3419,
+      "step": 859
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.914209115281501e-05,
+      "loss": 0.0114,
+      "step": 860
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.912868632707775e-05,
+      "loss": 0.1905,
+      "step": 861
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.911528150134048e-05,
+      "loss": 0.0218,
+      "step": 862
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.910187667560322e-05,
+      "loss": 0.4374,
+      "step": 863
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.908847184986595e-05,
+      "loss": 0.1844,
+      "step": 864
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.907506702412869e-05,
+      "loss": 0.2427,
+      "step": 865
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.906166219839142e-05,
+      "loss": 0.2749,
+      "step": 866
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.904825737265416e-05,
+      "loss": 0.2089,
+      "step": 867
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.903485254691689e-05,
+      "loss": 0.06,
+      "step": 868
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.902144772117963e-05,
+      "loss": 0.1951,
+      "step": 869
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.900804289544236e-05,
+      "loss": 0.0252,
+      "step": 870
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.89946380697051e-05,
+      "loss": 0.0299,
+      "step": 871
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.898123324396783e-05,
+      "loss": 0.0298,
+      "step": 872
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.896782841823057e-05,
+      "loss": 0.5186,
+      "step": 873
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.89544235924933e-05,
+      "loss": 0.2704,
+      "step": 874
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.894101876675603e-05,
+      "loss": 0.2435,
+      "step": 875
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.892761394101877e-05,
+      "loss": 0.039,
+      "step": 876
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.89142091152815e-05,
+      "loss": 0.0275,
+      "step": 877
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.890080428954424e-05,
+      "loss": 0.1164,
+      "step": 878
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.888739946380697e-05,
+      "loss": 0.1551,
+      "step": 879
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.887399463806971e-05,
+      "loss": 0.0215,
+      "step": 880
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.886058981233244e-05,
+      "loss": 0.0379,
+      "step": 881
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.884718498659518e-05,
+      "loss": 0.0553,
+      "step": 882
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.883378016085791e-05,
+      "loss": 0.1073,
+      "step": 883
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.882037533512065e-05,
+      "loss": 0.3525,
+      "step": 884
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8806970509383376e-05,
+      "loss": 0.2646,
+      "step": 885
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.879356568364612e-05,
+      "loss": 0.5758,
+      "step": 886
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8780160857908846e-05,
+      "loss": 0.9312,
+      "step": 887
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.876675603217159e-05,
+      "loss": 0.2748,
+      "step": 888
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.8753351206434316e-05,
+      "loss": 0.33,
+      "step": 889
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.873994638069705e-05,
+      "loss": 0.0312,
+      "step": 890
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.8726541554959786e-05,
+      "loss": 0.0449,
+      "step": 891
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.871313672922252e-05,
+      "loss": 0.1197,
+      "step": 892
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.869973190348526e-05,
+      "loss": 0.0913,
+      "step": 893
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.868632707774799e-05,
+      "loss": 0.0284,
+      "step": 894
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.867292225201073e-05,
+      "loss": 0.3769,
+      "step": 895
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.865951742627346e-05,
+      "loss": 0.0947,
+      "step": 896
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.86461126005362e-05,
+      "loss": 0.4282,
+      "step": 897
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.863270777479893e-05,
+      "loss": 0.0049,
+      "step": 898
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.861930294906167e-05,
+      "loss": 0.3632,
+      "step": 899
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.8605898123324396e-05,
+      "loss": 0.0421,
+      "step": 900
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.859249329758714e-05,
+      "loss": 0.5793,
+      "step": 901
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8579088471849866e-05,
+      "loss": 0.1695,
+      "step": 902
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.856568364611261e-05,
+      "loss": 0.3082,
+      "step": 903
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.8552278820375336e-05,
+      "loss": 0.0151,
+      "step": 904
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.853887399463807e-05,
+      "loss": 0.3463,
+      "step": 905
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.852546916890081e-05,
+      "loss": 0.4573,
+      "step": 906
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.851206434316354e-05,
+      "loss": 0.1281,
+      "step": 907
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.849865951742628e-05,
+      "loss": 0.3168,
+      "step": 908
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.848525469168901e-05,
+      "loss": 0.0331,
+      "step": 909
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.847184986595174e-05,
+      "loss": 0.1825,
+      "step": 910
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.845844504021448e-05,
+      "loss": 0.6238,
+      "step": 911
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.844504021447721e-05,
+      "loss": 0.0663,
+      "step": 912
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.843163538873995e-05,
+      "loss": 0.2827,
+      "step": 913
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.841823056300268e-05,
+      "loss": 0.0133,
+      "step": 914
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.8404825737265416e-05,
+      "loss": 0.0415,
+      "step": 915
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.839142091152815e-05,
+      "loss": 0.2969,
+      "step": 916
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8378016085790886e-05,
+      "loss": 0.0145,
+      "step": 917
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.836461126005362e-05,
+      "loss": 0.0223,
+      "step": 918
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8351206434316357e-05,
+      "loss": 1.042,
+      "step": 919
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8337801608579085e-05,
+      "loss": 0.5061,
+      "step": 920
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.832439678284183e-05,
+      "loss": 0.406,
+      "step": 921
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8310991957104555e-05,
+      "loss": 0.1337,
+      "step": 922
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.82975871313673e-05,
+      "loss": 0.2057,
+      "step": 923
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8284182305630025e-05,
+      "loss": 0.0987,
+      "step": 924
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.827077747989276e-05,
+      "loss": 0.4611,
+      "step": 925
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8257372654155496e-05,
+      "loss": 0.0152,
+      "step": 926
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.824396782841823e-05,
+      "loss": 0.4478,
+      "step": 927
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8230563002680966e-05,
+      "loss": 0.0579,
+      "step": 928
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.82171581769437e-05,
+      "loss": 0.1506,
+      "step": 929
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.8203753351206436e-05,
+      "loss": 0.5514,
+      "step": 930
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.819034852546917e-05,
+      "loss": 0.1505,
+      "step": 931
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8176943699731906e-05,
+      "loss": 0.1413,
+      "step": 932
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.816353887399464e-05,
+      "loss": 0.4758,
+      "step": 933
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.815013404825738e-05,
+      "loss": 0.0863,
+      "step": 934
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.8136729222520105e-05,
+      "loss": 0.3561,
+      "step": 935
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.812332439678285e-05,
+      "loss": 0.7784,
+      "step": 936
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.8109919571045575e-05,
+      "loss": 0.2243,
+      "step": 937
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.809651474530832e-05,
+      "loss": 0.1013,
+      "step": 938
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.8083109919571046e-05,
+      "loss": 0.132,
+      "step": 939
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.806970509383378e-05,
+      "loss": 0.3971,
+      "step": 940
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.8056300268096516e-05,
+      "loss": 0.0637,
+      "step": 941
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.804289544235925e-05,
+      "loss": 0.5178,
+      "step": 942
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.8029490616621986e-05,
+      "loss": 0.1982,
+      "step": 943
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.801608579088472e-05,
+      "loss": 0.2225,
+      "step": 944
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.800268096514745e-05,
+      "loss": 0.1425,
+      "step": 945
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.798927613941019e-05,
+      "loss": 0.0621,
+      "step": 946
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.797587131367292e-05,
+      "loss": 0.1556,
+      "step": 947
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.796246648793566e-05,
+      "loss": 0.3805,
+      "step": 948
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.794906166219839e-05,
+      "loss": 0.7049,
+      "step": 949
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7935656836461125e-05,
+      "loss": 0.1055,
+      "step": 950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.792225201072386e-05,
+      "loss": 0.0489,
+      "step": 951
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7908847184986596e-05,
+      "loss": 0.1881,
+      "step": 952
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.789544235924933e-05,
+      "loss": 0.0202,
+      "step": 953
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.7882037533512066e-05,
+      "loss": 0.1043,
+      "step": 954
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.78686327077748e-05,
+      "loss": 0.2093,
+      "step": 955
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7855227882037536e-05,
+      "loss": 0.0395,
+      "step": 956
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.784182305630027e-05,
+      "loss": 0.1459,
+      "step": 957
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7828418230563006e-05,
+      "loss": 0.0338,
+      "step": 958
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.781501340482574e-05,
+      "loss": 0.4741,
+      "step": 959
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.780160857908847e-05,
+      "loss": 0.2049,
+      "step": 960
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.778820375335121e-05,
+      "loss": 0.309,
+      "step": 961
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.777479892761394e-05,
+      "loss": 0.0253,
+      "step": 962
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.776139410187668e-05,
+      "loss": 0.4832,
+      "step": 963
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.774798927613941e-05,
+      "loss": 0.2111,
+      "step": 964
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.7734584450402145e-05,
+      "loss": 0.1788,
+      "step": 965
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.772117962466488e-05,
+      "loss": 0.5252,
+      "step": 966
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7707774798927616e-05,
+      "loss": 0.4711,
+      "step": 967
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.769436997319035e-05,
+      "loss": 0.5184,
+      "step": 968
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7680965147453086e-05,
+      "loss": 0.2164,
+      "step": 969
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7667560321715814e-05,
+      "loss": 0.5393,
+      "step": 970
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7654155495978556e-05,
+      "loss": 0.2588,
+      "step": 971
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7640750670241285e-05,
+      "loss": 0.164,
+      "step": 972
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.7627345844504027e-05,
+      "loss": 0.2896,
+      "step": 973
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7613941018766755e-05,
+      "loss": 0.039,
+      "step": 974
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.760053619302949e-05,
+      "loss": 0.16,
+      "step": 975
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.7587131367292225e-05,
+      "loss": 0.1832,
+      "step": 976
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.757372654155496e-05,
+      "loss": 0.0812,
+      "step": 977
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7560321715817695e-05,
+      "loss": 0.1476,
+      "step": 978
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.754691689008043e-05,
+      "loss": 0.1853,
+      "step": 979
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7533512064343166e-05,
+      "loss": 0.2875,
+      "step": 980
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.75201072386059e-05,
+      "loss": 0.1918,
+      "step": 981
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7506702412868636e-05,
+      "loss": 0.2445,
+      "step": 982
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.749329758713137e-05,
+      "loss": 0.4653,
+      "step": 983
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.7479892761394106e-05,
+      "loss": 0.0614,
+      "step": 984
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.746648793565684e-05,
+      "loss": 0.2818,
+      "step": 985
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7453083109919576e-05,
+      "loss": 0.1363,
+      "step": 986
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.7439678284182305e-05,
+      "loss": 0.3244,
+      "step": 987
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.742627345844505e-05,
+      "loss": 0.081,
+      "step": 988
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7412868632707775e-05,
+      "loss": 0.0488,
+      "step": 989
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.739946380697052e-05,
+      "loss": 0.2057,
+      "step": 990
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7386058981233245e-05,
+      "loss": 0.2598,
+      "step": 991
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.737265415549598e-05,
+      "loss": 0.1318,
+      "step": 992
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7359249329758716e-05,
+      "loss": 0.0482,
+      "step": 993
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.734584450402145e-05,
+      "loss": 0.2586,
+      "step": 994
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7332439678284186e-05,
+      "loss": 0.2533,
+      "step": 995
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.731903485254692e-05,
+      "loss": 0.5292,
+      "step": 996
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.730563002680965e-05,
+      "loss": 0.1677,
+      "step": 997
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.729222520107239e-05,
+      "loss": 0.1869,
+      "step": 998
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.727882037533512e-05,
+      "loss": 0.0645,
+      "step": 999
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.726541554959786e-05,
+      "loss": 0.4107,
+      "step": 1000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.725201072386059e-05,
+      "loss": 0.0484,
+      "step": 1001
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.7238605898123325e-05,
+      "loss": 0.0813,
+      "step": 1002
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.722520107238606e-05,
+      "loss": 0.2467,
+      "step": 1003
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7211796246648795e-05,
+      "loss": 0.0324,
+      "step": 1004
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.719839142091153e-05,
+      "loss": 0.0536,
+      "step": 1005
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7184986595174266e-05,
+      "loss": 0.0399,
+      "step": 1006
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7171581769436994e-05,
+      "loss": 0.0257,
+      "step": 1007
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7158176943699736e-05,
+      "loss": 0.0407,
+      "step": 1008
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7144772117962464e-05,
+      "loss": 0.0173,
+      "step": 1009
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7131367292225206e-05,
+      "loss": 0.0166,
+      "step": 1010
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7117962466487934e-05,
+      "loss": 0.1898,
+      "step": 1011
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.710455764075067e-05,
+      "loss": 0.0525,
+      "step": 1012
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.7091152815013405e-05,
+      "loss": 0.043,
+      "step": 1013
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.707774798927614e-05,
+      "loss": 0.3994,
+      "step": 1014
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7064343163538875e-05,
+      "loss": 0.0372,
+      "step": 1015
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.705093833780161e-05,
+      "loss": 0.2909,
+      "step": 1016
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.7037533512064345e-05,
+      "loss": 0.8221,
+      "step": 1017
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.702412868632708e-05,
+      "loss": 0.0084,
+      "step": 1018
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.7010723860589815e-05,
+      "loss": 0.3058,
+      "step": 1019
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.699731903485255e-05,
+      "loss": 1.0774,
+      "step": 1020
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6983914209115286e-05,
+      "loss": 0.2018,
+      "step": 1021
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6970509383378014e-05,
+      "loss": 0.0537,
+      "step": 1022
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6957104557640756e-05,
+      "loss": 0.444,
+      "step": 1023
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6943699731903484e-05,
+      "loss": 0.6497,
+      "step": 1024
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6930294906166226e-05,
+      "loss": 0.0097,
+      "step": 1025
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.6916890080428955e-05,
+      "loss": 0.0082,
+      "step": 1026
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.690348525469169e-05,
+      "loss": 0.0387,
+      "step": 1027
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6890080428954425e-05,
+      "loss": 0.3969,
+      "step": 1028
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.687667560321716e-05,
+      "loss": 0.0136,
+      "step": 1029
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6863270777479895e-05,
+      "loss": 0.0099,
+      "step": 1030
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.684986595174263e-05,
+      "loss": 0.3509,
+      "step": 1031
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.683646112600536e-05,
+      "loss": 0.0257,
+      "step": 1032
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.68230563002681e-05,
+      "loss": 0.3158,
+      "step": 1033
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.680965147453083e-05,
+      "loss": 0.0677,
+      "step": 1034
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.679624664879357e-05,
+      "loss": 0.3887,
+      "step": 1035
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.67828418230563e-05,
+      "loss": 0.0714,
+      "step": 1036
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6769436997319034e-05,
+      "loss": 0.1066,
+      "step": 1037
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.675603217158177e-05,
+      "loss": 0.6238,
+      "step": 1038
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6742627345844504e-05,
+      "loss": 0.0405,
+      "step": 1039
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.672922252010724e-05,
+      "loss": 0.0223,
+      "step": 1040
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6715817694369975e-05,
+      "loss": 0.2737,
+      "step": 1041
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.670241286863271e-05,
+      "loss": 0.015,
+      "step": 1042
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6689008042895445e-05,
+      "loss": 0.1709,
+      "step": 1043
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.667560321715818e-05,
+      "loss": 0.2649,
+      "step": 1044
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6662198391420915e-05,
+      "loss": 0.1524,
+      "step": 1045
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.664879356568365e-05,
+      "loss": 0.2461,
+      "step": 1046
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.663538873994638e-05,
+      "loss": 0.3425,
+      "step": 1047
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.662198391420912e-05,
+      "loss": 0.2689,
+      "step": 1048
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.660857908847185e-05,
+      "loss": 0.0066,
+      "step": 1049
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.659517426273459e-05,
+      "loss": 0.0328,
+      "step": 1050
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.658176943699732e-05,
+      "loss": 0.1273,
+      "step": 1051
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6568364611260054e-05,
+      "loss": 0.2346,
+      "step": 1052
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.655495978552279e-05,
+      "loss": 0.0118,
+      "step": 1053
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6541554959785525e-05,
+      "loss": 0.0287,
+      "step": 1054
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.652815013404826e-05,
+      "loss": 0.264,
+      "step": 1055
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.6514745308310995e-05,
+      "loss": 0.0216,
+      "step": 1056
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.650134048257372e-05,
+      "loss": 0.0261,
+      "step": 1057
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6487935656836465e-05,
+      "loss": 0.1911,
+      "step": 1058
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6474530831099194e-05,
+      "loss": 0.029,
+      "step": 1059
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.6461126005361935e-05,
+      "loss": 0.0393,
+      "step": 1060
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6447721179624664e-05,
+      "loss": 0.1044,
+      "step": 1061
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.64343163538874e-05,
+      "loss": 0.6364,
+      "step": 1062
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.6420911528150134e-05,
+      "loss": 0.3589,
+      "step": 1063
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.640750670241287e-05,
+      "loss": 0.1128,
+      "step": 1064
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6394101876675604e-05,
+      "loss": 0.2219,
+      "step": 1065
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.638069705093834e-05,
+      "loss": 0.211,
+      "step": 1066
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.6367292225201075e-05,
+      "loss": 0.387,
+      "step": 1067
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.635388739946381e-05,
+      "loss": 0.0725,
+      "step": 1068
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6340482573726545e-05,
+      "loss": 0.0268,
+      "step": 1069
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.632707774798928e-05,
+      "loss": 0.516,
+      "step": 1070
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.6313672922252015e-05,
+      "loss": 0.0746,
+      "step": 1071
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6300268096514743e-05,
+      "loss": 0.2486,
+      "step": 1072
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6286863270777485e-05,
+      "loss": 0.1584,
+      "step": 1073
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6273458445040214e-05,
+      "loss": 0.1301,
+      "step": 1074
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6260053619302956e-05,
+      "loss": 0.0122,
+      "step": 1075
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6246648793565684e-05,
+      "loss": 0.0215,
+      "step": 1076
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.623324396782842e-05,
+      "loss": 0.2068,
+      "step": 1077
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6219839142091154e-05,
+      "loss": 0.1882,
+      "step": 1078
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.620643431635389e-05,
+      "loss": 0.368,
+      "step": 1079
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.6193029490616625e-05,
+      "loss": 0.7537,
+      "step": 1080
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.617962466487936e-05,
+      "loss": 0.1614,
+      "step": 1081
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.616621983914209e-05,
+      "loss": 0.0585,
+      "step": 1082
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.615281501340483e-05,
+      "loss": 0.0535,
+      "step": 1083
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.613941018766756e-05,
+      "loss": 0.6518,
+      "step": 1084
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.61260053619303e-05,
+      "loss": 0.4885,
+      "step": 1085
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.611260053619303e-05,
+      "loss": 0.4078,
+      "step": 1086
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.6099195710455764e-05,
+      "loss": 0.2101,
+      "step": 1087
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.60857908847185e-05,
+      "loss": 0.0192,
+      "step": 1088
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.6072386058981234e-05,
+      "loss": 0.3885,
+      "step": 1089
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.605898123324397e-05,
+      "loss": 0.0393,
+      "step": 1090
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.6045576407506704e-05,
+      "loss": 0.2179,
+      "step": 1091
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.603217158176944e-05,
+      "loss": 0.1814,
+      "step": 1092
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.6018766756032174e-05,
+      "loss": 0.0647,
+      "step": 1093
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.600536193029491e-05,
+      "loss": 0.0657,
+      "step": 1094
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.5991957104557645e-05,
+      "loss": 0.1062,
+      "step": 1095
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.597855227882038e-05,
+      "loss": 0.4314,
+      "step": 1096
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.596514745308311e-05,
+      "loss": 0.3074,
+      "step": 1097
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.595174262734585e-05,
+      "loss": 0.0159,
+      "step": 1098
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.593833780160858e-05,
+      "loss": 0.3829,
+      "step": 1099
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.592493297587132e-05,
+      "loss": 0.3277,
+      "step": 1100
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.591152815013405e-05,
+      "loss": 0.3785,
+      "step": 1101
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5898123324396784e-05,
+      "loss": 0.0162,
+      "step": 1102
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.588471849865952e-05,
+      "loss": 0.0772,
+      "step": 1103
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5871313672922254e-05,
+      "loss": 0.0292,
+      "step": 1104
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.585790884718499e-05,
+      "loss": 0.0748,
+      "step": 1105
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5844504021447724e-05,
+      "loss": 0.2276,
+      "step": 1106
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.583109919571046e-05,
+      "loss": 0.3174,
+      "step": 1107
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5817694369973195e-05,
+      "loss": 0.134,
+      "step": 1108
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.580428954423592e-05,
+      "loss": 0.3488,
+      "step": 1109
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5790884718498665e-05,
+      "loss": 0.1003,
+      "step": 1110
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.577747989276139e-05,
+      "loss": 0.1972,
+      "step": 1111
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5764075067024135e-05,
+      "loss": 0.2151,
+      "step": 1112
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.5750670241286863e-05,
+      "loss": 0.3725,
+      "step": 1113
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.57372654155496e-05,
+      "loss": 0.0227,
+      "step": 1114
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5723860589812334e-05,
+      "loss": 0.2063,
+      "step": 1115
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.571045576407507e-05,
+      "loss": 0.0316,
+      "step": 1116
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.5697050938337804e-05,
+      "loss": 0.1545,
+      "step": 1117
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.568364611260054e-05,
+      "loss": 0.4582,
+      "step": 1118
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.567024128686327e-05,
+      "loss": 0.5452,
+      "step": 1119
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.565683646112601e-05,
+      "loss": 0.2009,
+      "step": 1120
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.564343163538874e-05,
+      "loss": 0.1201,
+      "step": 1121
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.563002680965148e-05,
+      "loss": 0.5343,
+      "step": 1122
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.561662198391421e-05,
+      "loss": 0.3925,
+      "step": 1123
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.560321715817694e-05,
+      "loss": 0.0375,
+      "step": 1124
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.558981233243968e-05,
+      "loss": 0.0411,
+      "step": 1125
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5576407506702413e-05,
+      "loss": 0.0338,
+      "step": 1126
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.556300268096515e-05,
+      "loss": 0.035,
+      "step": 1127
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5549597855227884e-05,
+      "loss": 0.0283,
+      "step": 1128
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.553619302949062e-05,
+      "loss": 0.034,
+      "step": 1129
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5522788203753354e-05,
+      "loss": 0.518,
+      "step": 1130
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.550938337801609e-05,
+      "loss": 0.0241,
+      "step": 1131
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5495978552278824e-05,
+      "loss": 0.3143,
+      "step": 1132
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.548257372654156e-05,
+      "loss": 0.8011,
+      "step": 1133
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.546916890080429e-05,
+      "loss": 0.4365,
+      "step": 1134
+    },
+    {
+      "epoch": 3.0,
+      "eval_f1": 0.7867219917012448,
+      "eval_loss": 0.613310694694519,
+      "eval_runtime": 1.9007,
+      "eval_samples_per_second": 796.041,
+      "eval_steps_per_second": 49.983,
+      "step": 1134
+    }
+  ],
+  "max_steps": 3780,
+  "num_train_epochs": 10,
+  "total_flos": 291221296732800.0,
+  "trial_name": null,
+  "trial_params": null
+}