diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,72021 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.942181340341656,
+  "eval_steps": 30000.0,
+  "global_step": 12000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.37636761487965e-08,
+      "loss": 51.6875,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.7527352297593e-08,
+      "loss": 54.375,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.312910284463895e-07,
+      "loss": 50.5625,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.75054704595186e-07,
+      "loss": 53.3125,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.188183807439825e-07,
+      "loss": 49.0625,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.62582056892779e-07,
+      "loss": 46.0,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.063457330415755e-07,
+      "loss": 38.75,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.50109409190372e-07,
+      "loss": 57.4375,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9387308533916855e-07,
+      "loss": 35.625,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.37636761487965e-07,
+      "loss": 38.1875,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.814004376367615e-07,
+      "loss": 46.25,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.25164113785558e-07,
+      "loss": 52.8125,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.689277899343545e-07,
+      "loss": 47.125,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.12691466083151e-07,
+      "loss": 57.3125,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.564551422319475e-07,
+      "loss": 54.75,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.00218818380744e-07,
+      "loss": 55.0,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.439824945295406e-07,
+      "loss": 47.375,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.877461706783371e-07,
+      "loss": 51.3125,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.315098468271335e-07,
+      "loss": 54.5,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.7527352297593e-07,
+      "loss": 46.25,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.190371991247265e-07,
+      "loss": 46.75,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.62800875273523e-07,
+      "loss": 48.625,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0065645514223197e-06,
+      "loss": 49.75,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.050328227571116e-06,
+      "loss": 55.875,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0940919037199126e-06,
+      "loss": 58.0625,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.137855579868709e-06,
+      "loss": 56.125,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1816192560175056e-06,
+      "loss": 47.625,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.225382932166302e-06,
+      "loss": 42.75,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2691466083150985e-06,
+      "loss": 44.8125,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.312910284463895e-06,
+      "loss": 46.5,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3566739606126916e-06,
+      "loss": 46.4688,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.400437636761488e-06,
+      "loss": 48.6562,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4442013129102847e-06,
+      "loss": 48.4375,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4879649890590811e-06,
+      "loss": 53.125,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5317286652078775e-06,
+      "loss": 49.125,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5754923413566742e-06,
+      "loss": 44.0,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6192560175054706e-06,
+      "loss": 44.75,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.663019693654267e-06,
+      "loss": 45.4375,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7067833698030637e-06,
+      "loss": 43.6875,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.75054704595186e-06,
+      "loss": 47.125,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7943107221006568e-06,
+      "loss": 49.8125,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.838074398249453e-06,
+      "loss": 54.5625,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8818380743982495e-06,
+      "loss": 46.4375,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.925601750547046e-06,
+      "loss": 53.5,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9693654266958425e-06,
+      "loss": 47.2812,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0131291028446394e-06,
+      "loss": 44.5625,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.056892778993436e-06,
+      "loss": 39.1875,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.100656455142232e-06,
+      "loss": 51.4375,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1444201312910287e-06,
+      "loss": 53.5625,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.188183807439825e-06,
+      "loss": 47.8125,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2319474835886216e-06,
+      "loss": 51.5,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.275711159737418e-06,
+      "loss": 47.3125,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3194748358862144e-06,
+      "loss": 51.5,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3632385120350113e-06,
+      "loss": 42.875,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4070021881838077e-06,
+      "loss": 48.5625,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.450765864332604e-06,
+      "loss": 47.6875,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4945295404814006e-06,
+      "loss": 54.5,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.538293216630197e-06,
+      "loss": 47.25,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5820568927789935e-06,
+      "loss": 46.9375,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.62582056892779e-06,
+      "loss": 52.3125,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6695842450765868e-06,
+      "loss": 50.875,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.713347921225383e-06,
+      "loss": 45.25,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7571115973741796e-06,
+      "loss": 48.0,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.800875273522976e-06,
+      "loss": 51.0,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8446389496717725e-06,
+      "loss": 50.125,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8884026258205694e-06,
+      "loss": 46.375,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.932166301969366e-06,
+      "loss": 47.0,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9759299781181622e-06,
+      "loss": 46.625,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0196936542669587e-06,
+      "loss": 50.5,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.063457330415755e-06,
+      "loss": 47.5625,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.107221006564552e-06,
+      "loss": 44.4375,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1509846827133484e-06,
+      "loss": 53.875,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.194748358862145e-06,
+      "loss": 45.375,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2385120350109413e-06,
+      "loss": 42.8125,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2822757111597377e-06,
+      "loss": 39.5,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.326039387308534e-06,
+      "loss": 44.0,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.369803063457331e-06,
+      "loss": 46.5625,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4135667396061274e-06,
+      "loss": 45.5,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.457330415754924e-06,
+      "loss": 46.6875,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.50109409190372e-06,
+      "loss": 44.5625,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5448577680525163e-06,
+      "loss": 38.875,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5886214442013136e-06,
+      "loss": 44.875,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.63238512035011e-06,
+      "loss": 45.5,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.676148796498906e-06,
+      "loss": 46.5625,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7199124726477025e-06,
+      "loss": 44.125,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.763676148796499e-06,
+      "loss": 52.25,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8074398249452958e-06,
+      "loss": 55.5625,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.851203501094092e-06,
+      "loss": 44.75,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.894967177242889e-06,
+      "loss": 41.3125,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.938730853391685e-06,
+      "loss": 36.9688,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.982494529540482e-06,
+      "loss": 50.6875,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.026258205689279e-06,
+      "loss": 43.875,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.070021881838075e-06,
+      "loss": 42.875,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.113785557986872e-06,
+      "loss": 46.5625,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.157549234135668e-06,
+      "loss": 42.0625,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.201312910284464e-06,
+      "loss": 44.0625,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2450765864332605e-06,
+      "loss": 39.3125,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.288840262582057e-06,
+      "loss": 45.9688,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.332603938730854e-06,
+      "loss": 42.75,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.37636761487965e-06,
+      "loss": 41.125,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.420131291028446e-06,
+      "loss": 42.4375,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.463894967177243e-06,
+      "loss": 39.9375,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.50765864332604e-06,
+      "loss": 30.3438,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.551422319474836e-06,
+      "loss": 37.0625,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.595185995623633e-06,
+      "loss": 45.0,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.638949671772429e-06,
+      "loss": 41.0,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.682713347921226e-06,
+      "loss": 43.0,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.726477024070023e-06,
+      "loss": 38.0312,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.770240700218819e-06,
+      "loss": 42.4375,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8140043763676155e-06,
+      "loss": 41.625,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8577680525164115e-06,
+      "loss": 41.25,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.901531728665208e-06,
+      "loss": 30.7656,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.945295404814005e-06,
+      "loss": 34.875,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.989059080962801e-06,
+      "loss": 30.625,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.032822757111597e-06,
+      "loss": 36.6875,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.076586433260394e-06,
+      "loss": 33.0625,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.120350109409191e-06,
+      "loss": 32.6875,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.164113785557987e-06,
+      "loss": 24.0469,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.207877461706784e-06,
+      "loss": 37.5,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.25164113785558e-06,
+      "loss": 31.8438,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.295404814004377e-06,
+      "loss": 31.375,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.3391684901531735e-06,
+      "loss": 31.4688,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.3829321663019695e-06,
+      "loss": 24.4844,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.426695842450766e-06,
+      "loss": 28.875,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.470459518599562e-06,
+      "loss": 26.25,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.514223194748359e-06,
+      "loss": 25.8438,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.557986870897156e-06,
+      "loss": 32.375,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.601750547045952e-06,
+      "loss": 24.2344,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.645514223194749e-06,
+      "loss": 26.4375,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.689277899343545e-06,
+      "loss": 29.5,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.733041575492342e-06,
+      "loss": 32.375,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.776805251641139e-06,
+      "loss": 20.8281,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.820568927789935e-06,
+      "loss": 26.6562,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.864332603938732e-06,
+      "loss": 22.4062,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.908096280087528e-06,
+      "loss": 29.75,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9518599562363245e-06,
+      "loss": 25.7188,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.995623632385121e-06,
+      "loss": 26.2188,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.039387308533917e-06,
+      "loss": 23.1562,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.083150984682714e-06,
+      "loss": 16.125,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.12691466083151e-06,
+      "loss": 27.0,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.170678336980307e-06,
+      "loss": 19.4688,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.214442013129104e-06,
+      "loss": 24.25,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.2582056892779e-06,
+      "loss": 18.4375,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.301969365426697e-06,
+      "loss": 21.6562,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.345733041575493e-06,
+      "loss": 20.9375,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.38949671772429e-06,
+      "loss": 18.875,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.433260393873086e-06,
+      "loss": 10.75,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.4770240700218825e-06,
+      "loss": 19.0156,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.520787746170679e-06,
+      "loss": 12.6562,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.564551422319475e-06,
+      "loss": 14.9062,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.608315098468272e-06,
+      "loss": 16.9688,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.652078774617068e-06,
+      "loss": 13.875,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.695842450765865e-06,
+      "loss": 16.6875,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.739606126914662e-06,
+      "loss": 17.1094,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.783369803063458e-06,
+      "loss": 9.375,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.827133479212255e-06,
+      "loss": 15.0469,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.87089715536105e-06,
+      "loss": 16.7812,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.914660831509848e-06,
+      "loss": 7.4375,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.9584245076586446e-06,
+      "loss": 7.625,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.00218818380744e-06,
+      "loss": 12.9531,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.0459518599562374e-06,
+      "loss": 9.5469,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.089715536105033e-06,
+      "loss": 16.125,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.1334792122538295e-06,
+      "loss": 15.9062,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.177242888402627e-06,
+      "loss": 12.2812,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.221006564551422e-06,
+      "loss": 14.0312,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.26477024070022e-06,
+      "loss": 10.9844,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.308533916849015e-06,
+      "loss": 19.1562,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.352297592997812e-06,
+      "loss": 13.2188,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.39606126914661e-06,
+      "loss": 12.625,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.439824945295405e-06,
+      "loss": 15.4375,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.483588621444202e-06,
+      "loss": 16.7344,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.527352297592998e-06,
+      "loss": 12.7969,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.571115973741795e-06,
+      "loss": 10.4844,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.6148796498905915e-06,
+      "loss": 15.125,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.658643326039388e-06,
+      "loss": 11.25,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.702407002188184e-06,
+      "loss": 11.3438,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.746170678336981e-06,
+      "loss": 8.9688,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.789934354485778e-06,
+      "loss": 11.0781,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.833698030634575e-06,
+      "loss": 16.2344,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.87746170678337e-06,
+      "loss": 8.3125,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.921225382932167e-06,
+      "loss": 13.3438,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.964989059080964e-06,
+      "loss": 9.2656,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.00875273522976e-06,
+      "loss": 7.7188,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.052516411378558e-06,
+      "loss": 12.9062,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.096280087527353e-06,
+      "loss": 6.4688,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.14004376367615e-06,
+      "loss": 9.0625,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.183807439824946e-06,
+      "loss": 19.7188,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.227571115973743e-06,
+      "loss": 11.6875,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.271334792122538e-06,
+      "loss": 4.5938,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.315098468271335e-06,
+      "loss": 8.2188,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.358862144420132e-06,
+      "loss": 11.7812,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.402625820568927e-06,
+      "loss": 11.0938,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.446389496717726e-06,
+      "loss": 7.5,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.490153172866521e-06,
+      "loss": 7.2188,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.533916849015318e-06,
+      "loss": 9.9219,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.577680525164115e-06,
+      "loss": 8.7188,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.62144420131291e-06,
+      "loss": 10.5312,
+      "step": 197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.665207877461708e-06,
+      "loss": 7.3438,
+      "step": 198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.708971553610504e-06,
+      "loss": 6.9375,
+      "step": 199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.7527352297593e-06,
+      "loss": 7.3125,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.796498905908097e-06,
+      "loss": 6.4062,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.840262582056893e-06,
+      "loss": 10.3438,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.88402625820569e-06,
+      "loss": 5.0312,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.927789934354486e-06,
+      "loss": 9.3438,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.971553610503283e-06,
+      "loss": 8.0,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.01531728665208e-06,
+      "loss": 13.4844,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.059080962800875e-06,
+      "loss": 10.1875,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.102844638949672e-06,
+      "loss": 5.0625,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.146608315098469e-06,
+      "loss": 12.7344,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.190371991247266e-06,
+      "loss": 3.9531,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.234135667396063e-06,
+      "loss": 12.6875,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.277899343544858e-06,
+      "loss": 6.0156,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.321663019693655e-06,
+      "loss": 10.5625,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.365426695842451e-06,
+      "loss": 5.7188,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.409190371991248e-06,
+      "loss": 5.6875,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.452954048140045e-06,
+      "loss": 6.2812,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.49671772428884e-06,
+      "loss": 7.4844,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.540481400437637e-06,
+      "loss": 6.0938,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.584245076586434e-06,
+      "loss": 4.8438,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.628008752735231e-06,
+      "loss": 10.1875,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.671772428884028e-06,
+      "loss": 6.9062,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.715536105032823e-06,
+      "loss": 6.1562,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.75929978118162e-06,
+      "loss": 7.6719,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.803063457330417e-06,
+      "loss": 9.9219,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.846827133479214e-06,
+      "loss": 5.375,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89059080962801e-06,
+      "loss": 10.3125,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.934354485776806e-06,
+      "loss": 8.0156,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.978118161925602e-06,
+      "loss": 9.75,
+      "step": 228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.00218818380744e-05,
+      "loss": 10.7969,
+      "step": 229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0065645514223194e-05,
+      "loss": 8.25,
+      "step": 230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0109409190371993e-05,
+      "loss": 5.5,
+      "step": 231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0153172866520788e-05,
+      "loss": 11.875,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0196936542669585e-05,
+      "loss": 7.9062,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0240700218818382e-05,
+      "loss": 13.875,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0284463894967179e-05,
+      "loss": 9.4062,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0328227571115974e-05,
+      "loss": 5.8125,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0371991247264772e-05,
+      "loss": 8.2188,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0415754923413568e-05,
+      "loss": 6.5781,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0459518599562364e-05,
+      "loss": 7.2188,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.050328227571116e-05,
+      "loss": 9.0625,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0547045951859958e-05,
+      "loss": 5.5469,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0590809628008753e-05,
+      "loss": 7.7188,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.063457330415755e-05,
+      "loss": 7.375,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0678336980306347e-05,
+      "loss": 12.4062,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0722100656455144e-05,
+      "loss": 9.8281,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0765864332603939e-05,
+      "loss": 8.0625,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0809628008752738e-05,
+      "loss": 8.9062,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0853391684901533e-05,
+      "loss": 7.9219,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.089715536105033e-05,
+      "loss": 7.375,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0940919037199125e-05,
+      "loss": 8.0156,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0984682713347923e-05,
+      "loss": 3.5938,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1028446389496719e-05,
+      "loss": 8.6719,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1072210065645514e-05,
+      "loss": 8.7812,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1115973741794312e-05,
+      "loss": 5.2812,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1159737417943109e-05,
+      "loss": 6.9062,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1203501094091904e-05,
+      "loss": 6.2812,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.12472647702407e-05,
+      "loss": 5.5312,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1291028446389498e-05,
+      "loss": 4.9375,
+      "step": 258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1334792122538293e-05,
+      "loss": 4.8125,
+      "step": 259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.137855579868709e-05,
+      "loss": 5.3125,
+      "step": 260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1422319474835889e-05,
+      "loss": 5.375,
+      "step": 261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1466083150984684e-05,
+      "loss": 5.1562,
+      "step": 262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1509846827133479e-05,
+      "loss": 9.25,
+      "step": 263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1553610503282277e-05,
+      "loss": 3.4375,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1597374179431074e-05,
+      "loss": 7.9375,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.164113785557987e-05,
+      "loss": 8.3125,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1684901531728665e-05,
+      "loss": 5.625,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1728665207877463e-05,
+      "loss": 9.0781,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1772428884026258e-05,
+      "loss": 7.3281,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1816192560175055e-05,
+      "loss": 6.1562,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1859956236323854e-05,
+      "loss": 8.0625,
+      "step": 271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1903719912472649e-05,
+      "loss": 7.3906,
+      "step": 272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1947483588621444e-05,
+      "loss": 6.7656,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1991247264770243e-05,
+      "loss": 4.2812,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2035010940919038e-05,
+      "loss": 5.2969,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2078774617067835e-05,
+      "loss": 7.6562,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.212253829321663e-05,
+      "loss": 4.8281,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2166301969365428e-05,
+      "loss": 7.1562,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2210065645514224e-05,
+      "loss": 8.0312,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.225382932166302e-05,
+      "loss": 5.0625,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2297592997811817e-05,
+      "loss": 6.1875,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2341356673960614e-05,
+      "loss": 6.625,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.238512035010941e-05,
+      "loss": 7.1406,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2428884026258208e-05,
+      "loss": 4.1719,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2472647702407003e-05,
+      "loss": 5.7812,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.25164113785558e-05,
+      "loss": 6.6562,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2560175054704595e-05,
+      "loss": 10.5625,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2603938730853394e-05,
+      "loss": 6.6875,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2647702407002189e-05,
+      "loss": 5.875,
+      "step": 289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2691466083150986e-05,
+      "loss": 5.7812,
+      "step": 290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2735229759299782e-05,
+      "loss": 6.9844,
+      "step": 291
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.277899343544858e-05,
+      "loss": 8.5469,
+      "step": 292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2822757111597374e-05,
+      "loss": 6.0156,
+      "step": 293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2866520787746171e-05,
+      "loss": 8.3438,
+      "step": 294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2910284463894968e-05,
+      "loss": 7.5938,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2954048140043765e-05,
+      "loss": 7.5781,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.299781181619256e-05,
+      "loss": 4.6562,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3041575492341359e-05,
+      "loss": 4.2812,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3085339168490154e-05,
+      "loss": 7.3125,
+      "step": 299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.312910284463895e-05,
+      "loss": 6.3594,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3172866520787748e-05,
+      "loss": 7.125,
+      "step": 301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3216630196936544e-05,
+      "loss": 8.2812,
+      "step": 302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.326039387308534e-05,
+      "loss": 6.0625,
+      "step": 303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3304157549234137e-05,
+      "loss": 2.625,
+      "step": 304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3347921225382933e-05,
+      "loss": 7.1719,
+      "step": 305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.339168490153173e-05,
+      "loss": 5.5938,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3435448577680525e-05,
+      "loss": 7.5625,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3479212253829324e-05,
+      "loss": 9.2188,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3522975929978119e-05,
+      "loss": 7.25,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3566739606126916e-05,
+      "loss": 8.1562,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3610503282275713e-05,
+      "loss": 8.5938,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.365426695842451e-05,
+      "loss": 5.7031,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3698030634573305e-05,
+      "loss": 7.2031,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.37417943107221e-05,
+      "loss": 10.125,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3785557986870899e-05,
+      "loss": 4.8125,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3829321663019695e-05,
+      "loss": 9.0469,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.387308533916849e-05,
+      "loss": 6.75,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3916849015317289e-05,
+      "loss": 5.2188,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3960612691466084e-05,
+      "loss": 10.9219,
+      "step": 319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.400437636761488e-05,
+      "loss": 8.6094,
+      "step": 320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4048140043763678e-05,
+      "loss": 9.2656,
+      "step": 321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4091903719912475e-05,
+      "loss": 3.8594,
+      "step": 322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.413566739606127e-05,
+      "loss": 6.3125,
+      "step": 323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4179431072210065e-05,
+      "loss": 10.5781,
+      "step": 324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4223194748358864e-05,
+      "loss": 5.9062,
+      "step": 325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4266958424507659e-05,
+      "loss": 4.875,
+      "step": 326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4310722100656456e-05,
+      "loss": 5.6875,
+      "step": 327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4354485776805254e-05,
+      "loss": 8.1094,
+      "step": 328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.439824945295405e-05,
+      "loss": 7.0938,
+      "step": 329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4442013129102845e-05,
+      "loss": 4.8906,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4485776805251643e-05,
+      "loss": 5.1719,
+      "step": 331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.452954048140044e-05,
+      "loss": 5.0625,
+      "step": 332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4573304157549235e-05,
+      "loss": 4.75,
+      "step": 333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.461706783369803e-05,
+      "loss": 5.4062,
+      "step": 334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4660831509846829e-05,
+      "loss": 4.9375,
+      "step": 335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4704595185995624e-05,
+      "loss": 6.8594,
+      "step": 336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4748358862144421e-05,
+      "loss": 6.6875,
+      "step": 337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.479212253829322e-05,
+      "loss": 6.25,
+      "step": 338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4835886214442015e-05,
+      "loss": 7.4844,
+      "step": 339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.487964989059081e-05,
+      "loss": 3.8125,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4923413566739607e-05,
+      "loss": 5.7188,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4967177242888404e-05,
+      "loss": 10.6406,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.50109409190372e-05,
+      "loss": 5.8438,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5054704595185996e-05,
+      "loss": 5.0156,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5098468271334794e-05,
+      "loss": 4.9062,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.514223194748359e-05,
+      "loss": 5.8438,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5185995623632386e-05,
+      "loss": 6.6094,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5229759299781183e-05,
+      "loss": 7.5625,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5273522975929978e-05,
+      "loss": 6.1875,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.5317286652078775e-05,
+      "loss": 6.9062,
+      "step": 350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5361050328227572e-05,
+      "loss": 5.75,
+      "step": 351
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.540481400437637e-05,
+      "loss": 6.125,
+      "step": 352
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5448577680525166e-05,
+      "loss": 6.4062,
+      "step": 353
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5492341356673962e-05,
+      "loss": 5.6719,
+      "step": 354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.553610503282276e-05,
+      "loss": 6.1719,
+      "step": 355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5579868708971556e-05,
+      "loss": 6.8125,
+      "step": 356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.562363238512035e-05,
+      "loss": 5.8438,
+      "step": 357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.566739606126915e-05,
+      "loss": 7.9375,
+      "step": 358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5711159737417943e-05,
+      "loss": 4.8125,
+      "step": 359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.575492341356674e-05,
+      "loss": 4.7188,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5798687089715537e-05,
+      "loss": 6.3281,
+      "step": 361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5842450765864334e-05,
+      "loss": 12.7344,
+      "step": 362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.588621444201313e-05,
+      "loss": 3.3594,
+      "step": 363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5929978118161928e-05,
+      "loss": 6.1562,
+      "step": 364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5973741794310725e-05,
+      "loss": 4.375,
+      "step": 365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.601750547045952e-05,
+      "loss": 10.8438,
+      "step": 366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6061269146608315e-05,
+      "loss": 8.9531,
+      "step": 367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6105032822757115e-05,
+      "loss": 6.9062,
+      "step": 368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.614879649890591e-05,
+      "loss": 5.4219,
+      "step": 369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6192560175054705e-05,
+      "loss": 2.7188,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6236323851203502e-05,
+      "loss": 6.3281,
+      "step": 371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.62800875273523e-05,
+      "loss": 7.125,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6323851203501096e-05,
+      "loss": 11.3125,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6367614879649893e-05,
+      "loss": 6.2188,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.641137855579869e-05,
+      "loss": 4.9844,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6455142231947487e-05,
+      "loss": 6.125,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.649890590809628e-05,
+      "loss": 6.9844,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6542669584245077e-05,
+      "loss": 7.0938,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6586433260393874e-05,
+      "loss": 11.5625,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.663019693654267e-05,
+      "loss": 6.2031,
+      "step": 380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6673960612691468e-05,
+      "loss": 5.4688,
+      "step": 381
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6717724288840264e-05,
+      "loss": 8.7188,
+      "step": 382
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.676148796498906e-05,
+      "loss": 5.5,
+      "step": 383
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6805251641137855e-05,
+      "loss": 6.5938,
+      "step": 384
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6849015317286655e-05,
+      "loss": 4.1875,
+      "step": 385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6892778993435452e-05,
+      "loss": 6.8125,
+      "step": 386
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6936542669584245e-05,
+      "loss": 6.7812,
+      "step": 387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6980306345733042e-05,
+      "loss": 4.2812,
+      "step": 388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.702407002188184e-05,
+      "loss": 6.6875,
+      "step": 389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7067833698030636e-05,
+      "loss": 5.1875,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7111597374179433e-05,
+      "loss": 4.0938,
+      "step": 391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.715536105032823e-05,
+      "loss": 6.6875,
+      "step": 392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7199124726477026e-05,
+      "loss": 5.25,
+      "step": 393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.724288840262582e-05,
+      "loss": 7.5938,
+      "step": 394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.728665207877462e-05,
+      "loss": 6.3125,
+      "step": 395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7330415754923417e-05,
+      "loss": 4.1875,
+      "step": 396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.737417943107221e-05,
+      "loss": 7.1719,
+      "step": 397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7417943107221007e-05,
+      "loss": 9.5781,
+      "step": 398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7461706783369804e-05,
+      "loss": 7.625,
+      "step": 399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.75054704595186e-05,
+      "loss": 5.0312,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7549234135667398e-05,
+      "loss": 4.6719,
+      "step": 401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7592997811816195e-05,
+      "loss": 7.0156,
+      "step": 402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.763676148796499e-05,
+      "loss": 7.1562,
+      "step": 403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7680525164113785e-05,
+      "loss": 4.8906,
+      "step": 404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7724288840262585e-05,
+      "loss": 5.75,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.776805251641138e-05,
+      "loss": 6.125,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7811816192560176e-05,
+      "loss": 4.2812,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7855579868708973e-05,
+      "loss": 6.3438,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.789934354485777e-05,
+      "loss": 7.6406,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7943107221006566e-05,
+      "loss": 11.5156,
+      "step": 410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7986870897155363e-05,
+      "loss": 6.1875,
+      "step": 411
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.803063457330416e-05,
+      "loss": 8.375,
+      "step": 412
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8074398249452957e-05,
+      "loss": 4.7812,
+      "step": 413
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.811816192560175e-05,
+      "loss": 5.1875,
+      "step": 414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.816192560175055e-05,
+      "loss": 5.0156,
+      "step": 415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8205689277899344e-05,
+      "loss": 7.0312,
+      "step": 416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.824945295404814e-05,
+      "loss": 5.6562,
+      "step": 417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8293216630196938e-05,
+      "loss": 6.3438,
+      "step": 418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8336980306345735e-05,
+      "loss": 4.625,
+      "step": 419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.838074398249453e-05,
+      "loss": 6.8125,
+      "step": 420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8424507658643328e-05,
+      "loss": 5.9844,
+      "step": 421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8468271334792125e-05,
+      "loss": 6.125,
+      "step": 422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8512035010940922e-05,
+      "loss": 2.8125,
+      "step": 423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8555798687089715e-05,
+      "loss": 3.1406,
+      "step": 424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8599562363238512e-05,
+      "loss": 4.0938,
+      "step": 425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.864332603938731e-05,
+      "loss": 5.625,
+      "step": 426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8687089715536106e-05,
+      "loss": 5.2188,
+      "step": 427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8730853391684903e-05,
+      "loss": 5.75,
+      "step": 428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.87746170678337e-05,
+      "loss": 5.8125,
+      "step": 429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8818380743982497e-05,
+      "loss": 3.0625,
+      "step": 430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8862144420131293e-05,
+      "loss": 3.75,
+      "step": 431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.890590809628009e-05,
+      "loss": 6.2188,
+      "step": 432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8949671772428887e-05,
+      "loss": 1.9688,
+      "step": 433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.899343544857768e-05,
+      "loss": 3.4219,
+      "step": 434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9037199124726478e-05,
+      "loss": 3.4062,
+      "step": 435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9080962800875274e-05,
+      "loss": 4.625,
+      "step": 436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.912472647702407e-05,
+      "loss": 7.0938,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9168490153172868e-05,
+      "loss": 7.4531,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9212253829321665e-05,
+      "loss": 4.8125,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9256017505470462e-05,
+      "loss": 4.625,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.929978118161926e-05,
+      "loss": 5.875,
+      "step": 441
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9343544857768056e-05,
+      "loss": 6.7188,
+      "step": 442
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9387308533916852e-05,
+      "loss": 11.3906,
+      "step": 443
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9431072210065646e-05,
+      "loss": 8.3906,
+      "step": 444
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9474835886214443e-05,
+      "loss": 8.8906,
+      "step": 445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.951859956236324e-05,
+      "loss": 4.5469,
+      "step": 446
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9562363238512036e-05,
+      "loss": 5.3125,
+      "step": 447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9606126914660833e-05,
+      "loss": 3.1562,
+      "step": 448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.964989059080963e-05,
+      "loss": 5.125,
+      "step": 449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9693654266958427e-05,
+      "loss": 4.625,
+      "step": 450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.973741794310722e-05,
+      "loss": 6.1875,
+      "step": 451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.978118161925602e-05,
+      "loss": 7.8906,
+      "step": 452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9824945295404818e-05,
+      "loss": 6.2031,
+      "step": 453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.986870897155361e-05,
+      "loss": 9.2969,
+      "step": 454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9912472647702408e-05,
+      "loss": 8.0469,
+      "step": 455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9956236323851205e-05,
+      "loss": 4.0625,
+      "step": 456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 4.1562,
+      "step": 457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.00437636761488e-05,
+      "loss": 10.4375,
+      "step": 458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0087527352297592e-05,
+      "loss": 7.0781,
+      "step": 459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.013129102844639e-05,
+      "loss": 4.0312,
+      "step": 460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.017505470459519e-05,
+      "loss": 5.9375,
+      "step": 461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0218818380743986e-05,
+      "loss": 10.2031,
+      "step": 462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0262582056892783e-05,
+      "loss": 8.8438,
+      "step": 463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0306345733041576e-05,
+      "loss": 5.8906,
+      "step": 464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0350109409190373e-05,
+      "loss": 7.5625,
+      "step": 465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.039387308533917e-05,
+      "loss": 5.9219,
+      "step": 466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0437636761487963e-05,
+      "loss": 6.5938,
+      "step": 467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0481400437636764e-05,
+      "loss": 8.1875,
+      "step": 468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.052516411378556e-05,
+      "loss": 7.0781,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.0568927789934357e-05,
+      "loss": 6.9531,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.061269146608315e-05,
+      "loss": 8.1875,
+      "step": 471
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0656455142231948e-05,
+      "loss": 5.5,
+      "step": 472
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0700218818380745e-05,
+      "loss": 5.2969,
+      "step": 473
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0743982494529545e-05,
+      "loss": 7.9688,
+      "step": 474
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0787746170678342e-05,
+      "loss": 6.4688,
+      "step": 475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0831509846827135e-05,
+      "loss": 4.6719,
+      "step": 476
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0875273522975932e-05,
+      "loss": 3.9219,
+      "step": 477
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.091903719912473e-05,
+      "loss": 9.3125,
+      "step": 478
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0962800875273522e-05,
+      "loss": 8.0156,
+      "step": 479
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.100656455142232e-05,
+      "loss": 5.1562,
+      "step": 480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.105032822757112e-05,
+      "loss": 3.3594,
+      "step": 481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1094091903719916e-05,
+      "loss": 7.1406,
+      "step": 482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.113785557986871e-05,
+      "loss": 5.125,
+      "step": 483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1181619256017507e-05,
+      "loss": 8.5781,
+      "step": 484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1225382932166304e-05,
+      "loss": 5.875,
+      "step": 485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.12691466083151e-05,
+      "loss": 6.5938,
+      "step": 486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1312910284463894e-05,
+      "loss": 8.4688,
+      "step": 487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1356673960612694e-05,
+      "loss": 6.1406,
+      "step": 488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.140043763676149e-05,
+      "loss": 6.4375,
+      "step": 489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1444201312910288e-05,
+      "loss": 6.0625,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.148796498905908e-05,
+      "loss": 6.1875,
+      "step": 491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1531728665207878e-05,
+      "loss": 4.3125,
+      "step": 492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1575492341356675e-05,
+      "loss": 5.5156,
+      "step": 493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1619256017505475e-05,
+      "loss": 4.0,
+      "step": 494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.166301969365427e-05,
+      "loss": 4.0469,
+      "step": 495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1706783369803066e-05,
+      "loss": 5.7031,
+      "step": 496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1750547045951862e-05,
+      "loss": 5.0625,
+      "step": 497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.179431072210066e-05,
+      "loss": 5.9219,
+      "step": 498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1838074398249453e-05,
+      "loss": 7.2969,
+      "step": 499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.188183807439825e-05,
+      "loss": 6.4375,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.192560175054705e-05,
+      "loss": 5.6094,
+      "step": 501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1969365426695847e-05,
+      "loss": 6.1094,
+      "step": 502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.201312910284464e-05,
+      "loss": 7.8438,
+      "step": 503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2056892778993437e-05,
+      "loss": 4.2812,
+      "step": 504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2100656455142234e-05,
+      "loss": 8.6562,
+      "step": 505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2144420131291027e-05,
+      "loss": 6.0312,
+      "step": 506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2188183807439824e-05,
+      "loss": 4.1875,
+      "step": 507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2231947483588624e-05,
+      "loss": 5.6562,
+      "step": 508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.227571115973742e-05,
+      "loss": 5.25,
+      "step": 509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2319474835886218e-05,
+      "loss": 3.6875,
+      "step": 510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.236323851203501e-05,
+      "loss": 8.4219,
+      "step": 511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.240700218818381e-05,
+      "loss": 3.75,
+      "step": 512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2450765864332605e-05,
+      "loss": 7.2812,
+      "step": 513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.24945295404814e-05,
+      "loss": 3.9688,
+      "step": 514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.25382932166302e-05,
+      "loss": 4.9375,
+      "step": 515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2582056892778996e-05,
+      "loss": 4.0312,
+      "step": 516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2625820568927793e-05,
+      "loss": 7.9688,
+      "step": 517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2669584245076586e-05,
+      "loss": 6.6406,
+      "step": 518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2713347921225383e-05,
+      "loss": 4.1562,
+      "step": 519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.275711159737418e-05,
+      "loss": 5.875,
+      "step": 520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.280087527352298e-05,
+      "loss": 8.6875,
+      "step": 521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2844638949671777e-05,
+      "loss": 5.625,
+      "step": 522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.288840262582057e-05,
+      "loss": 3.75,
+      "step": 523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2932166301969367e-05,
+      "loss": 4.8906,
+      "step": 524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.2975929978118164e-05,
+      "loss": 7.4375,
+      "step": 525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.3019693654266958e-05,
+      "loss": 7.9531,
+      "step": 526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.3063457330415755e-05,
+      "loss": 7.4375,
+      "step": 527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.3107221006564555e-05,
+      "loss": 3.8125,
+      "step": 528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.3150984682713352e-05,
+      "loss": 5.5469,
+      "step": 529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.319474835886215e-05,
+      "loss": 5.875,
+      "step": 530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.3238512035010942e-05,
+      "loss": 6.2812,
+      "step": 531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.328227571115974e-05,
+      "loss": 9.1406,
+      "step": 532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3326039387308536e-05,
+      "loss": 5.375,
+      "step": 533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.336980306345733e-05,
+      "loss": 5.125,
+      "step": 534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.341356673960613e-05,
+      "loss": 3.8125,
+      "step": 535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3457330415754926e-05,
+      "loss": 5.3438,
+      "step": 536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3501094091903723e-05,
+      "loss": 5.2812,
+      "step": 537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3544857768052517e-05,
+      "loss": 6.9375,
+      "step": 538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3588621444201314e-05,
+      "loss": 4.1562,
+      "step": 539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.363238512035011e-05,
+      "loss": 3.625,
+      "step": 540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3676148796498907e-05,
+      "loss": 5.4062,
+      "step": 541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3719912472647707e-05,
+      "loss": 10.7188,
+      "step": 542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.37636761487965e-05,
+      "loss": 8.3594,
+      "step": 543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3807439824945298e-05,
+      "loss": 4.6719,
+      "step": 544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3851203501094095e-05,
+      "loss": 5.4844,
+      "step": 545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3894967177242888e-05,
+      "loss": 6.8438,
+      "step": 546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3938730853391685e-05,
+      "loss": 5.0312,
+      "step": 547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3982494529540485e-05,
+      "loss": 3.2656,
+      "step": 548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4026258205689282e-05,
+      "loss": 9.375,
+      "step": 549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4070021881838076e-05,
+      "loss": 5.7188,
+      "step": 550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4113785557986872e-05,
+      "loss": 8.0625,
+      "step": 551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.415754923413567e-05,
+      "loss": 8.1875,
+      "step": 552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4201312910284466e-05,
+      "loss": 5.0156,
+      "step": 553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.424507658643326e-05,
+      "loss": 9.7188,
+      "step": 554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.428884026258206e-05,
+      "loss": 4.375,
+      "step": 555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4332603938730857e-05,
+      "loss": 5.875,
+      "step": 556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4376367614879654e-05,
+      "loss": 5.4844,
+      "step": 557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4420131291028447e-05,
+      "loss": 5.6875,
+      "step": 558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4463894967177244e-05,
+      "loss": 7.9531,
+      "step": 559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.450765864332604e-05,
+      "loss": 4.7656,
+      "step": 560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4551422319474834e-05,
+      "loss": 7.0781,
+      "step": 561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4595185995623634e-05,
+      "loss": 10.0625,
+      "step": 562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.463894967177243e-05,
+      "loss": 7.0312,
+      "step": 563
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4682713347921228e-05,
+      "loss": 7.6719,
+      "step": 564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4726477024070025e-05,
+      "loss": 4.4375,
+      "step": 565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.477024070021882e-05,
+      "loss": 6.1875,
+      "step": 566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4814004376367615e-05,
+      "loss": 5.7188,
+      "step": 567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4857768052516416e-05,
+      "loss": 6.9688,
+      "step": 568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4901531728665213e-05,
+      "loss": 3.1562,
+      "step": 569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4945295404814006e-05,
+      "loss": 4.3438,
+      "step": 570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4989059080962803e-05,
+      "loss": 4.1562,
+      "step": 571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.50328227571116e-05,
+      "loss": 7.4062,
+      "step": 572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5076586433260393e-05,
+      "loss": 8.25,
+      "step": 573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.512035010940919e-05,
+      "loss": 5.25,
+      "step": 574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.516411378555799e-05,
+      "loss": 7.5,
+      "step": 575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5207877461706787e-05,
+      "loss": 10.9062,
+      "step": 576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5251641137855584e-05,
+      "loss": 8.4688,
+      "step": 577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5295404814004377e-05,
+      "loss": 10.5781,
+      "step": 578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5339168490153174e-05,
+      "loss": 4.0312,
+      "step": 579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.538293216630197e-05,
+      "loss": 6.625,
+      "step": 580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5426695842450765e-05,
+      "loss": 6.7969,
+      "step": 581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5470459518599565e-05,
+      "loss": 5.9688,
+      "step": 582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5514223194748362e-05,
+      "loss": 4.5312,
+      "step": 583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.555798687089716e-05,
+      "loss": 5.9062,
+      "step": 584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5601750547045952e-05,
+      "loss": 6.1562,
+      "step": 585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.564551422319475e-05,
+      "loss": 5.4531,
+      "step": 586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5689277899343546e-05,
+      "loss": 9.0625,
+      "step": 587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5733041575492343e-05,
+      "loss": 5.8281,
+      "step": 588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5776805251641143e-05,
+      "loss": 4.375,
+      "step": 589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5820568927789936e-05,
+      "loss": 4.75,
+      "step": 590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5864332603938733e-05,
+      "loss": 6.5312,
+      "step": 591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.590809628008753e-05,
+      "loss": 7.0625,
+      "step": 592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.5951859956236324e-05,
+      "loss": 5.5625,
+      "step": 593
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.599562363238512e-05,
+      "loss": 4.7812,
+      "step": 594
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.603938730853392e-05,
+      "loss": 7.5,
+      "step": 595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6083150984682718e-05,
+      "loss": 6.9062,
+      "step": 596
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6126914660831514e-05,
+      "loss": 6.8281,
+      "step": 597
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6170678336980308e-05,
+      "loss": 6.875,
+      "step": 598
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6214442013129105e-05,
+      "loss": 4.4062,
+      "step": 599
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.62582056892779e-05,
+      "loss": 5.5,
+      "step": 600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6301969365426695e-05,
+      "loss": 1.75,
+      "step": 601
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6345733041575495e-05,
+      "loss": 8.4531,
+      "step": 602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6389496717724292e-05,
+      "loss": 4.6562,
+      "step": 603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.643326039387309e-05,
+      "loss": 3.9688,
+      "step": 604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6477024070021882e-05,
+      "loss": 8.125,
+      "step": 605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.652078774617068e-05,
+      "loss": 7.0938,
+      "step": 606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6564551422319476e-05,
+      "loss": 11.7812,
+      "step": 607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6608315098468273e-05,
+      "loss": 7.6094,
+      "step": 608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6652078774617073e-05,
+      "loss": 5.5312,
+      "step": 609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6695842450765867e-05,
+      "loss": 5.0625,
+      "step": 610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6739606126914664e-05,
+      "loss": 5.1406,
+      "step": 611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.678336980306346e-05,
+      "loss": 4.9375,
+      "step": 612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6827133479212254e-05,
+      "loss": 6.0156,
+      "step": 613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.687089715536105e-05,
+      "loss": 4.75,
+      "step": 614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.691466083150985e-05,
+      "loss": 8.5,
+      "step": 615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6958424507658648e-05,
+      "loss": 7.125,
+      "step": 616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.700218818380744e-05,
+      "loss": 6.8906,
+      "step": 617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7045951859956238e-05,
+      "loss": 4.0469,
+      "step": 618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7089715536105035e-05,
+      "loss": 5.3594,
+      "step": 619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7133479212253832e-05,
+      "loss": 6.0,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7177242888402625e-05,
+      "loss": 7.8281,
+      "step": 621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7221006564551426e-05,
+      "loss": 7.6406,
+      "step": 622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7264770240700223e-05,
+      "loss": 8.0625,
+      "step": 623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.730853391684902e-05,
+      "loss": 3.875,
+      "step": 624
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7352297592997813e-05,
+      "loss": 5.2812,
+      "step": 625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.739606126914661e-05,
+      "loss": 5.2812,
+      "step": 626
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7439824945295407e-05,
+      "loss": 5.8125,
+      "step": 627
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.74835886214442e-05,
+      "loss": 7.7969,
+      "step": 628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7527352297593e-05,
+      "loss": 6.3594,
+      "step": 629
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7571115973741797e-05,
+      "loss": 7.7812,
+      "step": 630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7614879649890594e-05,
+      "loss": 4.25,
+      "step": 631
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.765864332603939e-05,
+      "loss": 6.3438,
+      "step": 632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7702407002188184e-05,
+      "loss": 4.7812,
+      "step": 633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.774617067833698e-05,
+      "loss": 6.6406,
+      "step": 634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7789934354485778e-05,
+      "loss": 7.5156,
+      "step": 635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7833698030634578e-05,
+      "loss": 5.2656,
+      "step": 636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7877461706783372e-05,
+      "loss": 2.7031,
+      "step": 637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.792122538293217e-05,
+      "loss": 5.7188,
+      "step": 638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7964989059080965e-05,
+      "loss": 5.8125,
+      "step": 639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.800875273522976e-05,
+      "loss": 5.9062,
+      "step": 640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8052516411378556e-05,
+      "loss": 4.7656,
+      "step": 641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8096280087527356e-05,
+      "loss": 5.3438,
+      "step": 642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8140043763676153e-05,
+      "loss": 6.2031,
+      "step": 643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.818380743982495e-05,
+      "loss": 6.9375,
+      "step": 644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8227571115973743e-05,
+      "loss": 3.1406,
+      "step": 645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.827133479212254e-05,
+      "loss": 5.7812,
+      "step": 646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8315098468271337e-05,
+      "loss": 4.4062,
+      "step": 647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.835886214442013e-05,
+      "loss": 2.7812,
+      "step": 648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.840262582056893e-05,
+      "loss": 5.5469,
+      "step": 649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8446389496717728e-05,
+      "loss": 4.5938,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8490153172866524e-05,
+      "loss": 3.8594,
+      "step": 651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8533916849015318e-05,
+      "loss": 6.9375,
+      "step": 652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.8577680525164115e-05,
+      "loss": 8.0938,
+      "step": 653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.862144420131291e-05,
+      "loss": 5.0156,
+      "step": 654
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.866520787746171e-05,
+      "loss": 3.4062,
+      "step": 655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.870897155361051e-05,
+      "loss": 6.4375,
+      "step": 656
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8752735229759302e-05,
+      "loss": 6.125,
+      "step": 657
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.87964989059081e-05,
+      "loss": 5.1562,
+      "step": 658
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8840262582056896e-05,
+      "loss": 3.6406,
+      "step": 659
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.888402625820569e-05,
+      "loss": 4.6875,
+      "step": 660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8927789934354486e-05,
+      "loss": 7.625,
+      "step": 661
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8971553610503286e-05,
+      "loss": 5.9688,
+      "step": 662
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9015317286652083e-05,
+      "loss": 8.7969,
+      "step": 663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.905908096280088e-05,
+      "loss": 4.2188,
+      "step": 664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9102844638949674e-05,
+      "loss": 5.0469,
+      "step": 665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.914660831509847e-05,
+      "loss": 4.5,
+      "step": 666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9190371991247267e-05,
+      "loss": 5.875,
+      "step": 667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.923413566739606e-05,
+      "loss": 3.2344,
+      "step": 668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.927789934354486e-05,
+      "loss": 7.7969,
+      "step": 669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9321663019693658e-05,
+      "loss": 11.4062,
+      "step": 670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9365426695842455e-05,
+      "loss": 3.7969,
+      "step": 671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9409190371991248e-05,
+      "loss": 6.9219,
+      "step": 672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9452954048140045e-05,
+      "loss": 7.2812,
+      "step": 673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9496717724288842e-05,
+      "loss": 8.5625,
+      "step": 674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.954048140043764e-05,
+      "loss": 4.4062,
+      "step": 675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.958424507658644e-05,
+      "loss": 8.625,
+      "step": 676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9628008752735233e-05,
+      "loss": 11.3438,
+      "step": 677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.967177242888403e-05,
+      "loss": 5.7344,
+      "step": 678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9715536105032826e-05,
+      "loss": 5.0625,
+      "step": 679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.975929978118162e-05,
+      "loss": 5.25,
+      "step": 680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9803063457330417e-05,
+      "loss": 5.4375,
+      "step": 681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9846827133479213e-05,
+      "loss": 7.375,
+      "step": 682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9890590809628014e-05,
+      "loss": 8.3594,
+      "step": 683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9934354485776807e-05,
+      "loss": 3.9844,
+      "step": 684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9978118161925604e-05,
+      "loss": 5.125,
+      "step": 685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.00218818380744e-05,
+      "loss": 4.4062,
+      "step": 686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0065645514223198e-05,
+      "loss": 3.3438,
+      "step": 687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.010940919037199e-05,
+      "loss": 4.8438,
+      "step": 688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.015317286652079e-05,
+      "loss": 3.5,
+      "step": 689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.019693654266959e-05,
+      "loss": 4.4688,
+      "step": 690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0240700218818385e-05,
+      "loss": 5.9531,
+      "step": 691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.028446389496718e-05,
+      "loss": 6.125,
+      "step": 692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0328227571115976e-05,
+      "loss": 4.2812,
+      "step": 693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0371991247264772e-05,
+      "loss": 7.9219,
+      "step": 694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0415754923413566e-05,
+      "loss": 5.125,
+      "step": 695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0459518599562366e-05,
+      "loss": 3.8125,
+      "step": 696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0503282275711163e-05,
+      "loss": 6.25,
+      "step": 697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0547045951859956e-05,
+      "loss": 3.5,
+      "step": 698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.059080962800876e-05,
+      "loss": 4.9062,
+      "step": 699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.063457330415755e-05,
+      "loss": 4.9688,
+      "step": 700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0678336980306344e-05,
+      "loss": 4.3125,
+      "step": 701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0722100656455144e-05,
+      "loss": 5.1875,
+      "step": 702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0765864332603944e-05,
+      "loss": 7.3125,
+      "step": 703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.080962800875274e-05,
+      "loss": 4.1875,
+      "step": 704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.085339168490154e-05,
+      "loss": 4.0156,
+      "step": 705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.089715536105033e-05,
+      "loss": 3.9844,
+      "step": 706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0940919037199125e-05,
+      "loss": 6.3906,
+      "step": 707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.0984682713347925e-05,
+      "loss": 5.0,
+      "step": 708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.102844638949672e-05,
+      "loss": 5.3438,
+      "step": 709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.107221006564552e-05,
+      "loss": 6.1562,
+      "step": 710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.111597374179431e-05,
+      "loss": 9.3438,
+      "step": 711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.115973741794311e-05,
+      "loss": 3.75,
+      "step": 712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.1203501094091906e-05,
+      "loss": 6.0625,
+      "step": 713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.12472647702407e-05,
+      "loss": 6.8125,
+      "step": 714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.12910284463895e-05,
+      "loss": 2.8438,
+      "step": 715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.13347921225383e-05,
+      "loss": 6.9062,
+      "step": 716
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.137855579868709e-05,
+      "loss": 7.75,
+      "step": 717
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.142231947483589e-05,
+      "loss": 4.9062,
+      "step": 718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.146608315098469e-05,
+      "loss": 8.5,
+      "step": 719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.150984682713348e-05,
+      "loss": 9.6875,
+      "step": 720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1553610503282274e-05,
+      "loss": 4.9688,
+      "step": 721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1597374179431074e-05,
+      "loss": 6.7344,
+      "step": 722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1641137855579874e-05,
+      "loss": 6.9219,
+      "step": 723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.168490153172867e-05,
+      "loss": 5.5156,
+      "step": 724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.172866520787747e-05,
+      "loss": 6.625,
+      "step": 725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.177242888402626e-05,
+      "loss": 2.4062,
+      "step": 726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1816192560175055e-05,
+      "loss": 3.9062,
+      "step": 727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.1859956236323855e-05,
+      "loss": 6.0625,
+      "step": 728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.190371991247265e-05,
+      "loss": 6.7188,
+      "step": 729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.194748358862145e-05,
+      "loss": 7.8125,
+      "step": 730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.199124726477024e-05,
+      "loss": 3.75,
+      "step": 731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.203501094091904e-05,
+      "loss": 2.625,
+      "step": 732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.2078774617067836e-05,
+      "loss": 4.4062,
+      "step": 733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.212253829321663e-05,
+      "loss": 6.4219,
+      "step": 734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.216630196936543e-05,
+      "loss": 3.6094,
+      "step": 735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.221006564551423e-05,
+      "loss": 7.9062,
+      "step": 736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.2253829321663024e-05,
+      "loss": 3.2188,
+      "step": 737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.229759299781182e-05,
+      "loss": 4.3906,
+      "step": 738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.234135667396062e-05,
+      "loss": 9.5938,
+      "step": 739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.238512035010941e-05,
+      "loss": 7.0938,
+      "step": 740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.2428884026258204e-05,
+      "loss": 4.3125,
+      "step": 741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.2472647702407005e-05,
+      "loss": 7.3906,
+      "step": 742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.2516411378555805e-05,
+      "loss": 3.5625,
+      "step": 743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.25601750547046e-05,
+      "loss": 8.3438,
+      "step": 744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.260393873085339e-05,
+      "loss": 6.2812,
+      "step": 745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.264770240700219e-05,
+      "loss": 11.0469,
+      "step": 746
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.2691466083150986e-05,
+      "loss": 11.5938,
+      "step": 747
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.2735229759299786e-05,
+      "loss": 8.7344,
+      "step": 748
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.277899343544858e-05,
+      "loss": 6.5312,
+      "step": 749
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.282275711159738e-05,
+      "loss": 7.3125,
+      "step": 750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.286652078774617e-05,
+      "loss": 5.8125,
+      "step": 751
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.291028446389497e-05,
+      "loss": 3.5312,
+      "step": 752
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.295404814004377e-05,
+      "loss": 7.7812,
+      "step": 753
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.299781181619256e-05,
+      "loss": 7.7812,
+      "step": 754
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.304157549234136e-05,
+      "loss": 7.3906,
+      "step": 755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.3085339168490154e-05,
+      "loss": 3.5625,
+      "step": 756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.3129102844638954e-05,
+      "loss": 3.7812,
+      "step": 757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.317286652078775e-05,
+      "loss": 5.875,
+      "step": 758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.321663019693655e-05,
+      "loss": 7.2031,
+      "step": 759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.326039387308534e-05,
+      "loss": 7.75,
+      "step": 760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.3304157549234135e-05,
+      "loss": 4.5312,
+      "step": 761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.3347921225382935e-05,
+      "loss": 5.2031,
+      "step": 762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.3391684901531735e-05,
+      "loss": 13.0,
+      "step": 763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.343544857768053e-05,
+      "loss": 4.9531,
+      "step": 764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.347921225382932e-05,
+      "loss": 4.5312,
+      "step": 765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.352297592997812e-05,
+      "loss": 5.0625,
+      "step": 766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.3566739606126916e-05,
+      "loss": 5.0625,
+      "step": 767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.361050328227571e-05,
+      "loss": 6.9062,
+      "step": 768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.365426695842451e-05,
+      "loss": 8.0469,
+      "step": 769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.369803063457331e-05,
+      "loss": 8.3125,
+      "step": 770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.37417943107221e-05,
+      "loss": 4.4062,
+      "step": 771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.3785557986870904e-05,
+      "loss": 4.9062,
+      "step": 772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.38293216630197e-05,
+      "loss": 3.4062,
+      "step": 773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.387308533916849e-05,
+      "loss": 8.2031,
+      "step": 774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.391684901531729e-05,
+      "loss": 6.125,
+      "step": 775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.3960612691466084e-05,
+      "loss": 8.9688,
+      "step": 776
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4004376367614885e-05,
+      "loss": 4.2969,
+      "step": 777
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.404814004376368e-05,
+      "loss": 7.4688,
+      "step": 778
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.409190371991248e-05,
+      "loss": 7.8438,
+      "step": 779
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.413566739606127e-05,
+      "loss": 8.2812,
+      "step": 780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4179431072210065e-05,
+      "loss": 4.3281,
+      "step": 781
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4223194748358865e-05,
+      "loss": 6.0,
+      "step": 782
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4266958424507666e-05,
+      "loss": 5.5625,
+      "step": 783
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.431072210065646e-05,
+      "loss": 3.625,
+      "step": 784
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.435448577680525e-05,
+      "loss": 5.2031,
+      "step": 785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.439824945295405e-05,
+      "loss": 7.0312,
+      "step": 786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4442013129102846e-05,
+      "loss": 5.2344,
+      "step": 787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.448577680525164e-05,
+      "loss": 7.3125,
+      "step": 788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.452954048140044e-05,
+      "loss": 4.7969,
+      "step": 789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.457330415754924e-05,
+      "loss": 3.9688,
+      "step": 790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4617067833698034e-05,
+      "loss": 5.0312,
+      "step": 791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4660831509846834e-05,
+      "loss": 7.9062,
+      "step": 792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.470459518599563e-05,
+      "loss": 3.9688,
+      "step": 793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.474835886214442e-05,
+      "loss": 6.9062,
+      "step": 794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.479212253829322e-05,
+      "loss": 8.3906,
+      "step": 795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4835886214442015e-05,
+      "loss": 5.9688,
+      "step": 796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4879649890590815e-05,
+      "loss": 3.25,
+      "step": 797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.492341356673961e-05,
+      "loss": 6.0469,
+      "step": 798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.496717724288841e-05,
+      "loss": 4.0,
+      "step": 799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.50109409190372e-05,
+      "loss": 3.4062,
+      "step": 800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.5054704595185996e-05,
+      "loss": 3.9219,
+      "step": 801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.5098468271334796e-05,
+      "loss": 5.25,
+      "step": 802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.514223194748359e-05,
+      "loss": 7.0,
+      "step": 803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.518599562363239e-05,
+      "loss": 3.8594,
+      "step": 804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.522975929978118e-05,
+      "loss": 7.0938,
+      "step": 805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.527352297592998e-05,
+      "loss": 5.125,
+      "step": 806
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.531728665207878e-05,
+      "loss": 5.875,
+      "step": 807
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.536105032822757e-05,
+      "loss": 9.1562,
+      "step": 808
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.540481400437637e-05,
+      "loss": 5.4062,
+      "step": 809
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.544857768052517e-05,
+      "loss": 3.3203,
+      "step": 810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.5492341356673964e-05,
+      "loss": 4.5,
+      "step": 811
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.553610503282276e-05,
+      "loss": 4.3125,
+      "step": 812
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.557986870897156e-05,
+      "loss": 9.125,
+      "step": 813
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.562363238512035e-05,
+      "loss": 7.6875,
+      "step": 814
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.566739606126915e-05,
+      "loss": 8.2031,
+      "step": 815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.5711159737417945e-05,
+      "loss": 5.9062,
+      "step": 816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.5754923413566745e-05,
+      "loss": 7.6406,
+      "step": 817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.579868708971554e-05,
+      "loss": 8.2969,
+      "step": 818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.584245076586434e-05,
+      "loss": 9.0312,
+      "step": 819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.588621444201313e-05,
+      "loss": 4.375,
+      "step": 820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.5929978118161926e-05,
+      "loss": 3.4062,
+      "step": 821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.5973741794310726e-05,
+      "loss": 6.1562,
+      "step": 822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.601750547045952e-05,
+      "loss": 9.1562,
+      "step": 823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.606126914660832e-05,
+      "loss": 5.7812,
+      "step": 824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.610503282275711e-05,
+      "loss": 9.7812,
+      "step": 825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6148796498905914e-05,
+      "loss": 5.6094,
+      "step": 826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.619256017505471e-05,
+      "loss": 5.6719,
+      "step": 827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.62363238512035e-05,
+      "loss": 5.9062,
+      "step": 828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.62800875273523e-05,
+      "loss": 6.75,
+      "step": 829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.63238512035011e-05,
+      "loss": 5.9766,
+      "step": 830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6367614879649895e-05,
+      "loss": 3.9688,
+      "step": 831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.641137855579869e-05,
+      "loss": 7.1562,
+      "step": 832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.645514223194749e-05,
+      "loss": 8.1094,
+      "step": 833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.649890590809628e-05,
+      "loss": 8.125,
+      "step": 834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6542669584245075e-05,
+      "loss": 8.3281,
+      "step": 835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6586433260393875e-05,
+      "loss": 7.2812,
+      "step": 836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6630196936542676e-05,
+      "loss": 7.2969,
+      "step": 837
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.667396061269147e-05,
+      "loss": 7.4844,
+      "step": 838
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.671772428884027e-05,
+      "loss": 5.9219,
+      "step": 839
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.676148796498906e-05,
+      "loss": 4.4531,
+      "step": 840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6805251641137856e-05,
+      "loss": 4.3125,
+      "step": 841
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6849015317286657e-05,
+      "loss": 5.5625,
+      "step": 842
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.689277899343545e-05,
+      "loss": 4.2656,
+      "step": 843
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.693654266958425e-05,
+      "loss": 8.25,
+      "step": 844
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6980306345733044e-05,
+      "loss": 5.9688,
+      "step": 845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7024070021881844e-05,
+      "loss": 8.5,
+      "step": 846
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.706783369803064e-05,
+      "loss": 3.9375,
+      "step": 847
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.711159737417943e-05,
+      "loss": 6.3438,
+      "step": 848
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.715536105032823e-05,
+      "loss": 3.9375,
+      "step": 849
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7199124726477025e-05,
+      "loss": 3.1875,
+      "step": 850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7242888402625825e-05,
+      "loss": 5.4688,
+      "step": 851
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.728665207877462e-05,
+      "loss": 8.2344,
+      "step": 852
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.733041575492342e-05,
+      "loss": 5.875,
+      "step": 853
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.737417943107221e-05,
+      "loss": 4.875,
+      "step": 854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7417943107221006e-05,
+      "loss": 4.0312,
+      "step": 855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7461706783369806e-05,
+      "loss": 4.7031,
+      "step": 856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7505470459518606e-05,
+      "loss": 5.0938,
+      "step": 857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.75492341356674e-05,
+      "loss": 4.5625,
+      "step": 858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.75929978118162e-05,
+      "loss": 4.9688,
+      "step": 859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.763676148796499e-05,
+      "loss": 3.6094,
+      "step": 860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.768052516411379e-05,
+      "loss": 5.75,
+      "step": 861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.772428884026259e-05,
+      "loss": 3.9062,
+      "step": 862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.776805251641138e-05,
+      "loss": 4.8438,
+      "step": 863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.781181619256018e-05,
+      "loss": 6.8125,
+      "step": 864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7855579868708974e-05,
+      "loss": 4.0312,
+      "step": 865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.7899343544857774e-05,
+      "loss": 3.0,
+      "step": 866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.794310722100657e-05,
+      "loss": 6.5938,
+      "step": 867
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.798687089715536e-05,
+      "loss": 6.3906,
+      "step": 868
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.803063457330416e-05,
+      "loss": 6.8125,
+      "step": 869
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8074398249452955e-05,
+      "loss": 4.5625,
+      "step": 870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8118161925601755e-05,
+      "loss": 6.6719,
+      "step": 871
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.816192560175055e-05,
+      "loss": 7.1719,
+      "step": 872
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.820568927789935e-05,
+      "loss": 5.625,
+      "step": 873
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.824945295404814e-05,
+      "loss": 5.5938,
+      "step": 874
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8293216630196936e-05,
+      "loss": 4.7344,
+      "step": 875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8336980306345736e-05,
+      "loss": 4.5938,
+      "step": 876
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.838074398249453e-05,
+      "loss": 8.0312,
+      "step": 877
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.842450765864333e-05,
+      "loss": 5.25,
+      "step": 878
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8468271334792123e-05,
+      "loss": 8.8125,
+      "step": 879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8512035010940924e-05,
+      "loss": 6.8125,
+      "step": 880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.855579868708972e-05,
+      "loss": 4.9219,
+      "step": 881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.859956236323852e-05,
+      "loss": 7.9375,
+      "step": 882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.864332603938731e-05,
+      "loss": 5.5781,
+      "step": 883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.868708971553611e-05,
+      "loss": 7.5156,
+      "step": 884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8730853391684905e-05,
+      "loss": 6.1875,
+      "step": 885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8774617067833705e-05,
+      "loss": 5.4688,
+      "step": 886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.88183807439825e-05,
+      "loss": 5.2031,
+      "step": 887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.886214442013129e-05,
+      "loss": 5.2188,
+      "step": 888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.890590809628009e-05,
+      "loss": 6.8438,
+      "step": 889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8949671772428885e-05,
+      "loss": 3.9531,
+      "step": 890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.8993435448577686e-05,
+      "loss": 2.625,
+      "step": 891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.903719912472648e-05,
+      "loss": 4.8125,
+      "step": 892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.908096280087528e-05,
+      "loss": 4.8438,
+      "step": 893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.912472647702407e-05,
+      "loss": 5.5156,
+      "step": 894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9168490153172866e-05,
+      "loss": 10.0312,
+      "step": 895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9212253829321667e-05,
+      "loss": 5.1562,
+      "step": 896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.925601750547046e-05,
+      "loss": 9.6562,
+      "step": 897
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.929978118161926e-05,
+      "loss": 6.4375,
+      "step": 898
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9343544857768054e-05,
+      "loss": 7.4844,
+      "step": 899
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9387308533916854e-05,
+      "loss": 10.0312,
+      "step": 900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.943107221006565e-05,
+      "loss": 7.5312,
+      "step": 901
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.947483588621444e-05,
+      "loss": 7.0625,
+      "step": 902
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.951859956236324e-05,
+      "loss": 10.375,
+      "step": 903
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.956236323851204e-05,
+      "loss": 10.875,
+      "step": 904
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9606126914660835e-05,
+      "loss": 6.6875,
+      "step": 905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9649890590809635e-05,
+      "loss": 3.5469,
+      "step": 906
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.969365426695843e-05,
+      "loss": 5.4375,
+      "step": 907
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.973741794310722e-05,
+      "loss": 2.2812,
+      "step": 908
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.978118161925602e-05,
+      "loss": 4.5312,
+      "step": 909
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9824945295404816e-05,
+      "loss": 5.5938,
+      "step": 910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9868708971553616e-05,
+      "loss": 6.0156,
+      "step": 911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.991247264770241e-05,
+      "loss": 7.5938,
+      "step": 912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.995623632385121e-05,
+      "loss": 7.0781,
+      "step": 913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4e-05,
+      "loss": 7.5,
+      "step": 914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999999988678851e-05,
+      "loss": 5.7812,
+      "step": 915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999999954715404e-05,
+      "loss": 5.0,
+      "step": 916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9999998981096576e-05,
+      "loss": 3.5,
+      "step": 917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9999998188616156e-05,
+      "loss": 7.0469,
+      "step": 918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999999716971276e-05,
+      "loss": 3.7188,
+      "step": 919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9999995924386425e-05,
+      "loss": 4.5938,
+      "step": 920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999999445263714e-05,
+      "loss": 4.5781,
+      "step": 921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999999275446494e-05,
+      "loss": 4.6875,
+      "step": 922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999999082986984e-05,
+      "loss": 2.625,
+      "step": 923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9999988678851856e-05,
+      "loss": 6.3125,
+      "step": 924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9999986301411014e-05,
+      "loss": 5.0938,
+      "step": 925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999998369754734e-05,
+      "loss": 6.4062,
+      "step": 926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9999980867260875e-05,
+      "loss": 6.7188,
+      "step": 927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9999977810551645e-05,
+      "loss": 2.9375,
+      "step": 928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999997452741967e-05,
+      "loss": 5.8281,
+      "step": 929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9999971017865006e-05,
+      "loss": 7.1406,
+      "step": 930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999996728188769e-05,
+      "loss": 8.25,
+      "step": 931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999996331948776e-05,
+      "loss": 5.6875,
+      "step": 932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999995913066525e-05,
+      "loss": 3.7188,
+      "step": 933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999995471542023e-05,
+      "loss": 6.2812,
+      "step": 934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999995007375273e-05,
+      "loss": 5.625,
+      "step": 935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999994520566282e-05,
+      "loss": 4.0938,
+      "step": 936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999994011115054e-05,
+      "loss": 4.5625,
+      "step": 937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999993479021596e-05,
+      "loss": 2.9062,
+      "step": 938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999992924285913e-05,
+      "loss": 5.3125,
+      "step": 939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999992346908011e-05,
+      "loss": 3.875,
+      "step": 940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999991746887898e-05,
+      "loss": 5.9375,
+      "step": 941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.99999112422558e-05,
+      "loss": 9.0781,
+      "step": 942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999990478921064e-05,
+      "loss": 3.7812,
+      "step": 943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999989810974358e-05,
+      "loss": 4.2031,
+      "step": 944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999989120385467e-05,
+      "loss": 5.1719,
+      "step": 945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999988407154403e-05,
+      "loss": 4.4062,
+      "step": 946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9999876712811704e-05,
+      "loss": 5.125,
+      "step": 947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.99998691276578e-05,
+      "loss": 5.7188,
+      "step": 948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999986131608239e-05,
+      "loss": 4.0,
+      "step": 949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9999853278085566e-05,
+      "loss": 5.0156,
+      "step": 950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999984501366741e-05,
+      "loss": 4.375,
+      "step": 951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999983652282803e-05,
+      "loss": 8.3125,
+      "step": 952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999982780556752e-05,
+      "loss": 5.8438,
+      "step": 953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9999818861885974e-05,
+      "loss": 5.5,
+      "step": 954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.999980969178349e-05,
+      "loss": 9.0625,
+      "step": 955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9999800295260184e-05,
+      "loss": 4.875,
+      "step": 956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9999790672316155e-05,
+      "loss": 6.9688,
+      "step": 957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9999780822951504e-05,
+      "loss": 9.9375,
+      "step": 958
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999977074716636e-05,
+      "loss": 6.8281,
+      "step": 959
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999976044496082e-05,
+      "loss": 2.3125,
+      "step": 960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999974991633501e-05,
+      "loss": 3.0,
+      "step": 961
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999973916128904e-05,
+      "loss": 7.125,
+      "step": 962
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999972817982305e-05,
+      "loss": 6.3125,
+      "step": 963
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999971697193715e-05,
+      "loss": 7.4062,
+      "step": 964
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999970553763147e-05,
+      "loss": 8.4531,
+      "step": 965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9999693876906136e-05,
+      "loss": 8.0469,
+      "step": 966
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999968198976129e-05,
+      "loss": 4.6406,
+      "step": 967
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999966987619705e-05,
+      "loss": 4.3438,
+      "step": 968
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9999657536213575e-05,
+      "loss": 7.8125,
+      "step": 969
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9999644969810986e-05,
+      "loss": 4.2188,
+      "step": 970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9999632176989435e-05,
+      "loss": 4.8125,
+      "step": 971
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999961915774906e-05,
+      "loss": 5.1875,
+      "step": 972
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999960591209002e-05,
+      "loss": 6.0938,
+      "step": 973
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999959244001245e-05,
+      "loss": 10.5625,
+      "step": 974
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999957874151651e-05,
+      "loss": 6.5,
+      "step": 975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9999564816602364e-05,
+      "loss": 4.4062,
+      "step": 976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999955066527015e-05,
+      "loss": 4.1719,
+      "step": 977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999953628752004e-05,
+      "loss": 6.7031,
+      "step": 978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9999521683352205e-05,
+      "loss": 5.2031,
+      "step": 979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9999506852766796e-05,
+      "loss": 7.2969,
+      "step": 980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999949179576398e-05,
+      "loss": 4.5625,
+      "step": 981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999947651234394e-05,
+      "loss": 4.6875,
+      "step": 982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999946100250684e-05,
+      "loss": 9.5625,
+      "step": 983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999944526625286e-05,
+      "loss": 8.3438,
+      "step": 984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999942930358217e-05,
+      "loss": 5.4375,
+      "step": 985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9999413114494966e-05,
+      "loss": 5.7812,
+      "step": 986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999939669899141e-05,
+      "loss": 4.3125,
+      "step": 987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.999938005707171e-05,
+      "loss": 5.1562,
+      "step": 988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9999363188736045e-05,
+      "loss": 4.4375,
+      "step": 989
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999934609398461e-05,
+      "loss": 3.7812,
+      "step": 990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999932877281758e-05,
+      "loss": 9.5469,
+      "step": 991
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999931122523517e-05,
+      "loss": 5.3594,
+      "step": 992
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9999293451237574e-05,
+      "loss": 5.5312,
+      "step": 993
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999927545082499e-05,
+      "loss": 6.6719,
+      "step": 994
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999925722399763e-05,
+      "loss": 4.8906,
+      "step": 995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.99992387707557e-05,
+      "loss": 7.4062,
+      "step": 996
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999922009109939e-05,
+      "loss": 5.5156,
+      "step": 997
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999920118502894e-05,
+      "loss": 7.375,
+      "step": 998
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9999182052544536e-05,
+      "loss": 8.1719,
+      "step": 999
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999916269364641e-05,
+      "loss": 5.2656,
+      "step": 1000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999914310833479e-05,
+      "loss": 4.9688,
+      "step": 1001
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999912329660988e-05,
+      "loss": 5.7188,
+      "step": 1002
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999910325847191e-05,
+      "loss": 4.1875,
+      "step": 1003
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9999082993921104e-05,
+      "loss": 4.8281,
+      "step": 1004
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9999062502957706e-05,
+      "loss": 5.625,
+      "step": 1005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9999041785581927e-05,
+      "loss": 9.1875,
+      "step": 1006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9999020841794024e-05,
+      "loss": 6.9219,
+      "step": 1007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999899967159421e-05,
+      "loss": 6.5312,
+      "step": 1008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999897827498274e-05,
+      "loss": 4.3438,
+      "step": 1009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999895665195985e-05,
+      "loss": 5.6875,
+      "step": 1010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.99989348025258e-05,
+      "loss": 4.1406,
+      "step": 1011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9998912726680814e-05,
+      "loss": 6.1094,
+      "step": 1012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9998890424425164e-05,
+      "loss": 6.75,
+      "step": 1013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.9998867895759084e-05,
+      "loss": 6.8125,
+      "step": 1014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999884514068283e-05,
+      "loss": 3.0,
+      "step": 1015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999882215919668e-05,
+      "loss": 8.8125,
+      "step": 1016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999879895130087e-05,
+      "loss": 3.0938,
+      "step": 1017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999877551699568e-05,
+      "loss": 5.1875,
+      "step": 1018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.999875185628136e-05,
+      "loss": 8.1094,
+      "step": 1019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999872796915819e-05,
+      "loss": 3.9688,
+      "step": 1020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999870385562644e-05,
+      "loss": 6.625,
+      "step": 1021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9998679515686374e-05,
+      "loss": 5.4688,
+      "step": 1022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9998654949338286e-05,
+      "loss": 5.9531,
+      "step": 1023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999863015658243e-05,
+      "loss": 6.3438,
+      "step": 1024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9998605137419095e-05,
+      "loss": 4.9531,
+      "step": 1025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999857989184858e-05,
+      "loss": 4.5312,
+      "step": 1026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999855441987115e-05,
+      "loss": 3.75,
+      "step": 1027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.99985287214871e-05,
+      "loss": 5.6406,
+      "step": 1028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999850279669673e-05,
+      "loss": 7.3125,
+      "step": 1029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9998476645500324e-05,
+      "loss": 4.4062,
+      "step": 1030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9998450267898174e-05,
+      "loss": 8.2344,
+      "step": 1031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999842366389059e-05,
+      "loss": 6.7188,
+      "step": 1032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999839683347786e-05,
+      "loss": 10.8594,
+      "step": 1033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9998369776660306e-05,
+      "loss": 7.4375,
+      "step": 1034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999834249343821e-05,
+      "loss": 6.25,
+      "step": 1035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9998314983811913e-05,
+      "loss": 5.6406,
+      "step": 1036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.99982872477817e-05,
+      "loss": 5.0,
+      "step": 1037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999825928534789e-05,
+      "loss": 5.9375,
+      "step": 1038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999823109651081e-05,
+      "loss": 3.25,
+      "step": 1039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999820268127076e-05,
+      "loss": 5.4688,
+      "step": 1040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999817403962808e-05,
+      "loss": 4.7812,
+      "step": 1041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9998145171583094e-05,
+      "loss": 5.0312,
+      "step": 1042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9998116077136115e-05,
+      "loss": 7.0,
+      "step": 1043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999808675628748e-05,
+      "loss": 3.4062,
+      "step": 1044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999805720903752e-05,
+      "loss": 4.375,
+      "step": 1045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999802743538658e-05,
+      "loss": 5.6094,
+      "step": 1046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999799743533498e-05,
+      "loss": 6.7812,
+      "step": 1047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9997967208883063e-05,
+      "loss": 6.0469,
+      "step": 1048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.999793675603119e-05,
+      "loss": 7.5,
+      "step": 1049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.9997906076779676e-05,
+      "loss": 8.2656,
+      "step": 1050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9997875171128896e-05,
+      "loss": 5.4688,
+      "step": 1051
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999784403907918e-05,
+      "loss": 4.0938,
+      "step": 1052
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999781268063089e-05,
+      "loss": 9.0938,
+      "step": 1053
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9997781095784376e-05,
+      "loss": 6.375,
+      "step": 1054
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9997749284539996e-05,
+      "loss": 5.2188,
+      "step": 1055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999771724689812e-05,
+      "loss": 6.9688,
+      "step": 1056
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.99976849828591e-05,
+      "loss": 13.5938,
+      "step": 1057
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999765249242331e-05,
+      "loss": 5.9297,
+      "step": 1058
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.99976197755911e-05,
+      "loss": 11.1562,
+      "step": 1059
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9997586832362865e-05,
+      "loss": 3.9688,
+      "step": 1060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999755366273896e-05,
+      "loss": 6.7344,
+      "step": 1061
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999752026671977e-05,
+      "loss": 3.625,
+      "step": 1062
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999748664430566e-05,
+      "loss": 6.3125,
+      "step": 1063
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999745279549703e-05,
+      "loss": 4.1094,
+      "step": 1064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9997418720294254e-05,
+      "loss": 8.7031,
+      "step": 1065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9997384418697704e-05,
+      "loss": 6.0938,
+      "step": 1066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9997349890707794e-05,
+      "loss": 5.8281,
+      "step": 1067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9997315136324896e-05,
+      "loss": 5.4062,
+      "step": 1068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999728015554941e-05,
+      "loss": 5.2969,
+      "step": 1069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999724494838174e-05,
+      "loss": 3.75,
+      "step": 1070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999720951482226e-05,
+      "loss": 3.875,
+      "step": 1071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.99971738548714e-05,
+      "loss": 3.4219,
+      "step": 1072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999713796852955e-05,
+      "loss": 4.75,
+      "step": 1073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9997101855797116e-05,
+      "loss": 4.0,
+      "step": 1074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999706551667451e-05,
+      "loss": 3.5,
+      "step": 1075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9997028951162136e-05,
+      "loss": 4.5625,
+      "step": 1076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999699215926042e-05,
+      "loss": 5.2031,
+      "step": 1077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999695514096976e-05,
+      "loss": 8.9531,
+      "step": 1078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.99969178962906e-05,
+      "loss": 7.9062,
+      "step": 1079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.999688042522334e-05,
+      "loss": 9.3906,
+      "step": 1080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999684272776842e-05,
+      "loss": 6.5469,
+      "step": 1081
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999680480392626e-05,
+      "loss": 4.6562,
+      "step": 1082
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999676665369729e-05,
+      "loss": 6.5938,
+      "step": 1083
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999672827708194e-05,
+      "loss": 4.5625,
+      "step": 1084
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9996689674080637e-05,
+      "loss": 4.7188,
+      "step": 1085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9996650844693835e-05,
+      "loss": 6.7656,
+      "step": 1086
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9996611788921967e-05,
+      "loss": 4.1875,
+      "step": 1087
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9996572506765466e-05,
+      "loss": 8.9062,
+      "step": 1088
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999653299822479e-05,
+      "loss": 4.2031,
+      "step": 1089
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999649326330038e-05,
+      "loss": 6.0312,
+      "step": 1090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9996453301992684e-05,
+      "loss": 5.6406,
+      "step": 1091
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9996413114302154e-05,
+      "loss": 5.6562,
+      "step": 1092
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999637270022925e-05,
+      "loss": 5.8594,
+      "step": 1093
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999633205977443e-05,
+      "loss": 4.3125,
+      "step": 1094
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999629119293815e-05,
+      "loss": 10.0,
+      "step": 1095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999625009972087e-05,
+      "loss": 5.1406,
+      "step": 1096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999620878012306e-05,
+      "loss": 4.7344,
+      "step": 1097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9996167234145194e-05,
+      "loss": 6.5938,
+      "step": 1098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999612546178773e-05,
+      "loss": 6.1562,
+      "step": 1099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999608346305114e-05,
+      "loss": 7.9688,
+      "step": 1100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999604123793591e-05,
+      "loss": 3.3125,
+      "step": 1101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999599878644252e-05,
+      "loss": 5.2188,
+      "step": 1102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9995956108571444e-05,
+      "loss": 2.4688,
+      "step": 1103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9995913204323154e-05,
+      "loss": 5.5,
+      "step": 1104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9995870073698154e-05,
+      "loss": 4.1875,
+      "step": 1105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999582671669693e-05,
+      "loss": 3.5,
+      "step": 1106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999578313331996e-05,
+      "loss": 3.1562,
+      "step": 1107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999573932356775e-05,
+      "loss": 4.6875,
+      "step": 1108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9995695287440794e-05,
+      "loss": 5.5,
+      "step": 1109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9995651024939584e-05,
+      "loss": 7.0938,
+      "step": 1110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9995606536064625e-05,
+      "loss": 2.9375,
+      "step": 1111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9995561820816424e-05,
+      "loss": 3.9062,
+      "step": 1112
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999551687919548e-05,
+      "loss": 9.1875,
+      "step": 1113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999547171120231e-05,
+      "loss": 5.5625,
+      "step": 1114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999542631683742e-05,
+      "loss": 5.0781,
+      "step": 1115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9995380696101324e-05,
+      "loss": 7.0312,
+      "step": 1116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999533484899454e-05,
+      "loss": 4.9688,
+      "step": 1117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999528877551759e-05,
+      "loss": 4.375,
+      "step": 1118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999524247567099e-05,
+      "loss": 7.375,
+      "step": 1119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999519594945526e-05,
+      "loss": 6.9531,
+      "step": 1120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999514919687094e-05,
+      "loss": 5.3594,
+      "step": 1121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9995102217918556e-05,
+      "loss": 5.0938,
+      "step": 1122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9995055012598635e-05,
+      "loss": 7.4375,
+      "step": 1123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999500758091171e-05,
+      "loss": 5.7344,
+      "step": 1124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999495992285832e-05,
+      "loss": 5.9844,
+      "step": 1125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9994912038439004e-05,
+      "loss": 3.1562,
+      "step": 1126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999486392765431e-05,
+      "loss": 4.375,
+      "step": 1127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999481559050478e-05,
+      "loss": 2.3438,
+      "step": 1128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999476702699095e-05,
+      "loss": 4.5625,
+      "step": 1129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999471823711339e-05,
+      "loss": 8.0938,
+      "step": 1130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999466922087264e-05,
+      "loss": 5.2812,
+      "step": 1131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999461997826925e-05,
+      "loss": 5.4375,
+      "step": 1132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9994570509303784e-05,
+      "loss": 7.125,
+      "step": 1133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999452081397681e-05,
+      "loss": 7.0625,
+      "step": 1134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999447089228887e-05,
+      "loss": 7.0312,
+      "step": 1135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999442074424055e-05,
+      "loss": 8.6094,
+      "step": 1136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999437036983241e-05,
+      "loss": 6.0469,
+      "step": 1137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999431976906502e-05,
+      "loss": 5.5625,
+      "step": 1138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999426894193895e-05,
+      "loss": 6.9375,
+      "step": 1139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.999421788845478e-05,
+      "loss": 5.5625,
+      "step": 1140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.9994166608613086e-05,
+      "loss": 3.9375,
+      "step": 1141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.999411510241445e-05,
+      "loss": 6.0625,
+      "step": 1142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9994063369859446e-05,
+      "loss": 6.3125,
+      "step": 1143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9994011410948673e-05,
+      "loss": 6.8594,
+      "step": 1144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.999395922568272e-05,
+      "loss": 4.125,
+      "step": 1145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9993906814062156e-05,
+      "loss": 5.5625,
+      "step": 1146
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9993854176087603e-05,
+      "loss": 4.7188,
+      "step": 1147
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.999380131175964e-05,
+      "loss": 4.75,
+      "step": 1148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9993748221078866e-05,
+      "loss": 6.6562,
+      "step": 1149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.999369490404589e-05,
+      "loss": 4.9375,
+      "step": 1150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9993641360661304e-05,
+      "loss": 7.6562,
+      "step": 1151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9993587590925726e-05,
+      "loss": 5.8438,
+      "step": 1152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.999353359483976e-05,
+      "loss": 9.2344,
+      "step": 1153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.999347937240402e-05,
+      "loss": 5.7969,
+      "step": 1154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.999342492361911e-05,
+      "loss": 3.9375,
+      "step": 1155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.999337024848566e-05,
+      "loss": 5.25,
+      "step": 1156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9993315347004275e-05,
+      "loss": 7.375,
+      "step": 1157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.999326021917559e-05,
+      "loss": 4.0938,
+      "step": 1158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9993204865000216e-05,
+      "loss": 6.7812,
+      "step": 1159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9993149284478796e-05,
+      "loss": 6.5156,
+      "step": 1160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9993093477611944e-05,
+      "loss": 5.6875,
+      "step": 1161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.999303744440029e-05,
+      "loss": 7.25,
+      "step": 1162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9992981184844486e-05,
+      "loss": 5.6719,
+      "step": 1163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.999292469894516e-05,
+      "loss": 4.7344,
+      "step": 1164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.999286798670295e-05,
+      "loss": 5.7188,
+      "step": 1165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.99928110481185e-05,
+      "loss": 5.4062,
+      "step": 1166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9992753883192444e-05,
+      "loss": 5.9375,
+      "step": 1167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9992696491925446e-05,
+      "loss": 8.4688,
+      "step": 1168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9992638874318146e-05,
+      "loss": 3.0781,
+      "step": 1169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9992581030371195e-05,
+      "loss": 9.625,
+      "step": 1170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.9992522960085256e-05,
+      "loss": 5.0469,
+      "step": 1171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999246466346098e-05,
+      "loss": 4.3438,
+      "step": 1172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999240614049903e-05,
+      "loss": 3.9844,
+      "step": 1173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999234739120007e-05,
+      "loss": 6.2031,
+      "step": 1174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999228841556475e-05,
+      "loss": 5.3281,
+      "step": 1175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999222921359376e-05,
+      "loss": 5.4375,
+      "step": 1176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9992169785287764e-05,
+      "loss": 7.0625,
+      "step": 1177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9992110130647426e-05,
+      "loss": 6.2031,
+      "step": 1178
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999205024967342e-05,
+      "loss": 5.1562,
+      "step": 1179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999199014236644e-05,
+      "loss": 6.4219,
+      "step": 1180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999192980872715e-05,
+      "loss": 4.3438,
+      "step": 1181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999186924875624e-05,
+      "loss": 4.6406,
+      "step": 1182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9991808462454396e-05,
+      "loss": 4.2969,
+      "step": 1183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999174744982231e-05,
+      "loss": 3.375,
+      "step": 1184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999168621086067e-05,
+      "loss": 5.5156,
+      "step": 1185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9991624745570154e-05,
+      "loss": 6.2188,
+      "step": 1186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9991563053951476e-05,
+      "loss": 4.8438,
+      "step": 1187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999150113600534e-05,
+      "loss": 9.25,
+      "step": 1188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999143899173243e-05,
+      "loss": 6.2188,
+      "step": 1189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9991376621133454e-05,
+      "loss": 6.3438,
+      "step": 1190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999131402420912e-05,
+      "loss": 6.25,
+      "step": 1191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999125120096015e-05,
+      "loss": 9.9219,
+      "step": 1192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999118815138723e-05,
+      "loss": 7.1875,
+      "step": 1193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999112487549109e-05,
+      "loss": 8.1875,
+      "step": 1194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9991061373272435e-05,
+      "loss": 9.9375,
+      "step": 1195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.9990997644732e-05,
+      "loss": 8.4062,
+      "step": 1196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.99909336898705e-05,
+      "loss": 5.9062,
+      "step": 1197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999086950868865e-05,
+      "loss": 6.1094,
+      "step": 1198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999080510118718e-05,
+      "loss": 5.625,
+      "step": 1199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999074046736683e-05,
+      "loss": 6.3438,
+      "step": 1200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.999067560722833e-05,
+      "loss": 7.0781,
+      "step": 1201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.99906105207724e-05,
+      "loss": 8.0469,
+      "step": 1202
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.999054520799979e-05,
+      "loss": 5.125,
+      "step": 1203
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.999047966891123e-05,
+      "loss": 8.25,
+      "step": 1204
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.999041390350747e-05,
+      "loss": 10.4062,
+      "step": 1205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.999034791178925e-05,
+      "loss": 4.875,
+      "step": 1206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.999028169375732e-05,
+      "loss": 10.2812,
+      "step": 1207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9990215249412426e-05,
+      "loss": 4.7969,
+      "step": 1208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9990148578755326e-05,
+      "loss": 4.9531,
+      "step": 1209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.999008168178678e-05,
+      "loss": 6.9688,
+      "step": 1210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.999001455850752e-05,
+      "loss": 6.9062,
+      "step": 1211
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.998994720891832e-05,
+      "loss": 6.5469,
+      "step": 1212
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.998987963301996e-05,
+      "loss": 5.4688,
+      "step": 1213
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9989811830813186e-05,
+      "loss": 5.8594,
+      "step": 1214
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.998974380229877e-05,
+      "loss": 7.0,
+      "step": 1215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.998967554747748e-05,
+      "loss": 5.0625,
+      "step": 1216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.998960706635009e-05,
+      "loss": 4.8438,
+      "step": 1217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9989538358917375e-05,
+      "loss": 6.5938,
+      "step": 1218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9989469425180117e-05,
+      "loss": 5.0312,
+      "step": 1219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.998940026513909e-05,
+      "loss": 6.1875,
+      "step": 1220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.998933087879508e-05,
+      "loss": 7.0,
+      "step": 1221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.998926126614887e-05,
+      "loss": 5.9688,
+      "step": 1222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.998919142720126e-05,
+      "loss": 4.0938,
+      "step": 1223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9989121361953025e-05,
+      "loss": 3.4375,
+      "step": 1224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9989051070404964e-05,
+      "loss": 8.3438,
+      "step": 1225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9988980552557875e-05,
+      "loss": 4.7812,
+      "step": 1226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.998890980841255e-05,
+      "loss": 7.1719,
+      "step": 1227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.99888388379698e-05,
+      "loss": 3.3281,
+      "step": 1228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9988767641230425e-05,
+      "loss": 7.0,
+      "step": 1229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.998869621819523e-05,
+      "loss": 3.8125,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9988624568865014e-05,
+      "loss": 3.7969,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.99885526932406e-05,
+      "loss": 3.8438,
+      "step": 1232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9988480591322805e-05,
+      "loss": 7.0312,
+      "step": 1233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9988408263112435e-05,
+      "loss": 3.8438,
+      "step": 1234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998833570861031e-05,
+      "loss": 4.2188,
+      "step": 1235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998826292781726e-05,
+      "loss": 7.0938,
+      "step": 1236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.99881899207341e-05,
+      "loss": 8.875,
+      "step": 1237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9988116687361654e-05,
+      "loss": 10.0312,
+      "step": 1238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998804322770076e-05,
+      "loss": 9.6875,
+      "step": 1239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998796954175225e-05,
+      "loss": 4.25,
+      "step": 1240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998789562951695e-05,
+      "loss": 4.4844,
+      "step": 1241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998782149099571e-05,
+      "loss": 2.9688,
+      "step": 1242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998774712618935e-05,
+      "loss": 7.9531,
+      "step": 1243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998767253509872e-05,
+      "loss": 5.9531,
+      "step": 1244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9987597717724676e-05,
+      "loss": 5.5312,
+      "step": 1245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998752267406805e-05,
+      "loss": 5.0312,
+      "step": 1246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9987447404129703e-05,
+      "loss": 4.5312,
+      "step": 1247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998737190791048e-05,
+      "loss": 3.6562,
+      "step": 1248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998729618541124e-05,
+      "loss": 4.9062,
+      "step": 1249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9987220236632833e-05,
+      "loss": 3.0781,
+      "step": 1250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9987144061576126e-05,
+      "loss": 4.375,
+      "step": 1251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9987067660241974e-05,
+      "loss": 5.3125,
+      "step": 1252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998699103263126e-05,
+      "loss": 6.8594,
+      "step": 1253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998691417874482e-05,
+      "loss": 3.625,
+      "step": 1254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998683709858355e-05,
+      "loss": 6.9375,
+      "step": 1255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9986759792148325e-05,
+      "loss": 3.1875,
+      "step": 1256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9986682259440004e-05,
+      "loss": 5.3438,
+      "step": 1257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998660450045947e-05,
+      "loss": 5.7031,
+      "step": 1258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.99865265152076e-05,
+      "loss": 7.3281,
+      "step": 1259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9986448303685285e-05,
+      "loss": 4.5938,
+      "step": 1260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.9986369865893405e-05,
+      "loss": 3.7656,
+      "step": 1261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998629120183285e-05,
+      "loss": 3.3906,
+      "step": 1262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.998621231150451e-05,
+      "loss": 5.1094,
+      "step": 1263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998613319490928e-05,
+      "loss": 8.5,
+      "step": 1264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998605385204806e-05,
+      "loss": 4.0156,
+      "step": 1265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998597428292173e-05,
+      "loss": 6.4375,
+      "step": 1266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998589448753121e-05,
+      "loss": 5.6406,
+      "step": 1267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9985814465877395e-05,
+      "loss": 3.75,
+      "step": 1268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998573421796119e-05,
+      "loss": 5.6875,
+      "step": 1269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998565374378351e-05,
+      "loss": 2.5312,
+      "step": 1270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998557304334527e-05,
+      "loss": 4.25,
+      "step": 1271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9985492116647364e-05,
+      "loss": 5.1094,
+      "step": 1272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998541096369072e-05,
+      "loss": 5.3281,
+      "step": 1273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998532958447626e-05,
+      "loss": 5.25,
+      "step": 1274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9985247979004895e-05,
+      "loss": 4.5938,
+      "step": 1275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998516614727757e-05,
+      "loss": 7.0938,
+      "step": 1276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998508408929518e-05,
+      "loss": 4.8594,
+      "step": 1277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9985001805058684e-05,
+      "loss": 8.125,
+      "step": 1278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9984919294569e-05,
+      "loss": 3.4531,
+      "step": 1279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998483655782706e-05,
+      "loss": 6.3438,
+      "step": 1280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.99847535948338e-05,
+      "loss": 5.2969,
+      "step": 1281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998467040559017e-05,
+      "loss": 7.875,
+      "step": 1282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9984586990097095e-05,
+      "loss": 3.8438,
+      "step": 1283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998450334835553e-05,
+      "loss": 6.1562,
+      "step": 1284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998441948036643e-05,
+      "loss": 8.0938,
+      "step": 1285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998433538613073e-05,
+      "loss": 4.6875,
+      "step": 1286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9984251065649385e-05,
+      "loss": 3.5469,
+      "step": 1287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998416651892336e-05,
+      "loss": 5.125,
+      "step": 1288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9984081745953596e-05,
+      "loss": 6.6094,
+      "step": 1289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9983996746741066e-05,
+      "loss": 3.0,
+      "step": 1290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998391152128673e-05,
+      "loss": 3.9688,
+      "step": 1291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.9983826069591535e-05,
+      "loss": 6.2188,
+      "step": 1292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.998374039165647e-05,
+      "loss": 4.125,
+      "step": 1293
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998365448748251e-05,
+      "loss": 4.5,
+      "step": 1294
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.99835683570706e-05,
+      "loss": 5.8281,
+      "step": 1295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998348200042174e-05,
+      "loss": 3.0312,
+      "step": 1296
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9983395417536896e-05,
+      "loss": 5.6562,
+      "step": 1297
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998330860841705e-05,
+      "loss": 4.7188,
+      "step": 1298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998322157306319e-05,
+      "loss": 8.0625,
+      "step": 1299
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9983134311476295e-05,
+      "loss": 5.5781,
+      "step": 1300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998304682365735e-05,
+      "loss": 7.7031,
+      "step": 1301
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998295910960736e-05,
+      "loss": 9.375,
+      "step": 1302
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.99828711693273e-05,
+      "loss": 4.4062,
+      "step": 1303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998278300281817e-05,
+      "loss": 9.5938,
+      "step": 1304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9982694610080985e-05,
+      "loss": 5.7188,
+      "step": 1305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998260599111673e-05,
+      "loss": 2.9531,
+      "step": 1306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.99825171459264e-05,
+      "loss": 9.0,
+      "step": 1307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9982428074511024e-05,
+      "loss": 4.1562,
+      "step": 1308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.99823387768716e-05,
+      "loss": 6.5,
+      "step": 1309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998224925300912e-05,
+      "loss": 5.0312,
+      "step": 1310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9982159502924626e-05,
+      "loss": 5.2188,
+      "step": 1311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9982069526619126e-05,
+      "loss": 4.375,
+      "step": 1312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998197932409364e-05,
+      "loss": 3.25,
+      "step": 1313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998188889534917e-05,
+      "loss": 6.875,
+      "step": 1314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998179824038676e-05,
+      "loss": 3.75,
+      "step": 1315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998170735920743e-05,
+      "loss": 3.5,
+      "step": 1316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9981616251812216e-05,
+      "loss": 5.2188,
+      "step": 1317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9981524918202135e-05,
+      "loss": 4.6562,
+      "step": 1318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998143335837824e-05,
+      "loss": 7.0,
+      "step": 1319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9981341572341547e-05,
+      "loss": 3.9375,
+      "step": 1320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.99812495600931e-05,
+      "loss": 9.125,
+      "step": 1321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998115732163395e-05,
+      "loss": 6.0,
+      "step": 1322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.9981064856965144e-05,
+      "loss": 3.4062,
+      "step": 1323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.998097216608772e-05,
+      "loss": 5.5781,
+      "step": 1324
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.998087924900272e-05,
+      "loss": 4.7188,
+      "step": 1325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.998078610571121e-05,
+      "loss": 2.2969,
+      "step": 1326
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9980692736214244e-05,
+      "loss": 7.6719,
+      "step": 1327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.998059914051287e-05,
+      "loss": 5.875,
+      "step": 1328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.998050531860815e-05,
+      "loss": 7.2812,
+      "step": 1329
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.998041127050115e-05,
+      "loss": 5.5625,
+      "step": 1330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.998031699619293e-05,
+      "loss": 7.625,
+      "step": 1331
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.998022249568457e-05,
+      "loss": 5.0781,
+      "step": 1332
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.998012776897712e-05,
+      "loss": 5.375,
+      "step": 1333
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.998003281607166e-05,
+      "loss": 6.875,
+      "step": 1334
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.997993763696927e-05,
+      "loss": 2.5156,
+      "step": 1335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9979842231671034e-05,
+      "loss": 10.3906,
+      "step": 1336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9979746600178014e-05,
+      "loss": 6.4844,
+      "step": 1337
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.99796507424913e-05,
+      "loss": 6.7812,
+      "step": 1338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9979554658611986e-05,
+      "loss": 4.8125,
+      "step": 1339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9979458348541146e-05,
+      "loss": 4.4688,
+      "step": 1340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.997936181227989e-05,
+      "loss": 4.1875,
+      "step": 1341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9979265049829284e-05,
+      "loss": 2.7344,
+      "step": 1342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.997916806119044e-05,
+      "loss": 3.0938,
+      "step": 1343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.997907084636445e-05,
+      "loss": 8.1562,
+      "step": 1344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.997897340535243e-05,
+      "loss": 6.3438,
+      "step": 1345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9978875738155464e-05,
+      "loss": 4.7812,
+      "step": 1346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9978777844774664e-05,
+      "loss": 5.2188,
+      "step": 1347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.997867972521114e-05,
+      "loss": 4.0938,
+      "step": 1348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9978581379466e-05,
+      "loss": 6.8125,
+      "step": 1349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.997848280754036e-05,
+      "loss": 3.8125,
+      "step": 1350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.9978384009435343e-05,
+      "loss": 7.7656,
+      "step": 1351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.997828498515205e-05,
+      "loss": 5.1875,
+      "step": 1352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.997818573469161e-05,
+      "loss": 8.2031,
+      "step": 1353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.997808625805516e-05,
+      "loss": 3.4062,
+      "step": 1354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.99779865552438e-05,
+      "loss": 4.375,
+      "step": 1355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997788662625869e-05,
+      "loss": 4.5156,
+      "step": 1356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997778647110093e-05,
+      "loss": 4.5938,
+      "step": 1357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997768608977168e-05,
+      "loss": 3.625,
+      "step": 1358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997758548227206e-05,
+      "loss": 4.4062,
+      "step": 1359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997748464860321e-05,
+      "loss": 1.8438,
+      "step": 1360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997738358876628e-05,
+      "loss": 4.0312,
+      "step": 1361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997728230276241e-05,
+      "loss": 4.9844,
+      "step": 1362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997718079059275e-05,
+      "loss": 4.75,
+      "step": 1363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997707905225844e-05,
+      "loss": 3.7188,
+      "step": 1364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9976977087760634e-05,
+      "loss": 4.6406,
+      "step": 1365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.99768748971005e-05,
+      "loss": 4.8125,
+      "step": 1366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997677248027918e-05,
+      "loss": 2.4062,
+      "step": 1367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997666983729783e-05,
+      "loss": 7.2812,
+      "step": 1368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997656696815763e-05,
+      "loss": 4.9844,
+      "step": 1369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997646387285973e-05,
+      "loss": 5.3125,
+      "step": 1370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9976360551405306e-05,
+      "loss": 2.1406,
+      "step": 1371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997625700379552e-05,
+      "loss": 3.4219,
+      "step": 1372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9976153230031546e-05,
+      "loss": 8.2031,
+      "step": 1373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997604923011457e-05,
+      "loss": 6.125,
+      "step": 1374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9975945004045756e-05,
+      "loss": 4.5,
+      "step": 1375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997584055182628e-05,
+      "loss": 3.8125,
+      "step": 1376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997573587345734e-05,
+      "loss": 7.2656,
+      "step": 1377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997563096894011e-05,
+      "loss": 9.125,
+      "step": 1378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9975525838275784e-05,
+      "loss": 5.4844,
+      "step": 1379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9975420481465546e-05,
+      "loss": 6.6875,
+      "step": 1380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.99753148985106e-05,
+      "loss": 5.5938,
+      "step": 1381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997520908941213e-05,
+      "loss": 6.1562,
+      "step": 1382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997510305417134e-05,
+      "loss": 4.2812,
+      "step": 1383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.997499679278942e-05,
+      "loss": 6.2656,
+      "step": 1384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.9974890305267586e-05,
+      "loss": 4.7031,
+      "step": 1385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997478359160704e-05,
+      "loss": 5.125,
+      "step": 1386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9974676651808986e-05,
+      "loss": 2.5781,
+      "step": 1387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9974569485874636e-05,
+      "loss": 4.2969,
+      "step": 1388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9974462093805203e-05,
+      "loss": 6.5625,
+      "step": 1389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9974354475601914e-05,
+      "loss": 5.4375,
+      "step": 1390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9974246631265975e-05,
+      "loss": 5.6875,
+      "step": 1391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.99741385607986e-05,
+      "loss": 6.5469,
+      "step": 1392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997403026420103e-05,
+      "loss": 4.7812,
+      "step": 1393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997392174147448e-05,
+      "loss": 4.7188,
+      "step": 1394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997381299262018e-05,
+      "loss": 6.7656,
+      "step": 1395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997370401763937e-05,
+      "loss": 8.0625,
+      "step": 1396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997359481653327e-05,
+      "loss": 5.4219,
+      "step": 1397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997348538930313e-05,
+      "loss": 4.5,
+      "step": 1398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997337573595018e-05,
+      "loss": 5.5625,
+      "step": 1399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9973265856475653e-05,
+      "loss": 4.9688,
+      "step": 1400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997315575088081e-05,
+      "loss": 5.625,
+      "step": 1401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997304541916689e-05,
+      "loss": 7.4062,
+      "step": 1402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997293486133515e-05,
+      "loss": 5.0,
+      "step": 1403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9972824077386824e-05,
+      "loss": 6.8594,
+      "step": 1404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997271306732319e-05,
+      "loss": 5.375,
+      "step": 1405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9972601831145476e-05,
+      "loss": 6.4375,
+      "step": 1406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997249036885496e-05,
+      "loss": 4.2188,
+      "step": 1407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997237868045291e-05,
+      "loss": 4.4688,
+      "step": 1408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997226676594057e-05,
+      "loss": 6.2969,
+      "step": 1409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9972154625319225e-05,
+      "loss": 6.6562,
+      "step": 1410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997204225859014e-05,
+      "loss": 4.1719,
+      "step": 1411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9971929665754576e-05,
+      "loss": 5.125,
+      "step": 1412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9971816846813816e-05,
+      "loss": 5.6562,
+      "step": 1413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.9971703801769143e-05,
+      "loss": 10.0938,
+      "step": 1414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.997159053062183e-05,
+      "loss": 3.25,
+      "step": 1415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.997147703337316e-05,
+      "loss": 7.3125,
+      "step": 1416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.997136331002442e-05,
+      "loss": 4.125,
+      "step": 1417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.99712493605769e-05,
+      "loss": 4.9062,
+      "step": 1418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.997113518503188e-05,
+      "loss": 4.4688,
+      "step": 1419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9971020783390654e-05,
+      "loss": 7.2344,
+      "step": 1420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.997090615565453e-05,
+      "loss": 5.8125,
+      "step": 1421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.99707913018248e-05,
+      "loss": 6.4531,
+      "step": 1422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.997067622190276e-05,
+      "loss": 5.7188,
+      "step": 1423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.997056091588971e-05,
+      "loss": 6.6562,
+      "step": 1424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9970445383786965e-05,
+      "loss": 2.9375,
+      "step": 1425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9970329625595826e-05,
+      "loss": 4.9844,
+      "step": 1426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.99702136413176e-05,
+      "loss": 6.3438,
+      "step": 1427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9970097430953617e-05,
+      "loss": 4.1875,
+      "step": 1428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.996998099450517e-05,
+      "loss": 7.5156,
+      "step": 1429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.99698643319736e-05,
+      "loss": 4.9062,
+      "step": 1430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9969747443360206e-05,
+      "loss": 5.3125,
+      "step": 1431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.996963032866633e-05,
+      "loss": 2.8438,
+      "step": 1432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.996951298789328e-05,
+      "loss": 4.7812,
+      "step": 1433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.99693954210424e-05,
+      "loss": 7.75,
+      "step": 1434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.996927762811502e-05,
+      "loss": 3.875,
+      "step": 1435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.996915960911246e-05,
+      "loss": 4.375,
+      "step": 1436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.996904136403607e-05,
+      "loss": 4.5312,
+      "step": 1437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9968922892887174e-05,
+      "loss": 8.3438,
+      "step": 1438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.996880419566713e-05,
+      "loss": 5.1562,
+      "step": 1439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9968685272377276e-05,
+      "loss": 5.3438,
+      "step": 1440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9968566123018955e-05,
+      "loss": 6.3438,
+      "step": 1441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9968446747593516e-05,
+      "loss": 4.8438,
+      "step": 1442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.996832714610231e-05,
+      "loss": 4.2656,
+      "step": 1443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.99682073185467e-05,
+      "loss": 3.9688,
+      "step": 1444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.9968087264928025e-05,
+      "loss": 4.375,
+      "step": 1445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9967966985247665e-05,
+      "loss": 4.5781,
+      "step": 1446
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996784647950697e-05,
+      "loss": 7.6562,
+      "step": 1447
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.99677257477073e-05,
+      "loss": 4.3281,
+      "step": 1448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9967604789850035e-05,
+      "loss": 6.2188,
+      "step": 1449
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9967483605936525e-05,
+      "loss": 5.625,
+      "step": 1450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996736219596816e-05,
+      "loss": 2.375,
+      "step": 1451
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996724055994632e-05,
+      "loss": 6.4062,
+      "step": 1452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996711869787235e-05,
+      "loss": 8.25,
+      "step": 1453
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9966996609747666e-05,
+      "loss": 5.0625,
+      "step": 1454
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9966874295573625e-05,
+      "loss": 5.6875,
+      "step": 1455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996675175535162e-05,
+      "loss": 3.5312,
+      "step": 1456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996662898908305e-05,
+      "loss": 5.0625,
+      "step": 1457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996650599676929e-05,
+      "loss": 4.7188,
+      "step": 1458
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9966382778411727e-05,
+      "loss": 7.7188,
+      "step": 1459
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9966259334011774e-05,
+      "loss": 2.5,
+      "step": 1460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996613566357082e-05,
+      "loss": 4.5469,
+      "step": 1461
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996601176709026e-05,
+      "loss": 6.0,
+      "step": 1462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996588764457151e-05,
+      "loss": 5.1875,
+      "step": 1463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996576329601596e-05,
+      "loss": 2.75,
+      "step": 1464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996563872142503e-05,
+      "loss": 7.4844,
+      "step": 1465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9965513920800116e-05,
+      "loss": 1.7188,
+      "step": 1466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996538889414265e-05,
+      "loss": 4.1875,
+      "step": 1467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996526364145403e-05,
+      "loss": 9.2031,
+      "step": 1468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996513816273569e-05,
+      "loss": 5.7656,
+      "step": 1469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9965012457989035e-05,
+      "loss": 4.5312,
+      "step": 1470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9964886527215494e-05,
+      "loss": 5.875,
+      "step": 1471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.99647603704165e-05,
+      "loss": 5.625,
+      "step": 1472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996463398759347e-05,
+      "loss": 5.1875,
+      "step": 1473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996450737874784e-05,
+      "loss": 7.7344,
+      "step": 1474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9964380543881046e-05,
+      "loss": 5.0938,
+      "step": 1475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.996425348299452e-05,
+      "loss": 4.4219,
+      "step": 1476
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.99641261960897e-05,
+      "loss": 5.9688,
+      "step": 1477
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9963998683168034e-05,
+      "loss": 6.0469,
+      "step": 1478
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.996387094423095e-05,
+      "loss": 5.4062,
+      "step": 1479
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9963742979279917e-05,
+      "loss": 4.9375,
+      "step": 1480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.996361478831636e-05,
+      "loss": 5.3125,
+      "step": 1481
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9963486371341744e-05,
+      "loss": 3.9375,
+      "step": 1482
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9963357728357526e-05,
+      "loss": 4.3125,
+      "step": 1483
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.996322885936515e-05,
+      "loss": 4.1875,
+      "step": 1484
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.996309976436608e-05,
+      "loss": 5.4688,
+      "step": 1485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9962970443361786e-05,
+      "loss": 3.1875,
+      "step": 1486
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.996284089635372e-05,
+      "loss": 3.4062,
+      "step": 1487
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.996271112334336e-05,
+      "loss": 7.5,
+      "step": 1488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.996258112433216e-05,
+      "loss": 4.3906,
+      "step": 1489
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9962450899321605e-05,
+      "loss": 3.875,
+      "step": 1490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.996232044831317e-05,
+      "loss": 5.9688,
+      "step": 1491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9962189771308315e-05,
+      "loss": 4.4375,
+      "step": 1492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9962058868308545e-05,
+      "loss": 3.4375,
+      "step": 1493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9961927739315315e-05,
+      "loss": 5.6094,
+      "step": 1494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9961796384330124e-05,
+      "loss": 5.0156,
+      "step": 1495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.996166480335446e-05,
+      "loss": 3.8594,
+      "step": 1496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.996153299638982e-05,
+      "loss": 5.0938,
+      "step": 1497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9961400963437676e-05,
+      "loss": 6.0,
+      "step": 1498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9961268704499534e-05,
+      "loss": 2.8125,
+      "step": 1499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.99611362195769e-05,
+      "loss": 4.0938,
+      "step": 1500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.996100350867125e-05,
+      "loss": 4.5,
+      "step": 1501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.996087057178411e-05,
+      "loss": 5.9062,
+      "step": 1502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.996073740891698e-05,
+      "loss": 4.375,
+      "step": 1503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9960604020071356e-05,
+      "loss": 7.0312,
+      "step": 1504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.9960470405248754e-05,
+      "loss": 3.8438,
+      "step": 1505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.99603365644507e-05,
+      "loss": 4.7344,
+      "step": 1506
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.996020249767868e-05,
+      "loss": 3.2188,
+      "step": 1507
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9960068204934244e-05,
+      "loss": 5.5781,
+      "step": 1508
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995993368621889e-05,
+      "loss": 2.6719,
+      "step": 1509
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9959798941534154e-05,
+      "loss": 5.8125,
+      "step": 1510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995966397088155e-05,
+      "loss": 7.5469,
+      "step": 1511
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995952877426262e-05,
+      "loss": 6.1406,
+      "step": 1512
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995939335167888e-05,
+      "loss": 5.2969,
+      "step": 1513
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995925770313187e-05,
+      "loss": 5.1719,
+      "step": 1514
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9959121828623126e-05,
+      "loss": 2.8281,
+      "step": 1515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995898572815419e-05,
+      "loss": 6.0312,
+      "step": 1516
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9958849401726593e-05,
+      "loss": 4.7969,
+      "step": 1517
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9958712849341885e-05,
+      "loss": 3.9688,
+      "step": 1518
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995857607100162e-05,
+      "loss": 3.2188,
+      "step": 1519
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995843906670733e-05,
+      "loss": 4.0,
+      "step": 1520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995830183646057e-05,
+      "loss": 6.125,
+      "step": 1521
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.99581643802629e-05,
+      "loss": 7.6562,
+      "step": 1522
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995802669811587e-05,
+      "loss": 4.875,
+      "step": 1523
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9957888790021056e-05,
+      "loss": 5.6875,
+      "step": 1524
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995775065597999e-05,
+      "loss": 3.1719,
+      "step": 1525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995761229599425e-05,
+      "loss": 7.6406,
+      "step": 1526
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9957473710065413e-05,
+      "loss": 5.6406,
+      "step": 1527
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995733489819504e-05,
+      "loss": 4.9062,
+      "step": 1528
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995719586038469e-05,
+      "loss": 5.7656,
+      "step": 1529
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995705659663595e-05,
+      "loss": 5.1875,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.99569171069504e-05,
+      "loss": 7.3438,
+      "step": 1531
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9956777391329615e-05,
+      "loss": 6.6875,
+      "step": 1532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995663744977517e-05,
+      "loss": 5.4219,
+      "step": 1533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995649728228866e-05,
+      "loss": 6.0938,
+      "step": 1534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995635688887166e-05,
+      "loss": 4.5781,
+      "step": 1535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.9956216269525775e-05,
+      "loss": 5.6562,
+      "step": 1536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.995607542425258e-05,
+      "loss": 4.8281,
+      "step": 1537
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.995593435305368e-05,
+      "loss": 7.4062,
+      "step": 1538
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.995579305593067e-05,
+      "loss": 2.5625,
+      "step": 1539
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.995565153288515e-05,
+      "loss": 5.5,
+      "step": 1540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9955509783918725e-05,
+      "loss": 4.4375,
+      "step": 1541
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9955367809032986e-05,
+      "loss": 3.1562,
+      "step": 1542
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.995522560822956e-05,
+      "loss": 7.6094,
+      "step": 1543
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9955083181510046e-05,
+      "loss": 4.4688,
+      "step": 1544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.995494052887606e-05,
+      "loss": 5.0,
+      "step": 1545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.995479765032921e-05,
+      "loss": 2.0625,
+      "step": 1546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9954654545871114e-05,
+      "loss": 4.25,
+      "step": 1547
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.99545112155034e-05,
+      "loss": 8.0312,
+      "step": 1548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9954367659227685e-05,
+      "loss": 4.5938,
+      "step": 1549
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.995422387704561e-05,
+      "loss": 4.9844,
+      "step": 1550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9954079868958775e-05,
+      "loss": 5.5469,
+      "step": 1551
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.995393563496883e-05,
+      "loss": 6.6094,
+      "step": 1552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9953791175077406e-05,
+      "loss": 4.0156,
+      "step": 1553
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.995364648928613e-05,
+      "loss": 4.9531,
+      "step": 1554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9953501577596645e-05,
+      "loss": 4.9844,
+      "step": 1555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9953356440010586e-05,
+      "loss": 6.2812,
+      "step": 1556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9953211076529616e-05,
+      "loss": 2.7031,
+      "step": 1557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.995306548715535e-05,
+      "loss": 8.3125,
+      "step": 1558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9952919671889465e-05,
+      "loss": 4.0625,
+      "step": 1559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.995277363073359e-05,
+      "loss": 4.1094,
+      "step": 1560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.99526273636894e-05,
+      "loss": 8.7578,
+      "step": 1561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.995248087075853e-05,
+      "loss": 4.5938,
+      "step": 1562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.995233415194265e-05,
+      "loss": 2.5,
+      "step": 1563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9952187207243414e-05,
+      "loss": 2.5469,
+      "step": 1564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.995204003666249e-05,
+      "loss": 4.1094,
+      "step": 1565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9951892640201546e-05,
+      "loss": 7.875,
+      "step": 1566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.9951745017862247e-05,
+      "loss": 6.0,
+      "step": 1567
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.995159716964627e-05,
+      "loss": 6.0156,
+      "step": 1568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.995144909555528e-05,
+      "loss": 4.5625,
+      "step": 1569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.995130079559096e-05,
+      "loss": 3.3594,
+      "step": 1570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.995115226975498e-05,
+      "loss": 6.5469,
+      "step": 1571
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.995100351804904e-05,
+      "loss": 3.0938,
+      "step": 1572
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.99508545404748e-05,
+      "loss": 5.0,
+      "step": 1573
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.995070533703397e-05,
+      "loss": 7.25,
+      "step": 1574
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9950555907728225e-05,
+      "loss": 4.375,
+      "step": 1575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.995040625255926e-05,
+      "loss": 5.5312,
+      "step": 1576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9950256371528765e-05,
+      "loss": 5.25,
+      "step": 1577
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.995010626463845e-05,
+      "loss": 5.4219,
+      "step": 1578
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.994995593189e-05,
+      "loss": 6.3438,
+      "step": 1579
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.994980537328512e-05,
+      "loss": 4.5312,
+      "step": 1580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.994965458882551e-05,
+      "loss": 6.7812,
+      "step": 1581
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.99495035785129e-05,
+      "loss": 8.1562,
+      "step": 1582
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9949352342348975e-05,
+      "loss": 6.625,
+      "step": 1583
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.994920088033546e-05,
+      "loss": 8.25,
+      "step": 1584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9949049192474055e-05,
+      "loss": 5.125,
+      "step": 1585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9948897278766496e-05,
+      "loss": 6.875,
+      "step": 1586
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.994874513921449e-05,
+      "loss": 10.0,
+      "step": 1587
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9948592773819774e-05,
+      "loss": 8.25,
+      "step": 1588
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.994844018258405e-05,
+      "loss": 3.7188,
+      "step": 1589
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9948287365509065e-05,
+      "loss": 5.5625,
+      "step": 1590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.994813432259654e-05,
+      "loss": 4.875,
+      "step": 1591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.994798105384821e-05,
+      "loss": 7.1562,
+      "step": 1592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9947827559265814e-05,
+      "loss": 5.7188,
+      "step": 1593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.994767383885108e-05,
+      "loss": 5.0312,
+      "step": 1594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.994751989260576e-05,
+      "loss": 12.8438,
+      "step": 1595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.9947365720531586e-05,
+      "loss": 2.9688,
+      "step": 1596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.994721132263031e-05,
+      "loss": 3.7188,
+      "step": 1597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.994705669890368e-05,
+      "loss": 6.8125,
+      "step": 1598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994690184935344e-05,
+      "loss": 3.1562,
+      "step": 1599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994674677398135e-05,
+      "loss": 4.2812,
+      "step": 1600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994659147278917e-05,
+      "loss": 5.4844,
+      "step": 1601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994643594577864e-05,
+      "loss": 4.7188,
+      "step": 1602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994628019295154e-05,
+      "loss": 5.0,
+      "step": 1603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994612421430963e-05,
+      "loss": 3.7188,
+      "step": 1604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9945968009854665e-05,
+      "loss": 5.5312,
+      "step": 1605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994581157958842e-05,
+      "loss": 7.7969,
+      "step": 1606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994565492351267e-05,
+      "loss": 7.5,
+      "step": 1607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994549804162918e-05,
+      "loss": 6.0,
+      "step": 1608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9945340933939735e-05,
+      "loss": 9.1562,
+      "step": 1609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994518360044611e-05,
+      "loss": 5.8438,
+      "step": 1610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994502604115009e-05,
+      "loss": 8.5312,
+      "step": 1611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994486825605344e-05,
+      "loss": 7.9688,
+      "step": 1612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994471024515797e-05,
+      "loss": 9.3125,
+      "step": 1613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9944552008465456e-05,
+      "loss": 2.9531,
+      "step": 1614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.99443935459777e-05,
+      "loss": 5.5,
+      "step": 1615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9944234857696485e-05,
+      "loss": 3.1875,
+      "step": 1616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9944075943623605e-05,
+      "loss": 4.2656,
+      "step": 1617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994391680376087e-05,
+      "loss": 3.0,
+      "step": 1618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994375743811008e-05,
+      "loss": 4.4688,
+      "step": 1619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994359784667303e-05,
+      "loss": 5.3125,
+      "step": 1620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994343802945154e-05,
+      "loss": 2.9844,
+      "step": 1621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994327798644741e-05,
+      "loss": 5.625,
+      "step": 1622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9943117717662455e-05,
+      "loss": 6.2188,
+      "step": 1623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.9942957223098486e-05,
+      "loss": 3.7812,
+      "step": 1624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994279650275732e-05,
+      "loss": 2.2188,
+      "step": 1625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994263555664079e-05,
+      "loss": 5.1719,
+      "step": 1626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.99424743847507e-05,
+      "loss": 6.9062,
+      "step": 1627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.994231298708888e-05,
+      "loss": 5.7812,
+      "step": 1628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.994215136365716e-05,
+      "loss": 5.8906,
+      "step": 1629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9941989514457375e-05,
+      "loss": 5.1562,
+      "step": 1630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9941827439491344e-05,
+      "loss": 6.4688,
+      "step": 1631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.994166513876091e-05,
+      "loss": 3.9062,
+      "step": 1632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9941502612267916e-05,
+      "loss": 5.25,
+      "step": 1633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.994133986001419e-05,
+      "loss": 6.4062,
+      "step": 1634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9941176882001575e-05,
+      "loss": 6.3125,
+      "step": 1635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.994101367823193e-05,
+      "loss": 6.6328,
+      "step": 1636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9940850248707094e-05,
+      "loss": 4.5,
+      "step": 1637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9940686593428915e-05,
+      "loss": 3.7812,
+      "step": 1638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.994052271239925e-05,
+      "loss": 4.125,
+      "step": 1639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9940358605619944e-05,
+      "loss": 1.7969,
+      "step": 1640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.994019427309287e-05,
+      "loss": 6.0312,
+      "step": 1641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.994002971481988e-05,
+      "loss": 5.1406,
+      "step": 1642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.993986493080284e-05,
+      "loss": 7.1562,
+      "step": 1643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9939699921043614e-05,
+      "loss": 4.1875,
+      "step": 1644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.993953468554407e-05,
+      "loss": 7.0,
+      "step": 1645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9939369224306076e-05,
+      "loss": 5.7188,
+      "step": 1646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9939203537331505e-05,
+      "loss": 3.1875,
+      "step": 1647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.993903762462224e-05,
+      "loss": 6.75,
+      "step": 1648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.993887148618015e-05,
+      "loss": 5.8125,
+      "step": 1649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.993870512200713e-05,
+      "loss": 5.2812,
+      "step": 1650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.993853853210505e-05,
+      "loss": 5.5781,
+      "step": 1651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.99383717164758e-05,
+      "loss": 6.6562,
+      "step": 1652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.993820467512127e-05,
+      "loss": 8.0,
+      "step": 1653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9938037408043354e-05,
+      "loss": 6.3438,
+      "step": 1654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.993786991524394e-05,
+      "loss": 6.1719,
+      "step": 1655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.9937702196724914e-05,
+      "loss": 5.1719,
+      "step": 1656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.99375342524882e-05,
+      "loss": 6.4375,
+      "step": 1657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.993736608253568e-05,
+      "loss": 6.5781,
+      "step": 1658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993719768686926e-05,
+      "loss": 8.7344,
+      "step": 1659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993702906549086e-05,
+      "loss": 8.625,
+      "step": 1660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9936860218402374e-05,
+      "loss": 7.9062,
+      "step": 1661
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993669114560572e-05,
+      "loss": 6.7656,
+      "step": 1662
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9936521847102814e-05,
+      "loss": 6.1719,
+      "step": 1663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993635232289556e-05,
+      "loss": 4.3438,
+      "step": 1664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9936182572985896e-05,
+      "loss": 3.5781,
+      "step": 1665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993601259737573e-05,
+      "loss": 2.8594,
+      "step": 1666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9935842396066995e-05,
+      "loss": 3.3125,
+      "step": 1667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993567196906161e-05,
+      "loss": 4.0,
+      "step": 1668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993550131636151e-05,
+      "loss": 2.875,
+      "step": 1669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9935330437968617e-05,
+      "loss": 2.3125,
+      "step": 1670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993515933388489e-05,
+      "loss": 4.3438,
+      "step": 1671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993498800411223e-05,
+      "loss": 5.4688,
+      "step": 1672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993481644865261e-05,
+      "loss": 5.5156,
+      "step": 1673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9934644667507955e-05,
+      "loss": 5.7188,
+      "step": 1674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9934472660680215e-05,
+      "loss": 5.4688,
+      "step": 1675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9934300428171335e-05,
+      "loss": 5.0156,
+      "step": 1676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9934127969983264e-05,
+      "loss": 8.0,
+      "step": 1677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993395528611796e-05,
+      "loss": 3.5,
+      "step": 1678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993378237657737e-05,
+      "loss": 4.2031,
+      "step": 1679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993360924136345e-05,
+      "loss": 5.6562,
+      "step": 1680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993343588047818e-05,
+      "loss": 6.2188,
+      "step": 1681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.99332622939235e-05,
+      "loss": 2.4531,
+      "step": 1682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993308848170138e-05,
+      "loss": 3.5,
+      "step": 1683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993291444381379e-05,
+      "loss": 7.5,
+      "step": 1684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993274018026271e-05,
+      "loss": 5.5312,
+      "step": 1685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.9932565691050104e-05,
+      "loss": 5.8438,
+      "step": 1686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993239097617794e-05,
+      "loss": 4.4844,
+      "step": 1687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993221603564821e-05,
+      "loss": 3.125,
+      "step": 1688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.993204086946289e-05,
+      "loss": 6.3906,
+      "step": 1689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9931865477623956e-05,
+      "loss": 3.7188,
+      "step": 1690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.99316898601334e-05,
+      "loss": 5.9688,
+      "step": 1691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.993151401699321e-05,
+      "loss": 3.4219,
+      "step": 1692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.993133794820538e-05,
+      "loss": 5.0781,
+      "step": 1693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.993116165377189e-05,
+      "loss": 4.9688,
+      "step": 1694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.993098513369475e-05,
+      "loss": 4.875,
+      "step": 1695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.993080838797595e-05,
+      "loss": 7.1094,
+      "step": 1696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9930631416617504e-05,
+      "loss": 4.125,
+      "step": 1697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.99304542196214e-05,
+      "loss": 3.8438,
+      "step": 1698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9930276796989645e-05,
+      "loss": 3.7344,
+      "step": 1699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.993009914872426e-05,
+      "loss": 6.0,
+      "step": 1700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9929921274827256e-05,
+      "loss": 3.0312,
+      "step": 1701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.992974317530063e-05,
+      "loss": 6.0938,
+      "step": 1702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.99295648501464e-05,
+      "loss": 6.0,
+      "step": 1703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.99293862993666e-05,
+      "loss": 6.0781,
+      "step": 1704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.992920752296325e-05,
+      "loss": 5.0156,
+      "step": 1705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.992902852093836e-05,
+      "loss": 4.25,
+      "step": 1706
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.992884929329397e-05,
+      "loss": 5.375,
+      "step": 1707
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.99286698400321e-05,
+      "loss": 8.1094,
+      "step": 1708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.992849016115479e-05,
+      "loss": 5.3125,
+      "step": 1709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.992831025666406e-05,
+      "loss": 5.0781,
+      "step": 1710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9928130126561966e-05,
+      "loss": 4.9375,
+      "step": 1711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9927949770850535e-05,
+      "loss": 3.3438,
+      "step": 1712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9927769189531804e-05,
+      "loss": 7.2031,
+      "step": 1713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.992758838260784e-05,
+      "loss": 4.3125,
+      "step": 1714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.992740735008066e-05,
+      "loss": 5.0312,
+      "step": 1715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9927226091952335e-05,
+      "loss": 6.3281,
+      "step": 1716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.99270446082249e-05,
+      "loss": 8.5312,
+      "step": 1717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.9926862898900435e-05,
+      "loss": 4.1875,
+      "step": 1718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.992668096398098e-05,
+      "loss": 2.9688,
+      "step": 1719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992649880346859e-05,
+      "loss": 5.3125,
+      "step": 1720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9926316417365335e-05,
+      "loss": 1.7188,
+      "step": 1721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992613380567328e-05,
+      "loss": 4.125,
+      "step": 1722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992595096839449e-05,
+      "loss": 6.6875,
+      "step": 1723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992576790553104e-05,
+      "loss": 3.6562,
+      "step": 1724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992558461708499e-05,
+      "loss": 5.7812,
+      "step": 1725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992540110305843e-05,
+      "loss": 6.8281,
+      "step": 1726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992521736345343e-05,
+      "loss": 8.1406,
+      "step": 1727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992503339827207e-05,
+      "loss": 3.25,
+      "step": 1728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992484920751644e-05,
+      "loss": 4.5625,
+      "step": 1729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992466479118861e-05,
+      "loss": 4.7188,
+      "step": 1730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992448014929069e-05,
+      "loss": 6.2031,
+      "step": 1731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992429528182475e-05,
+      "loss": 5.6094,
+      "step": 1732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9924110188792883e-05,
+      "loss": 8.0625,
+      "step": 1733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9923924870197196e-05,
+      "loss": 5.625,
+      "step": 1734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9923739326039785e-05,
+      "loss": 3.5,
+      "step": 1735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992355355632275e-05,
+      "loss": 6.875,
+      "step": 1736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992336756104819e-05,
+      "loss": 7.1406,
+      "step": 1737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9923181340218216e-05,
+      "loss": 6.9688,
+      "step": 1738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9922994893834934e-05,
+      "loss": 5.6406,
+      "step": 1739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9922808221900445e-05,
+      "loss": 4.0312,
+      "step": 1740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992262132441688e-05,
+      "loss": 6.2812,
+      "step": 1741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992243420138635e-05,
+      "loss": 3.7969,
+      "step": 1742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992224685281095e-05,
+      "loss": 2.4688,
+      "step": 1743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992205927869284e-05,
+      "loss": 6.5312,
+      "step": 1744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992187147903412e-05,
+      "loss": 4.2344,
+      "step": 1745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992168345383692e-05,
+      "loss": 7.1406,
+      "step": 1746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9921495203103376e-05,
+      "loss": 9.2031,
+      "step": 1747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.99213067268356e-05,
+      "loss": 10.0,
+      "step": 1748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992111802503575e-05,
+      "loss": 11.6875,
+      "step": 1749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.992092909770594e-05,
+      "loss": 5.7031,
+      "step": 1750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9920739944848324e-05,
+      "loss": 9.3125,
+      "step": 1751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.992055056646504e-05,
+      "loss": 7.375,
+      "step": 1752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.992036096255823e-05,
+      "loss": 10.8125,
+      "step": 1753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.992017113313004e-05,
+      "loss": 5.0938,
+      "step": 1754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991998107818262e-05,
+      "loss": 5.2812,
+      "step": 1755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991979079771812e-05,
+      "loss": 6.0938,
+      "step": 1756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.99196002917387e-05,
+      "loss": 6.625,
+      "step": 1757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991940956024651e-05,
+      "loss": 5.2969,
+      "step": 1758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991921860324372e-05,
+      "loss": 5.9219,
+      "step": 1759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991902742073248e-05,
+      "loss": 2.8125,
+      "step": 1760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9918836012714954e-05,
+      "loss": 6.3281,
+      "step": 1761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991864437919332e-05,
+      "loss": 8.1875,
+      "step": 1762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991845252016973e-05,
+      "loss": 4.3125,
+      "step": 1763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991826043564638e-05,
+      "loss": 4.5,
+      "step": 1764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991806812562543e-05,
+      "loss": 4.9062,
+      "step": 1765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9917875590109055e-05,
+      "loss": 3.2188,
+      "step": 1766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991768282909944e-05,
+      "loss": 6.0781,
+      "step": 1767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991748984259876e-05,
+      "loss": 6.4375,
+      "step": 1768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991729663060921e-05,
+      "loss": 3.875,
+      "step": 1769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991710319313298e-05,
+      "loss": 6.375,
+      "step": 1770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991690953017225e-05,
+      "loss": 5.3125,
+      "step": 1771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991671564172921e-05,
+      "loss": 3.0938,
+      "step": 1772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991652152780606e-05,
+      "loss": 6.6406,
+      "step": 1773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9916327188405006e-05,
+      "loss": 10.7188,
+      "step": 1774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991613262352823e-05,
+      "loss": 4.875,
+      "step": 1775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991593783317795e-05,
+      "loss": 2.8438,
+      "step": 1776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991574281735637e-05,
+      "loss": 8.6875,
+      "step": 1777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991554757606569e-05,
+      "loss": 3.5312,
+      "step": 1778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991535210930813e-05,
+      "loss": 3.0,
+      "step": 1779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991515641708589e-05,
+      "loss": 7.8125,
+      "step": 1780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991496049940119e-05,
+      "loss": 5.75,
+      "step": 1781
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991476435625625e-05,
+      "loss": 5.25,
+      "step": 1782
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.99145679876533e-05,
+      "loss": 3.6094,
+      "step": 1783
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991437139359455e-05,
+      "loss": 4.4531,
+      "step": 1784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991417457408224e-05,
+      "loss": 3.4688,
+      "step": 1785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991397752911858e-05,
+      "loss": 3.5156,
+      "step": 1786
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9913780258705806e-05,
+      "loss": 3.3125,
+      "step": 1787
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991358276284616e-05,
+      "loss": 6.4062,
+      "step": 1788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9913385041541866e-05,
+      "loss": 5.1719,
+      "step": 1789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991318709479517e-05,
+      "loss": 4.8125,
+      "step": 1790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991298892260832e-05,
+      "loss": 7.6094,
+      "step": 1791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991279052498354e-05,
+      "loss": 3.3281,
+      "step": 1792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.99125919019231e-05,
+      "loss": 11.5,
+      "step": 1793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991239305342923e-05,
+      "loss": 5.9688,
+      "step": 1794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991219397950418e-05,
+      "loss": 5.6875,
+      "step": 1795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9911994680150215e-05,
+      "loss": 9.25,
+      "step": 1796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991179515536959e-05,
+      "loss": 3.9688,
+      "step": 1797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991159540516456e-05,
+      "loss": 6.25,
+      "step": 1798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991139542953739e-05,
+      "loss": 4.3281,
+      "step": 1799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991119522849034e-05,
+      "loss": 7.9219,
+      "step": 1800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991099480202568e-05,
+      "loss": 3.3438,
+      "step": 1801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991079415014568e-05,
+      "loss": 6.6406,
+      "step": 1802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9910593272852593e-05,
+      "loss": 3.9062,
+      "step": 1803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991039217014873e-05,
+      "loss": 3.2188,
+      "step": 1804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991019084203633e-05,
+      "loss": 2.9219,
+      "step": 1805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9909989288517696e-05,
+      "loss": 4.9531,
+      "step": 1806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.99097875095951e-05,
+      "loss": 7.0,
+      "step": 1807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.990958550527083e-05,
+      "loss": 6.6562,
+      "step": 1808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.990938327554717e-05,
+      "loss": 2.4375,
+      "step": 1809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9909180820426415e-05,
+      "loss": 2.5625,
+      "step": 1810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.990897813991085e-05,
+      "loss": 3.3906,
+      "step": 1811
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990877523400278e-05,
+      "loss": 5.5938,
+      "step": 1812
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990857210270449e-05,
+      "loss": 6.5469,
+      "step": 1813
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9908368746018276e-05,
+      "loss": 2.6875,
+      "step": 1814
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990816516394646e-05,
+      "loss": 5.5,
+      "step": 1815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990796135649133e-05,
+      "loss": 2.2188,
+      "step": 1816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9907757323655206e-05,
+      "loss": 4.1094,
+      "step": 1817
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9907553065440385e-05,
+      "loss": 4.1875,
+      "step": 1818
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9907348581849186e-05,
+      "loss": 3.375,
+      "step": 1819
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9907143872883926e-05,
+      "loss": 3.4375,
+      "step": 1820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990693893854692e-05,
+      "loss": 4.1562,
+      "step": 1821
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990673377884049e-05,
+      "loss": 4.9688,
+      "step": 1822
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9906528393766953e-05,
+      "loss": 6.2344,
+      "step": 1823
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990632278332864e-05,
+      "loss": 6.4375,
+      "step": 1824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990611694752787e-05,
+      "loss": 5.6875,
+      "step": 1825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990591088636698e-05,
+      "loss": 4.6875,
+      "step": 1826
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990570459984832e-05,
+      "loss": 5.4062,
+      "step": 1827
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990549808797419e-05,
+      "loss": 5.8906,
+      "step": 1828
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990529135074695e-05,
+      "loss": 4.1875,
+      "step": 1829
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990508438816894e-05,
+      "loss": 5.2812,
+      "step": 1830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.99048772002425e-05,
+      "loss": 5.1562,
+      "step": 1831
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9904669786969976e-05,
+      "loss": 4.3125,
+      "step": 1832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9904462148353714e-05,
+      "loss": 8.1719,
+      "step": 1833
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9904254284396067e-05,
+      "loss": 3.0938,
+      "step": 1834
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9904046195099384e-05,
+      "loss": 3.1875,
+      "step": 1835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9903837880466026e-05,
+      "loss": 4.6719,
+      "step": 1836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990362934049835e-05,
+      "loss": 4.6562,
+      "step": 1837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990342057519872e-05,
+      "loss": 4.2812,
+      "step": 1838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990321158456949e-05,
+      "loss": 4.3281,
+      "step": 1839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.990300236861303e-05,
+      "loss": 5.6875,
+      "step": 1840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9902792927331716e-05,
+      "loss": 4.3125,
+      "step": 1841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9902583260727915e-05,
+      "loss": 5.0312,
+      "step": 1842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9902373368804e-05,
+      "loss": 4.75,
+      "step": 1843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990216325156234e-05,
+      "loss": 8.7188,
+      "step": 1844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990195290900532e-05,
+      "loss": 4.625,
+      "step": 1845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990174234113533e-05,
+      "loss": 10.1094,
+      "step": 1846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990153154795474e-05,
+      "loss": 4.0625,
+      "step": 1847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990132052946594e-05,
+      "loss": 5.7812,
+      "step": 1848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990110928567132e-05,
+      "loss": 6.8594,
+      "step": 1849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9900897816573285e-05,
+      "loss": 4.7812,
+      "step": 1850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990068612217421e-05,
+      "loss": 2.5,
+      "step": 1851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9900474202476495e-05,
+      "loss": 7.25,
+      "step": 1852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.990026205748254e-05,
+      "loss": 2.7969,
+      "step": 1853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9900049687194755e-05,
+      "loss": 4.9688,
+      "step": 1854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989983709161553e-05,
+      "loss": 4.1094,
+      "step": 1855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989962427074729e-05,
+      "loss": 4.9219,
+      "step": 1856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989941122459243e-05,
+      "loss": 4.1562,
+      "step": 1857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9899197953153366e-05,
+      "loss": 4.6094,
+      "step": 1858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9898984456432514e-05,
+      "loss": 2.4375,
+      "step": 1859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9898770734432286e-05,
+      "loss": 3.1562,
+      "step": 1860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9898556787155115e-05,
+      "loss": 4.6562,
+      "step": 1861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98983426146034e-05,
+      "loss": 7.4375,
+      "step": 1862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989812821677959e-05,
+      "loss": 4.0625,
+      "step": 1863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98979135936861e-05,
+      "loss": 5.3281,
+      "step": 1864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989769874532536e-05,
+      "loss": 4.4688,
+      "step": 1865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.98974836716998e-05,
+      "loss": 4.75,
+      "step": 1866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989726837281186e-05,
+      "loss": 3.8438,
+      "step": 1867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9897052848663984e-05,
+      "loss": 5.4688,
+      "step": 1868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9896837099258596e-05,
+      "loss": 5.8594,
+      "step": 1869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989662112459815e-05,
+      "loss": 5.4531,
+      "step": 1870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989640492468508e-05,
+      "loss": 3.9688,
+      "step": 1871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.989618849952185e-05,
+      "loss": 4.4219,
+      "step": 1872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.98959718491109e-05,
+      "loss": 6.2188,
+      "step": 1873
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989575497345469e-05,
+      "loss": 5.7656,
+      "step": 1874
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9895537872555655e-05,
+      "loss": 4.375,
+      "step": 1875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989532054641627e-05,
+      "loss": 8.5625,
+      "step": 1876
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9895102995039e-05,
+      "loss": 8.0312,
+      "step": 1877
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.98948852184263e-05,
+      "loss": 3.6562,
+      "step": 1878
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989466721658063e-05,
+      "loss": 3.6094,
+      "step": 1879
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989444898950447e-05,
+      "loss": 3.5312,
+      "step": 1880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989423053720028e-05,
+      "loss": 3.7344,
+      "step": 1881
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989401185967055e-05,
+      "loss": 5.8594,
+      "step": 1882
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9893792956917724e-05,
+      "loss": 4.3594,
+      "step": 1883
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9893573828944315e-05,
+      "loss": 4.7812,
+      "step": 1884
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989335447575278e-05,
+      "loss": 7.1562,
+      "step": 1885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989313489734561e-05,
+      "loss": 3.125,
+      "step": 1886
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989291509372529e-05,
+      "loss": 7.2656,
+      "step": 1887
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989269506489432e-05,
+      "loss": 6.0469,
+      "step": 1888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9892474810855174e-05,
+      "loss": 6.2031,
+      "step": 1889
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9892254331610355e-05,
+      "loss": 4.75,
+      "step": 1890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989203362716236e-05,
+      "loss": 6.5781,
+      "step": 1891
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989181269751368e-05,
+      "loss": 7.7188,
+      "step": 1892
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989159154266682e-05,
+      "loss": 6.6875,
+      "step": 1893
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989137016262429e-05,
+      "loss": 6.7031,
+      "step": 1894
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989114855738859e-05,
+      "loss": 3.9688,
+      "step": 1895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989092672696222e-05,
+      "loss": 6.1875,
+      "step": 1896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.98907046713477e-05,
+      "loss": 5.7812,
+      "step": 1897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989048239054755e-05,
+      "loss": 8.4062,
+      "step": 1898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989025988456429e-05,
+      "loss": 9.8125,
+      "step": 1899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.989003715340042e-05,
+      "loss": 11.5312,
+      "step": 1900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.988981419705847e-05,
+      "loss": 7.375,
+      "step": 1901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9889591015540975e-05,
+      "loss": 7.5,
+      "step": 1902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988936760885044e-05,
+      "loss": 6.1719,
+      "step": 1903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988914397698942e-05,
+      "loss": 9.0312,
+      "step": 1904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988892011996042e-05,
+      "loss": 4.8281,
+      "step": 1905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9888696037766e-05,
+      "loss": 5.8906,
+      "step": 1906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988847173040868e-05,
+      "loss": 7.9688,
+      "step": 1907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9888247197891e-05,
+      "loss": 3.8125,
+      "step": 1908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988802244021551e-05,
+      "loss": 2.7812,
+      "step": 1909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988779745738475e-05,
+      "loss": 4.9062,
+      "step": 1910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988757224940127e-05,
+      "loss": 4.5469,
+      "step": 1911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988734681626762e-05,
+      "loss": 8.1094,
+      "step": 1912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988712115798634e-05,
+      "loss": 8.2812,
+      "step": 1913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9886895274559996e-05,
+      "loss": 6.25,
+      "step": 1914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988666916599114e-05,
+      "loss": 4.1875,
+      "step": 1915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988644283228235e-05,
+      "loss": 4.9688,
+      "step": 1916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988621627343616e-05,
+      "loss": 5.5781,
+      "step": 1917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988598948945516e-05,
+      "loss": 4.0156,
+      "step": 1918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9885762480341894e-05,
+      "loss": 5.3438,
+      "step": 1919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988553524609895e-05,
+      "loss": 3.6562,
+      "step": 1920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988530778672889e-05,
+      "loss": 8.7031,
+      "step": 1921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.98850801022343e-05,
+      "loss": 6.7031,
+      "step": 1922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988485219261775e-05,
+      "loss": 7.2188,
+      "step": 1923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9884624057881813e-05,
+      "loss": 2.7812,
+      "step": 1924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988439569802909e-05,
+      "loss": 4.1875,
+      "step": 1925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988416711306215e-05,
+      "loss": 5.7188,
+      "step": 1926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9883938302983594e-05,
+      "loss": 3.875,
+      "step": 1927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9883709267795995e-05,
+      "loss": 4.4844,
+      "step": 1928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988348000750197e-05,
+      "loss": 5.125,
+      "step": 1929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.98832505221041e-05,
+      "loss": 6.2969,
+      "step": 1930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988302081160497e-05,
+      "loss": 4.25,
+      "step": 1931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.988279087600721e-05,
+      "loss": 4.7969,
+      "step": 1932
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.98825607153134e-05,
+      "loss": 4.5625,
+      "step": 1933
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.988233032952615e-05,
+      "loss": 7.9375,
+      "step": 1934
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.988209971864808e-05,
+      "loss": 5.8125,
+      "step": 1935
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9881868882681793e-05,
+      "loss": 6.75,
+      "step": 1936
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9881637821629896e-05,
+      "loss": 7.2812,
+      "step": 1937
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.988140653549501e-05,
+      "loss": 5.7344,
+      "step": 1938
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.988117502427976e-05,
+      "loss": 5.3438,
+      "step": 1939
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.988094328798676e-05,
+      "loss": 7.875,
+      "step": 1940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.988071132661863e-05,
+      "loss": 5.8125,
+      "step": 1941
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9880479140178005e-05,
+      "loss": 6.0,
+      "step": 1942
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.988024672866751e-05,
+      "loss": 7.4062,
+      "step": 1943
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.988001409208977e-05,
+      "loss": 7.6875,
+      "step": 1944
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.987978123044743e-05,
+      "loss": 12.0938,
+      "step": 1945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.987954814374312e-05,
+      "loss": 8.1562,
+      "step": 1946
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.987931483197947e-05,
+      "loss": 8.0625,
+      "step": 1947
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.987908129515915e-05,
+      "loss": 4.3438,
+      "step": 1948
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9878847533284764e-05,
+      "loss": 7.0,
+      "step": 1949
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9878613546358996e-05,
+      "loss": 4.4531,
+      "step": 1950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.987837933438447e-05,
+      "loss": 2.7188,
+      "step": 1951
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.987814489736384e-05,
+      "loss": 7.5,
+      "step": 1952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9877910235299777e-05,
+      "loss": 7.9062,
+      "step": 1953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.987767534819492e-05,
+      "loss": 5.4219,
+      "step": 1954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9877440236051936e-05,
+      "loss": 5.3281,
+      "step": 1955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.987720489887348e-05,
+      "loss": 2.5469,
+      "step": 1956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.987696933666223e-05,
+      "loss": 4.9531,
+      "step": 1957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.987673354942084e-05,
+      "loss": 8.0938,
+      "step": 1958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.987649753715198e-05,
+      "loss": 3.4375,
+      "step": 1959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.987626129985833e-05,
+      "loss": 3.375,
+      "step": 1960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.987602483754256e-05,
+      "loss": 3.3125,
+      "step": 1961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.987578815020735e-05,
+      "loss": 5.1562,
+      "step": 1962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9875551237855384e-05,
+      "loss": 3.625,
+      "step": 1963
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9875314100489325e-05,
+      "loss": 4.0938,
+      "step": 1964
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9875076738111864e-05,
+      "loss": 6.125,
+      "step": 1965
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9874839150725705e-05,
+      "loss": 5.4219,
+      "step": 1966
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.987460133833352e-05,
+      "loss": 7.3438,
+      "step": 1967
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.987436330093801e-05,
+      "loss": 4.5469,
+      "step": 1968
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.987412503854187e-05,
+      "loss": 5.1719,
+      "step": 1969
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.98738865511478e-05,
+      "loss": 2.7188,
+      "step": 1970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9873647838758485e-05,
+      "loss": 7.6406,
+      "step": 1971
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.987340890137664e-05,
+      "loss": 2.0312,
+      "step": 1972
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.987316973900497e-05,
+      "loss": 6.0625,
+      "step": 1973
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9872930351646187e-05,
+      "loss": 3.75,
+      "step": 1974
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.987269073930298e-05,
+      "loss": 4.3594,
+      "step": 1975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.987245090197808e-05,
+      "loss": 4.4062,
+      "step": 1976
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.987221083967421e-05,
+      "loss": 4.9531,
+      "step": 1977
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9871970552394066e-05,
+      "loss": 5.2188,
+      "step": 1978
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.987173004014038e-05,
+      "loss": 7.6875,
+      "step": 1979
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9871489302915864e-05,
+      "loss": 3.875,
+      "step": 1980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.987124834072327e-05,
+      "loss": 6.875,
+      "step": 1981
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9871007153565295e-05,
+      "loss": 9.7969,
+      "step": 1982
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.987076574144469e-05,
+      "loss": 4.7656,
+      "step": 1983
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.987052410436418e-05,
+      "loss": 5.125,
+      "step": 1984
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.98702822423265e-05,
+      "loss": 3.1875,
+      "step": 1985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.987004015533439e-05,
+      "loss": 7.4688,
+      "step": 1986
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.986979784339059e-05,
+      "loss": 3.9688,
+      "step": 1987
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.986955530649785e-05,
+      "loss": 4.3438,
+      "step": 1988
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.986931254465891e-05,
+      "loss": 2.125,
+      "step": 1989
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.986906955787651e-05,
+      "loss": 4.9375,
+      "step": 1990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.986882634615341e-05,
+      "loss": 3.0625,
+      "step": 1991
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.986858290949236e-05,
+      "loss": 5.0938,
+      "step": 1992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.986833924789613e-05,
+      "loss": 6.0312,
+      "step": 1993
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986809536136746e-05,
+      "loss": 4.5,
+      "step": 1994
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986785124990911e-05,
+      "loss": 7.3125,
+      "step": 1995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986760691352387e-05,
+      "loss": 5.25,
+      "step": 1996
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986736235221447e-05,
+      "loss": 5.875,
+      "step": 1997
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9867117565983705e-05,
+      "loss": 5.9062,
+      "step": 1998
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986687255483434e-05,
+      "loss": 4.9688,
+      "step": 1999
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986662731876914e-05,
+      "loss": 4.4375,
+      "step": 2000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986638185779089e-05,
+      "loss": 7.1094,
+      "step": 2001
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9866136171902375e-05,
+      "loss": 5.6406,
+      "step": 2002
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986589026110636e-05,
+      "loss": 6.4219,
+      "step": 2003
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986564412540564e-05,
+      "loss": 3.8438,
+      "step": 2004
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9865397764802993e-05,
+      "loss": 3.9688,
+      "step": 2005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986515117930122e-05,
+      "loss": 5.5,
+      "step": 2006
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9864904368903106e-05,
+      "loss": 6.125,
+      "step": 2007
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986465733361145e-05,
+      "loss": 4.2344,
+      "step": 2008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9864410073429033e-05,
+      "loss": 2.1562,
+      "step": 2009
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986416258835867e-05,
+      "loss": 3.8281,
+      "step": 2010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986391487840316e-05,
+      "loss": 5.1875,
+      "step": 2011
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986366694356531e-05,
+      "loss": 5.6719,
+      "step": 2012
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986341878384792e-05,
+      "loss": 6.1094,
+      "step": 2013
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9863170399253804e-05,
+      "loss": 5.8125,
+      "step": 2014
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986292178978577e-05,
+      "loss": 4.7656,
+      "step": 2015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986267295544663e-05,
+      "loss": 5.7188,
+      "step": 2016
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986242389623921e-05,
+      "loss": 4.8125,
+      "step": 2017
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9862174612166325e-05,
+      "loss": 5.0312,
+      "step": 2018
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.98619251032308e-05,
+      "loss": 8.3594,
+      "step": 2019
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986167536943545e-05,
+      "loss": 2.8438,
+      "step": 2020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986142541078312e-05,
+      "loss": 2.2188,
+      "step": 2021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986117522727663e-05,
+      "loss": 3.1562,
+      "step": 2022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.98609248189188e-05,
+      "loss": 3.6875,
+      "step": 2023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.986067418571249e-05,
+      "loss": 5.1641,
+      "step": 2024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.986042332766051e-05,
+      "loss": 5.9844,
+      "step": 2025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.986017224476572e-05,
+      "loss": 4.4375,
+      "step": 2026
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.985992093703096e-05,
+      "loss": 7.6562,
+      "step": 2027
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.985966940445907e-05,
+      "loss": 2.6875,
+      "step": 2028
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.98594176470529e-05,
+      "loss": 4.3125,
+      "step": 2029
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.98591656648153e-05,
+      "loss": 6.3125,
+      "step": 2030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.985891345774913e-05,
+      "loss": 4.5781,
+      "step": 2031
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9858661025857224e-05,
+      "loss": 4.25,
+      "step": 2032
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.985840836914246e-05,
+      "loss": 4.1875,
+      "step": 2033
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.985815548760769e-05,
+      "loss": 3.4219,
+      "step": 2034
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.985790238125578e-05,
+      "loss": 2.875,
+      "step": 2035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9857649050089595e-05,
+      "loss": 6.7969,
+      "step": 2036
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9857395494112e-05,
+      "loss": 4.8281,
+      "step": 2037
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9857141713325876e-05,
+      "loss": 2.75,
+      "step": 2038
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9856887707734075e-05,
+      "loss": 11.75,
+      "step": 2039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.985663347733949e-05,
+      "loss": 3.5156,
+      "step": 2040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9856379022144995e-05,
+      "loss": 3.8438,
+      "step": 2041
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9856124342153475e-05,
+      "loss": 8.125,
+      "step": 2042
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.98558694373678e-05,
+      "loss": 6.5625,
+      "step": 2043
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.985561430779088e-05,
+      "loss": 3.5781,
+      "step": 2044
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9855358953425574e-05,
+      "loss": 3.9688,
+      "step": 2045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.985510337427479e-05,
+      "loss": 3.6562,
+      "step": 2046
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.985484757034142e-05,
+      "loss": 5.5312,
+      "step": 2047
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.985459154162835e-05,
+      "loss": 3.7188,
+      "step": 2048
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.98543352881385e-05,
+      "loss": 5.125,
+      "step": 2049
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9854078809874745e-05,
+      "loss": 5.5312,
+      "step": 2050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9853822106840014e-05,
+      "loss": 3.625,
+      "step": 2051
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.9853565179037197e-05,
+      "loss": 2.4062,
+      "step": 2052
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.98533080264692e-05,
+      "loss": 7.875,
+      "step": 2053
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.985305064913894e-05,
+      "loss": 4.875,
+      "step": 2054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.985279304704934e-05,
+      "loss": 6.3125,
+      "step": 2055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9852535220203305e-05,
+      "loss": 3.5938,
+      "step": 2056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.985227716860376e-05,
+      "loss": 6.1875,
+      "step": 2057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.985201889225361e-05,
+      "loss": 2.0,
+      "step": 2058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9851760391155804e-05,
+      "loss": 5.9062,
+      "step": 2059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9851501665313256e-05,
+      "loss": 4.9531,
+      "step": 2060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.985124271472889e-05,
+      "loss": 6.8125,
+      "step": 2061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.985098353940565e-05,
+      "loss": 4.1562,
+      "step": 2062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9850724139346454e-05,
+      "loss": 4.25,
+      "step": 2063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9850464514554256e-05,
+      "loss": 3.7188,
+      "step": 2064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9850204665031985e-05,
+      "loss": 7.5625,
+      "step": 2065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.984994459078259e-05,
+      "loss": 4.0312,
+      "step": 2066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9849684291809005e-05,
+      "loss": 4.4688,
+      "step": 2067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9849423768114184e-05,
+      "loss": 4.8438,
+      "step": 2068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.984916301970107e-05,
+      "loss": 3.5312,
+      "step": 2069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.984890204657263e-05,
+      "loss": 4.8906,
+      "step": 2070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9848640848731797e-05,
+      "loss": 5.9688,
+      "step": 2071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.984837942618155e-05,
+      "loss": 3.6875,
+      "step": 2072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9848117778924826e-05,
+      "loss": 4.9375,
+      "step": 2073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.984785590696461e-05,
+      "loss": 3.4375,
+      "step": 2074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.984759381030386e-05,
+      "loss": 4.7188,
+      "step": 2075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9847331488945525e-05,
+      "loss": 2.9844,
+      "step": 2076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.98470689428926e-05,
+      "loss": 3.0312,
+      "step": 2077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.984680617214804e-05,
+      "loss": 2.2188,
+      "step": 2078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.984654317671482e-05,
+      "loss": 4.75,
+      "step": 2079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.984627995659594e-05,
+      "loss": 4.625,
+      "step": 2080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.984601651179435e-05,
+      "loss": 4.875,
+      "step": 2081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.984575284231305e-05,
+      "loss": 4.625,
+      "step": 2082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.9845488948155024e-05,
+      "loss": 3.9375,
+      "step": 2083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.984522482932325e-05,
+      "loss": 7.5625,
+      "step": 2084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.984496048582073e-05,
+      "loss": 5.75,
+      "step": 2085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.984469591765045e-05,
+      "loss": 2.6875,
+      "step": 2086
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.9844431124815406e-05,
+      "loss": 4.6562,
+      "step": 2087
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.98441661073186e-05,
+      "loss": 4.5938,
+      "step": 2088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.984390086516302e-05,
+      "loss": 3.75,
+      "step": 2089
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.984363539835169e-05,
+      "loss": 5.5156,
+      "step": 2090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.98433697068876e-05,
+      "loss": 2.9531,
+      "step": 2091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.984310379077375e-05,
+      "loss": 7.2344,
+      "step": 2092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.984283765001317e-05,
+      "loss": 6.1875,
+      "step": 2093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.984257128460886e-05,
+      "loss": 2.9375,
+      "step": 2094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.9842304694563846e-05,
+      "loss": 3.625,
+      "step": 2095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.984203787988113e-05,
+      "loss": 3.6875,
+      "step": 2096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.984177084056376e-05,
+      "loss": 4.0156,
+      "step": 2097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.984150357661473e-05,
+      "loss": 6.1875,
+      "step": 2098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.984123608803707e-05,
+      "loss": 6.2812,
+      "step": 2099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.9840968374833824e-05,
+      "loss": 4.6406,
+      "step": 2100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.9840700437008014e-05,
+      "loss": 7.1094,
+      "step": 2101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.984043227456268e-05,
+      "loss": 3.4062,
+      "step": 2102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.9840163887500845e-05,
+      "loss": 2.8438,
+      "step": 2103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.983989527582556e-05,
+      "loss": 5.8594,
+      "step": 2104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.983962643953986e-05,
+      "loss": 5.2188,
+      "step": 2105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.983935737864678e-05,
+      "loss": 4.9844,
+      "step": 2106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.9839088093149375e-05,
+      "loss": 3.3125,
+      "step": 2107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.98388185830507e-05,
+      "loss": 7.3906,
+      "step": 2108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.98385488483538e-05,
+      "loss": 6.6562,
+      "step": 2109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.983827888906173e-05,
+      "loss": 7.7188,
+      "step": 2110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.9838008705177537e-05,
+      "loss": 4.0938,
+      "step": 2111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.9837738296704295e-05,
+      "loss": 7.1719,
+      "step": 2112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.983746766364506e-05,
+      "loss": 5.1562,
+      "step": 2113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.983719680600289e-05,
+      "loss": 3.625,
+      "step": 2114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.983692572378085e-05,
+      "loss": 3.9062,
+      "step": 2115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983665441698202e-05,
+      "loss": 4.2344,
+      "step": 2116
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9836382885609465e-05,
+      "loss": 3.4844,
+      "step": 2117
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9836111129666256e-05,
+      "loss": 5.4062,
+      "step": 2118
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983583914915548e-05,
+      "loss": 5.5469,
+      "step": 2119
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9835566944080204e-05,
+      "loss": 4.4062,
+      "step": 2120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9835294514443514e-05,
+      "loss": 5.2812,
+      "step": 2121
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9835021860248496e-05,
+      "loss": 7.0938,
+      "step": 2122
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983474898149824e-05,
+      "loss": 5.0938,
+      "step": 2123
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983447587819583e-05,
+      "loss": 5.0312,
+      "step": 2124
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9834202550344354e-05,
+      "loss": 7.0781,
+      "step": 2125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983392899794692e-05,
+      "loss": 4.3438,
+      "step": 2126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983365522100661e-05,
+      "loss": 3.0312,
+      "step": 2127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983338121952653e-05,
+      "loss": 4.6562,
+      "step": 2128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9833106993509785e-05,
+      "loss": 4.7969,
+      "step": 2129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983283254295947e-05,
+      "loss": 6.0,
+      "step": 2130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983255786787871e-05,
+      "loss": 7.4688,
+      "step": 2131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9832282968270595e-05,
+      "loss": 4.5781,
+      "step": 2132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983200784413825e-05,
+      "loss": 4.8594,
+      "step": 2133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983173249548478e-05,
+      "loss": 2.7812,
+      "step": 2134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983145692231332e-05,
+      "loss": 3.9375,
+      "step": 2135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983118112462696e-05,
+      "loss": 3.0469,
+      "step": 2136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983090510242885e-05,
+      "loss": 4.0625,
+      "step": 2137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.98306288557221e-05,
+      "loss": 4.2812,
+      "step": 2138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983035238450985e-05,
+      "loss": 3.5938,
+      "step": 2139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.983007568879521e-05,
+      "loss": 6.125,
+      "step": 2140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.982979876858133e-05,
+      "loss": 4.0625,
+      "step": 2141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9829521623871336e-05,
+      "loss": 3.5312,
+      "step": 2142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.982924425466838e-05,
+      "loss": 4.625,
+      "step": 2143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.982896666097558e-05,
+      "loss": 4.3594,
+      "step": 2144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9828688842796096e-05,
+      "loss": 7.5625,
+      "step": 2145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.982841080013306e-05,
+      "loss": 4.3906,
+      "step": 2146
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9828132532989634e-05,
+      "loss": 5.9688,
+      "step": 2147
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982785404136895e-05,
+      "loss": 2.4375,
+      "step": 2148
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982757532527419e-05,
+      "loss": 7.9688,
+      "step": 2149
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982729638470848e-05,
+      "loss": 3.9141,
+      "step": 2150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982701721967499e-05,
+      "loss": 4.3594,
+      "step": 2151
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982673783017688e-05,
+      "loss": 4.2969,
+      "step": 2152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982645821621732e-05,
+      "loss": 7.0938,
+      "step": 2153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982617837779946e-05,
+      "loss": 4.1875,
+      "step": 2154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982589831492647e-05,
+      "loss": 4.5,
+      "step": 2155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982561802760154e-05,
+      "loss": 5.375,
+      "step": 2156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982533751582782e-05,
+      "loss": 2.6719,
+      "step": 2157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982505677960851e-05,
+      "loss": 5.6562,
+      "step": 2158
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982477581894676e-05,
+      "loss": 4.125,
+      "step": 2159
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9824494633845774e-05,
+      "loss": 6.7031,
+      "step": 2160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9824213224308724e-05,
+      "loss": 3.8438,
+      "step": 2161
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.98239315903388e-05,
+      "loss": 4.2188,
+      "step": 2162
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9823649731939184e-05,
+      "loss": 7.7188,
+      "step": 2163
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982336764911307e-05,
+      "loss": 2.875,
+      "step": 2164
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982308534186366e-05,
+      "loss": 3.5938,
+      "step": 2165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9822802810194136e-05,
+      "loss": 6.2812,
+      "step": 2166
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982252005410771e-05,
+      "loss": 7.0156,
+      "step": 2167
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982223707360757e-05,
+      "loss": 4.3281,
+      "step": 2168
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982195386869693e-05,
+      "loss": 9.125,
+      "step": 2169
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982167043937899e-05,
+      "loss": 5.9688,
+      "step": 2170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9821386785656973e-05,
+      "loss": 4.8594,
+      "step": 2171
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982110290753407e-05,
+      "loss": 2.7812,
+      "step": 2172
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9820818805013505e-05,
+      "loss": 4.875,
+      "step": 2173
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982053447809849e-05,
+      "loss": 4.3281,
+      "step": 2174
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982024992679225e-05,
+      "loss": 5.4531,
+      "step": 2175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9819965151098e-05,
+      "loss": 5.0469,
+      "step": 2176
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9819680151018974e-05,
+      "loss": 3.4375,
+      "step": 2177
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981939492655839e-05,
+      "loss": 3.8438,
+      "step": 2178
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981910947771947e-05,
+      "loss": 2.6875,
+      "step": 2179
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9818823804505465e-05,
+      "loss": 3.9062,
+      "step": 2180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981853790691959e-05,
+      "loss": 3.2344,
+      "step": 2181
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9818251784965094e-05,
+      "loss": 3.0,
+      "step": 2182
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9817965438645215e-05,
+      "loss": 4.5312,
+      "step": 2183
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981767886796319e-05,
+      "loss": 3.1562,
+      "step": 2184
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9817392072922267e-05,
+      "loss": 9.0469,
+      "step": 2185
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981710505352569e-05,
+      "loss": 5.5312,
+      "step": 2186
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981681780977671e-05,
+      "loss": 2.4688,
+      "step": 2187
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981653034167858e-05,
+      "loss": 4.3125,
+      "step": 2188
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9816242649234554e-05,
+      "loss": 7.2031,
+      "step": 2189
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981595473244789e-05,
+      "loss": 6.7656,
+      "step": 2190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981566659132184e-05,
+      "loss": 2.7188,
+      "step": 2191
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981537822585967e-05,
+      "loss": 3.25,
+      "step": 2192
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981508963606466e-05,
+      "loss": 5.0,
+      "step": 2193
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9814800821940054e-05,
+      "loss": 6.8906,
+      "step": 2194
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9814511783489125e-05,
+      "loss": 4.2344,
+      "step": 2195
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981422252071516e-05,
+      "loss": 5.9688,
+      "step": 2196
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9813933033621427e-05,
+      "loss": 5.7188,
+      "step": 2197
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.98136433222112e-05,
+      "loss": 4.3594,
+      "step": 2198
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981335338648776e-05,
+      "loss": 6.2812,
+      "step": 2199
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981306322645439e-05,
+      "loss": 3.9219,
+      "step": 2200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981277284211437e-05,
+      "loss": 7.9844,
+      "step": 2201
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9812482233471e-05,
+      "loss": 5.6562,
+      "step": 2202
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9812191400527566e-05,
+      "loss": 4.1875,
+      "step": 2203
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981190034328735e-05,
+      "loss": 5.2969,
+      "step": 2204
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.981160906175365e-05,
+      "loss": 3.9375,
+      "step": 2205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9811317555929776e-05,
+      "loss": 6.6406,
+      "step": 2206
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.981102582581902e-05,
+      "loss": 5.2812,
+      "step": 2207
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.981073387142468e-05,
+      "loss": 4.0,
+      "step": 2208
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.981044169275006e-05,
+      "loss": 3.4688,
+      "step": 2209
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.981014928979849e-05,
+      "loss": 6.4375,
+      "step": 2210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9809856662573254e-05,
+      "loss": 6.0469,
+      "step": 2211
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.980956381107768e-05,
+      "loss": 6.2031,
+      "step": 2212
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.980927073531507e-05,
+      "loss": 3.6562,
+      "step": 2213
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9808977435288756e-05,
+      "loss": 4.8125,
+      "step": 2214
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.980868391100205e-05,
+      "loss": 6.2188,
+      "step": 2215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9808390162458286e-05,
+      "loss": 3.5938,
+      "step": 2216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.980809618966078e-05,
+      "loss": 5.3281,
+      "step": 2217
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.980780199261285e-05,
+      "loss": 3.2969,
+      "step": 2218
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.980750757131785e-05,
+      "loss": 6.875,
+      "step": 2219
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.98072129257791e-05,
+      "loss": 6.9844,
+      "step": 2220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.980691805599993e-05,
+      "loss": 3.75,
+      "step": 2221
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9806622961983686e-05,
+      "loss": 3.7969,
+      "step": 2222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9806327643733716e-05,
+      "loss": 6.9062,
+      "step": 2223
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.980603210125335e-05,
+      "loss": 4.5938,
+      "step": 2224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.980573633454594e-05,
+      "loss": 3.9062,
+      "step": 2225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.980544034361483e-05,
+      "loss": 3.9688,
+      "step": 2226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9805144128463385e-05,
+      "loss": 4.7969,
+      "step": 2227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9804847689094944e-05,
+      "loss": 3.8125,
+      "step": 2228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9804551025512864e-05,
+      "loss": 6.375,
+      "step": 2229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.980425413772051e-05,
+      "loss": 6.875,
+      "step": 2230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9803957025721236e-05,
+      "loss": 5.2188,
+      "step": 2231
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.980365968951841e-05,
+      "loss": 4.0312,
+      "step": 2232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.980336212911541e-05,
+      "loss": 5.9688,
+      "step": 2233
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9803064344515574e-05,
+      "loss": 6.25,
+      "step": 2234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9802766335722306e-05,
+      "loss": 6.7344,
+      "step": 2235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.9802468102738953e-05,
+      "loss": 6.7188,
+      "step": 2236
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.980216964556892e-05,
+      "loss": 5.1719,
+      "step": 2237
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.980187096421555e-05,
+      "loss": 3.25,
+      "step": 2238
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.980157205868226e-05,
+      "loss": 5.9062,
+      "step": 2239
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.980127292897241e-05,
+      "loss": 3.9688,
+      "step": 2240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.98009735750894e-05,
+      "loss": 4.625,
+      "step": 2241
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.980067399703661e-05,
+      "loss": 5.0,
+      "step": 2242
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9800374194817444e-05,
+      "loss": 5.6875,
+      "step": 2243
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9800074168435274e-05,
+      "loss": 5.9375,
+      "step": 2244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9799773917893526e-05,
+      "loss": 8.8281,
+      "step": 2245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979947344319557e-05,
+      "loss": 3.3438,
+      "step": 2246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9799172744344826e-05,
+      "loss": 5.4688,
+      "step": 2247
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979887182134469e-05,
+      "loss": 3.3594,
+      "step": 2248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979857067419858e-05,
+      "loss": 3.6719,
+      "step": 2249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979826930290989e-05,
+      "loss": 3.2812,
+      "step": 2250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979796770748204e-05,
+      "loss": 3.375,
+      "step": 2251
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9797665887918446e-05,
+      "loss": 4.7969,
+      "step": 2252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979736384422252e-05,
+      "loss": 5.7188,
+      "step": 2253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9797061576397686e-05,
+      "loss": 3.8438,
+      "step": 2254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979675908444736e-05,
+      "loss": 5.125,
+      "step": 2255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979645636837497e-05,
+      "loss": 5.0781,
+      "step": 2256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979615342818395e-05,
+      "loss": 4.25,
+      "step": 2257
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9795850263877725e-05,
+      "loss": 5.2344,
+      "step": 2258
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979554687545972e-05,
+      "loss": 6.4688,
+      "step": 2259
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979524326293338e-05,
+      "loss": 3.6094,
+      "step": 2260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979493942630213e-05,
+      "loss": 3.2188,
+      "step": 2261
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979463536556942e-05,
+      "loss": 3.9375,
+      "step": 2262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.97943310807387e-05,
+      "loss": 4.0312,
+      "step": 2263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9794026571813386e-05,
+      "loss": 3.4844,
+      "step": 2264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979372183879696e-05,
+      "loss": 3.4375,
+      "step": 2265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.9793416881692846e-05,
+      "loss": 4.8125,
+      "step": 2266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.979311170050451e-05,
+      "loss": 7.9062,
+      "step": 2267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.97928062952354e-05,
+      "loss": 3.7188,
+      "step": 2268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.979250066588898e-05,
+      "loss": 2.9688,
+      "step": 2269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.97921948124687e-05,
+      "loss": 5.4688,
+      "step": 2270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.979188873497804e-05,
+      "loss": 4.3438,
+      "step": 2271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.979158243342045e-05,
+      "loss": 3.0781,
+      "step": 2272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.97912759077994e-05,
+      "loss": 3.375,
+      "step": 2273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.9790969158118365e-05,
+      "loss": 5.4844,
+      "step": 2274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.979066218438082e-05,
+      "loss": 4.25,
+      "step": 2275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.979035498659022e-05,
+      "loss": 5.0781,
+      "step": 2276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.979004756475007e-05,
+      "loss": 4.7188,
+      "step": 2277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.9789739918863845e-05,
+      "loss": 7.3438,
+      "step": 2278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.978943204893501e-05,
+      "loss": 5.9844,
+      "step": 2279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.9789123954967074e-05,
+      "loss": 3.6875,
+      "step": 2280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.97888156369635e-05,
+      "loss": 6.3125,
+      "step": 2281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.9788507094927795e-05,
+      "loss": 3.5469,
+      "step": 2282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.978819832886345e-05,
+      "loss": 5.0312,
+      "step": 2283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.978788933877396e-05,
+      "loss": 2.5625,
+      "step": 2284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.978758012466282e-05,
+      "loss": 5.6406,
+      "step": 2285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.9787270686533535e-05,
+      "loss": 6.0156,
+      "step": 2286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.978696102438961e-05,
+      "loss": 4.1562,
+      "step": 2287
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.9786651138234546e-05,
+      "loss": 5.9062,
+      "step": 2288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.978634102807185e-05,
+      "loss": 4.1562,
+      "step": 2289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.9786030693905034e-05,
+      "loss": 4.0156,
+      "step": 2290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.978572013573761e-05,
+      "loss": 4.125,
+      "step": 2291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.97854093535731e-05,
+      "loss": 5.0938,
+      "step": 2292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.9785098347415016e-05,
+      "loss": 4.0312,
+      "step": 2293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.9784787117266884e-05,
+      "loss": 4.3281,
+      "step": 2294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.9784475663132225e-05,
+      "loss": 4.8438,
+      "step": 2295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.978416398501456e-05,
+      "loss": 5.2031,
+      "step": 2296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.978385208291743e-05,
+      "loss": 4.0625,
+      "step": 2297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.9783539956844355e-05,
+      "loss": 4.25,
+      "step": 2298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.9783227606798875e-05,
+      "loss": 4.5,
+      "step": 2299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.9782915032784514e-05,
+      "loss": 3.9062,
+      "step": 2300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.978260223480483e-05,
+      "loss": 5.3438,
+      "step": 2301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.9782289212863354e-05,
+      "loss": 6.0156,
+      "step": 2302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.978197596696363e-05,
+      "loss": 5.5156,
+      "step": 2303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.97816624971092e-05,
+      "loss": 3.0781,
+      "step": 2304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.978134880330363e-05,
+      "loss": 3.5312,
+      "step": 2305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.978103488555044e-05,
+      "loss": 3.7344,
+      "step": 2306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.978072074385321e-05,
+      "loss": 7.5781,
+      "step": 2307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.978040637821549e-05,
+      "loss": 7.5625,
+      "step": 2308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.978009178864084e-05,
+      "loss": 5.7656,
+      "step": 2309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.977977697513282e-05,
+      "loss": 10.375,
+      "step": 2310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.977946193769499e-05,
+      "loss": 4.2656,
+      "step": 2311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.9779146676330925e-05,
+      "loss": 8.9375,
+      "step": 2312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.9778831191044186e-05,
+      "loss": 2.9062,
+      "step": 2313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.977851548183835e-05,
+      "loss": 5.375,
+      "step": 2314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.9778199548716985e-05,
+      "loss": 5.7812,
+      "step": 2315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.977788339168367e-05,
+      "loss": 4.9844,
+      "step": 2316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.977756701074199e-05,
+      "loss": 5.1719,
+      "step": 2317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.977725040589553e-05,
+      "loss": 5.4062,
+      "step": 2318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.977693357714786e-05,
+      "loss": 5.8906,
+      "step": 2319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.977661652450257e-05,
+      "loss": 3.0938,
+      "step": 2320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.977629924796326e-05,
+      "loss": 3.2344,
+      "step": 2321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.977598174753352e-05,
+      "loss": 4.6094,
+      "step": 2322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.977566402321693e-05,
+      "loss": 3.75,
+      "step": 2323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.97753460750171e-05,
+      "loss": 4.5,
+      "step": 2324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.977502790293763e-05,
+      "loss": 5.6875,
+      "step": 2325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.977470950698211e-05,
+      "loss": 3.2969,
+      "step": 2326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.977439088715416e-05,
+      "loss": 5.3281,
+      "step": 2327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.9774072043457384e-05,
+      "loss": 3.5625,
+      "step": 2328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.977375297589539e-05,
+      "loss": 5.375,
+      "step": 2329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.977343368447177e-05,
+      "loss": 4.0625,
+      "step": 2330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.9773114169190175e-05,
+      "loss": 5.3438,
+      "step": 2331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.9772794430054196e-05,
+      "loss": 5.8906,
+      "step": 2332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.977247446706747e-05,
+      "loss": 7.25,
+      "step": 2333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.977215428023359e-05,
+      "loss": 4.75,
+      "step": 2334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.9771833869556225e-05,
+      "loss": 3.9844,
+      "step": 2335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.9771513235038964e-05,
+      "loss": 2.75,
+      "step": 2336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.977119237668546e-05,
+      "loss": 3.5625,
+      "step": 2337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.977087129449933e-05,
+      "loss": 4.75,
+      "step": 2338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.977054998848422e-05,
+      "loss": 6.2969,
+      "step": 2339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.977022845864376e-05,
+      "loss": 3.625,
+      "step": 2340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.9769906704981595e-05,
+      "loss": 4.3906,
+      "step": 2341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.976958472750137e-05,
+      "loss": 6.625,
+      "step": 2342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.976926252620673e-05,
+      "loss": 9.2031,
+      "step": 2343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.976894010110131e-05,
+      "loss": 5.5312,
+      "step": 2344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.976861745218877e-05,
+      "loss": 4.5625,
+      "step": 2345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.976829457947277e-05,
+      "loss": 7.25,
+      "step": 2346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.976797148295695e-05,
+      "loss": 6.1875,
+      "step": 2347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.976764816264498e-05,
+      "loss": 2.7031,
+      "step": 2348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.976732461854052e-05,
+      "loss": 6.5,
+      "step": 2349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.976700085064722e-05,
+      "loss": 3.2656,
+      "step": 2350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.9766676858968754e-05,
+      "loss": 6.5312,
+      "step": 2351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.9766352643508794e-05,
+      "loss": 5.0156,
+      "step": 2352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.9766028204271e-05,
+      "loss": 8.25,
+      "step": 2353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.976570354125907e-05,
+      "loss": 1.9062,
+      "step": 2354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.9765378654476646e-05,
+      "loss": 4.0625,
+      "step": 2355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.976505354392742e-05,
+      "loss": 4.25,
+      "step": 2356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.976472820961507e-05,
+      "loss": 4.625,
+      "step": 2357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.97644026515433e-05,
+      "loss": 5.5625,
+      "step": 2358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.9764076869715767e-05,
+      "loss": 4.25,
+      "step": 2359
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.976375086413617e-05,
+      "loss": 5.4062,
+      "step": 2360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9763424634808204e-05,
+      "loss": 5.7344,
+      "step": 2361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9763098181735554e-05,
+      "loss": 5.0469,
+      "step": 2362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.976277150492193e-05,
+      "loss": 5.8594,
+      "step": 2363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9762444604371016e-05,
+      "loss": 4.875,
+      "step": 2364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9762117480086514e-05,
+      "loss": 3.7656,
+      "step": 2365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.976179013207214e-05,
+      "loss": 6.1875,
+      "step": 2366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9761462560331585e-05,
+      "loss": 10.2344,
+      "step": 2367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9761134764868564e-05,
+      "loss": 3.75,
+      "step": 2368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9760806745686794e-05,
+      "loss": 4.5312,
+      "step": 2369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9760478502789975e-05,
+      "loss": 5.8906,
+      "step": 2370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.976015003618183e-05,
+      "loss": 7.5156,
+      "step": 2371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.975982134586608e-05,
+      "loss": 4.6719,
+      "step": 2372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.975949243184645e-05,
+      "loss": 5.6562,
+      "step": 2373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.975916329412665e-05,
+      "loss": 4.3281,
+      "step": 2374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.975883393271042e-05,
+      "loss": 4.75,
+      "step": 2375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.975850434760148e-05,
+      "loss": 6.5625,
+      "step": 2376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9758174538803564e-05,
+      "loss": 6.8438,
+      "step": 2377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.97578445063204e-05,
+      "loss": 6.75,
+      "step": 2378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.975751425015574e-05,
+      "loss": 5.8594,
+      "step": 2379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.975718377031332e-05,
+      "loss": 6.2812,
+      "step": 2380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.975685306679686e-05,
+      "loss": 7.375,
+      "step": 2381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.975652213961012e-05,
+      "loss": 7.1562,
+      "step": 2382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.975619098875685e-05,
+      "loss": 4.7188,
+      "step": 2383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.975585961424079e-05,
+      "loss": 5.0469,
+      "step": 2384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.97555280160657e-05,
+      "loss": 3.9531,
+      "step": 2385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.975519619423533e-05,
+      "loss": 4.0,
+      "step": 2386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9754864148753435e-05,
+      "loss": 4.4375,
+      "step": 2387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.975453187962377e-05,
+      "loss": 6.8125,
+      "step": 2388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.9754199386850115e-05,
+      "loss": 3.5938,
+      "step": 2389
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.975386667043621e-05,
+      "loss": 5.6719,
+      "step": 2390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.975353373038583e-05,
+      "loss": 2.6562,
+      "step": 2391
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9753200566702756e-05,
+      "loss": 7.5938,
+      "step": 2392
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.975286717939075e-05,
+      "loss": 7.8906,
+      "step": 2393
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9752533568453595e-05,
+      "loss": 5.8125,
+      "step": 2394
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.975219973389505e-05,
+      "loss": 6.9375,
+      "step": 2395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.97518656757189e-05,
+      "loss": 5.5781,
+      "step": 2396
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.975153139392894e-05,
+      "loss": 5.1719,
+      "step": 2397
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9751196888528945e-05,
+      "loss": 4.5312,
+      "step": 2398
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9750862159522706e-05,
+      "loss": 5.4688,
+      "step": 2399
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9750527206914e-05,
+      "loss": 3.9531,
+      "step": 2400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9750192030706634e-05,
+      "loss": 5.0312,
+      "step": 2401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9749856630904397e-05,
+      "loss": 2.8281,
+      "step": 2402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.974952100751109e-05,
+      "loss": 3.7812,
+      "step": 2403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9749185160530505e-05,
+      "loss": 4.2188,
+      "step": 2404
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.974884908996645e-05,
+      "loss": 6.9219,
+      "step": 2405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9748512795822725e-05,
+      "loss": 5.5625,
+      "step": 2406
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9748176278103144e-05,
+      "loss": 7.0312,
+      "step": 2407
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.974783953681151e-05,
+      "loss": 3.0625,
+      "step": 2408
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9747502571951636e-05,
+      "loss": 5.6328,
+      "step": 2409
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.974716538352734e-05,
+      "loss": 5.5938,
+      "step": 2410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.974682797154244e-05,
+      "loss": 4.75,
+      "step": 2411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9746490336000754e-05,
+      "loss": 4.7344,
+      "step": 2412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.97461524769061e-05,
+      "loss": 2.8125,
+      "step": 2413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.974581439426232e-05,
+      "loss": 3.9219,
+      "step": 2414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.974547608807322e-05,
+      "loss": 3.9688,
+      "step": 2415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.974513755834264e-05,
+      "loss": 3.5,
+      "step": 2416
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.974479880507441e-05,
+      "loss": 2.75,
+      "step": 2417
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.9744459828272374e-05,
+      "loss": 6.5,
+      "step": 2418
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.974412062794035e-05,
+      "loss": 3.7031,
+      "step": 2419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9743781204082206e-05,
+      "loss": 7.2812,
+      "step": 2420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9743441556701754e-05,
+      "loss": 4.2188,
+      "step": 2421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.974310168580286e-05,
+      "loss": 4.0625,
+      "step": 2422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9742761591389366e-05,
+      "loss": 5.75,
+      "step": 2423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.974242127346513e-05,
+      "loss": 4.8906,
+      "step": 2424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9742080732033987e-05,
+      "loss": 3.2188,
+      "step": 2425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9741739967099806e-05,
+      "loss": 3.9688,
+      "step": 2426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.974139897866644e-05,
+      "loss": 5.0625,
+      "step": 2427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9741057766737755e-05,
+      "loss": 6.1094,
+      "step": 2428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.97407163313176e-05,
+      "loss": 4.125,
+      "step": 2429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.974037467240986e-05,
+      "loss": 7.1094,
+      "step": 2430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9740032790018384e-05,
+      "loss": 4.7188,
+      "step": 2431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973969068414706e-05,
+      "loss": 7.2812,
+      "step": 2432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973934835479975e-05,
+      "loss": 8.5,
+      "step": 2433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9739005801980336e-05,
+      "loss": 5.5469,
+      "step": 2434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973866302569268e-05,
+      "loss": 9.2656,
+      "step": 2435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973832002594068e-05,
+      "loss": 3.9062,
+      "step": 2436
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9737976802728224e-05,
+      "loss": 4.0781,
+      "step": 2437
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973763335605918e-05,
+      "loss": 2.8438,
+      "step": 2438
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973728968593744e-05,
+      "loss": 3.5,
+      "step": 2439
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.97369457923669e-05,
+      "loss": 6.9688,
+      "step": 2440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9736601675351455e-05,
+      "loss": 5.9531,
+      "step": 2441
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9736257334895e-05,
+      "loss": 7.4375,
+      "step": 2442
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9735912771001426e-05,
+      "loss": 5.5,
+      "step": 2443
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973556798367464e-05,
+      "loss": 5.8125,
+      "step": 2444
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973522297291855e-05,
+      "loss": 4.7812,
+      "step": 2445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973487773873704e-05,
+      "loss": 5.7969,
+      "step": 2446
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973453228113405e-05,
+      "loss": 4.75,
+      "step": 2447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973418660011347e-05,
+      "loss": 6.4688,
+      "step": 2448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973384069567922e-05,
+      "loss": 3.3125,
+      "step": 2449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973349456783521e-05,
+      "loss": 2.1875,
+      "step": 2450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9733148216585375e-05,
+      "loss": 3.25,
+      "step": 2451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.973280164193361e-05,
+      "loss": 3.6562,
+      "step": 2452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.973245484388387e-05,
+      "loss": 5.6406,
+      "step": 2453
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9732107822440055e-05,
+      "loss": 3.0156,
+      "step": 2454
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.973176057760611e-05,
+      "loss": 4.6719,
+      "step": 2455
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.973141310938595e-05,
+      "loss": 4.25,
+      "step": 2456
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.973106541778352e-05,
+      "loss": 4.25,
+      "step": 2457
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.973071750280276e-05,
+      "loss": 9.2344,
+      "step": 2458
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.97303693644476e-05,
+      "loss": 4.8594,
+      "step": 2459
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.973002100272198e-05,
+      "loss": 4.6875,
+      "step": 2460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.972967241762985e-05,
+      "loss": 5.6875,
+      "step": 2461
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.972932360917516e-05,
+      "loss": 4.9375,
+      "step": 2462
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.972897457736185e-05,
+      "loss": 5.9219,
+      "step": 2463
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.972862532219388e-05,
+      "loss": 4.9062,
+      "step": 2464
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9728275843675196e-05,
+      "loss": 7.3438,
+      "step": 2465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9727926141809755e-05,
+      "loss": 3.5,
+      "step": 2466
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.972757621660152e-05,
+      "loss": 5.7188,
+      "step": 2467
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.972722606805445e-05,
+      "loss": 3.5312,
+      "step": 2468
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.972687569617251e-05,
+      "loss": 4.6094,
+      "step": 2469
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9726525100959675e-05,
+      "loss": 6.3125,
+      "step": 2470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9726174282419895e-05,
+      "loss": 3.7812,
+      "step": 2471
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.972582324055716e-05,
+      "loss": 5.2031,
+      "step": 2472
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.972547197537544e-05,
+      "loss": 4.5469,
+      "step": 2473
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.97251204868787e-05,
+      "loss": 4.5,
+      "step": 2474
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.972476877507093e-05,
+      "loss": 9.0625,
+      "step": 2475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.972441683995611e-05,
+      "loss": 4.2969,
+      "step": 2476
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.972406468153822e-05,
+      "loss": 5.3281,
+      "step": 2477
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.972371229982126e-05,
+      "loss": 3.9688,
+      "step": 2478
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.97233596948092e-05,
+      "loss": 4.9844,
+      "step": 2479
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9723006866506054e-05,
+      "loss": 3.9219,
+      "step": 2480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.97226538149158e-05,
+      "loss": 4.3125,
+      "step": 2481
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9722300540042434e-05,
+      "loss": 3.9844,
+      "step": 2482
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9721947041889965e-05,
+      "loss": 6.7344,
+      "step": 2483
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.972159332046239e-05,
+      "loss": 5.5469,
+      "step": 2484
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.972123937576371e-05,
+      "loss": 5.2344,
+      "step": 2485
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.972088520779795e-05,
+      "loss": 2.9688,
+      "step": 2486
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.972053081656909e-05,
+      "loss": 8.25,
+      "step": 2487
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.972017620208117e-05,
+      "loss": 10.1719,
+      "step": 2488
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971982136433819e-05,
+      "loss": 7.6875,
+      "step": 2489
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971946630334417e-05,
+      "loss": 3.8438,
+      "step": 2490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971911101910313e-05,
+      "loss": 3.9062,
+      "step": 2491
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971875551161909e-05,
+      "loss": 8.4219,
+      "step": 2492
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971839978089607e-05,
+      "loss": 5.0625,
+      "step": 2493
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971804382693812e-05,
+      "loss": 3.9531,
+      "step": 2494
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9717687649749235e-05,
+      "loss": 6.8906,
+      "step": 2495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971733124933348e-05,
+      "loss": 7.5625,
+      "step": 2496
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971697462569487e-05,
+      "loss": 3.25,
+      "step": 2497
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9716617778837456e-05,
+      "loss": 4.9844,
+      "step": 2498
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971626070876526e-05,
+      "loss": 4.7969,
+      "step": 2499
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9715903415482344e-05,
+      "loss": 4.7969,
+      "step": 2500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971554589899274e-05,
+      "loss": 6.6875,
+      "step": 2501
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.97151881593005e-05,
+      "loss": 4.6875,
+      "step": 2502
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9714830196409664e-05,
+      "loss": 3.7031,
+      "step": 2503
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9714472010324306e-05,
+      "loss": 2.7188,
+      "step": 2504
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971411360104846e-05,
+      "loss": 5.5625,
+      "step": 2505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.97137549685862e-05,
+      "loss": 3.6875,
+      "step": 2506
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971339611294157e-05,
+      "loss": 6.375,
+      "step": 2507
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971303703411865e-05,
+      "loss": 4.5469,
+      "step": 2508
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.97126777321215e-05,
+      "loss": 5.7188,
+      "step": 2509
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971231820695417e-05,
+      "loss": 4.4062,
+      "step": 2510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971195845862076e-05,
+      "loss": 4.6875,
+      "step": 2511
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9711598487125316e-05,
+      "loss": 3.6875,
+      "step": 2512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.971123829247192e-05,
+      "loss": 6.375,
+      "step": 2513
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.971087787466466e-05,
+      "loss": 5.0312,
+      "step": 2514
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.971051723370762e-05,
+      "loss": 6.375,
+      "step": 2515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9710156369604864e-05,
+      "loss": 6.1094,
+      "step": 2516
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.970979528236049e-05,
+      "loss": 4.5,
+      "step": 2517
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.970943397197858e-05,
+      "loss": 8.875,
+      "step": 2518
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.970907243846322e-05,
+      "loss": 7.2344,
+      "step": 2519
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.970871068181853e-05,
+      "loss": 3.625,
+      "step": 2520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9708348702048566e-05,
+      "loss": 1.5938,
+      "step": 2521
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.970798649915746e-05,
+      "loss": 3.8125,
+      "step": 2522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9707624073149296e-05,
+      "loss": 2.6719,
+      "step": 2523
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9707261424028175e-05,
+      "loss": 4.0938,
+      "step": 2524
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.970689855179821e-05,
+      "loss": 4.4531,
+      "step": 2525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.97065354564635e-05,
+      "loss": 5.4375,
+      "step": 2526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9706172138028165e-05,
+      "loss": 3.0,
+      "step": 2527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.970580859649632e-05,
+      "loss": 3.125,
+      "step": 2528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.970544483187207e-05,
+      "loss": 4.8438,
+      "step": 2529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.970508084415954e-05,
+      "loss": 3.9844,
+      "step": 2530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9704716633362845e-05,
+      "loss": 2.8438,
+      "step": 2531
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.970435219948612e-05,
+      "loss": 6.1406,
+      "step": 2532
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.970398754253348e-05,
+      "loss": 3.375,
+      "step": 2533
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.970362266250906e-05,
+      "loss": 3.6562,
+      "step": 2534
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9703257559416986e-05,
+      "loss": 4.125,
+      "step": 2535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.97028922332614e-05,
+      "loss": 4.75,
+      "step": 2536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9702526684046426e-05,
+      "loss": 4.0781,
+      "step": 2537
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.97021609117762e-05,
+      "loss": 3.8438,
+      "step": 2538
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9701794916454883e-05,
+      "loss": 4.5469,
+      "step": 2539
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.9701428698086604e-05,
+      "loss": 3.875,
+      "step": 2540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.97010622566755e-05,
+      "loss": 3.5312,
+      "step": 2541
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.970069559222574e-05,
+      "loss": 3.3281,
+      "step": 2542
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.970032870474147e-05,
+      "loss": 5.0312,
+      "step": 2543
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969996159422683e-05,
+      "loss": 4.9219,
+      "step": 2544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969959426068599e-05,
+      "loss": 4.5312,
+      "step": 2545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9699226704123106e-05,
+      "loss": 6.75,
+      "step": 2546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969885892454233e-05,
+      "loss": 9.4688,
+      "step": 2547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969849092194784e-05,
+      "loss": 6.5938,
+      "step": 2548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969812269634379e-05,
+      "loss": 4.7656,
+      "step": 2549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969775424773436e-05,
+      "loss": 2.9219,
+      "step": 2550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969738557612371e-05,
+      "loss": 5.4375,
+      "step": 2551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969701668151603e-05,
+      "loss": 5.4062,
+      "step": 2552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969664756391547e-05,
+      "loss": 3.5469,
+      "step": 2553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969627822332623e-05,
+      "loss": 7.5781,
+      "step": 2554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969590865975249e-05,
+      "loss": 4.5938,
+      "step": 2555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969553887319842e-05,
+      "loss": 5.9375,
+      "step": 2556
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969516886366822e-05,
+      "loss": 6.2031,
+      "step": 2557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969479863116608e-05,
+      "loss": 3.4844,
+      "step": 2558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9694428175696176e-05,
+      "loss": 7.5156,
+      "step": 2559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969405749726272e-05,
+      "loss": 9.4375,
+      "step": 2560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.96936865958699e-05,
+      "loss": 4.3125,
+      "step": 2561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9693315471521906e-05,
+      "loss": 4.875,
+      "step": 2562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969294412422296e-05,
+      "loss": 6.3281,
+      "step": 2563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969257255397725e-05,
+      "loss": 4.4375,
+      "step": 2564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9692200760788994e-05,
+      "loss": 5.9062,
+      "step": 2565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969182874466239e-05,
+      "loss": 2.5625,
+      "step": 2566
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.9691456505601655e-05,
+      "loss": 2.8125,
+      "step": 2567
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969108404361101e-05,
+      "loss": 2.2188,
+      "step": 2568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969071135869466e-05,
+      "loss": 7.5156,
+      "step": 2569
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.969033845085682e-05,
+      "loss": 5.5781,
+      "step": 2570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.968996532010173e-05,
+      "loss": 4.4062,
+      "step": 2571
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.968959196643361e-05,
+      "loss": 5.0312,
+      "step": 2572
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968921838985667e-05,
+      "loss": 6.4688,
+      "step": 2573
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968884459037515e-05,
+      "loss": 4.2344,
+      "step": 2574
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968847056799329e-05,
+      "loss": 9.2188,
+      "step": 2575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968809632271531e-05,
+      "loss": 3.75,
+      "step": 2576
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968772185454546e-05,
+      "loss": 12.1719,
+      "step": 2577
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968734716348797e-05,
+      "loss": 3.6562,
+      "step": 2578
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968697224954709e-05,
+      "loss": 4.4688,
+      "step": 2579
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9686597112727053e-05,
+      "loss": 6.9375,
+      "step": 2580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968622175303212e-05,
+      "loss": 6.5,
+      "step": 2581
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968584617046653e-05,
+      "loss": 7.7812,
+      "step": 2582
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968547036503454e-05,
+      "loss": 5.75,
+      "step": 2583
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.96850943367404e-05,
+      "loss": 4.6562,
+      "step": 2584
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968471808558837e-05,
+      "loss": 5.6562,
+      "step": 2585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968434161158271e-05,
+      "loss": 7.0312,
+      "step": 2586
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968396491472768e-05,
+      "loss": 3.8438,
+      "step": 2587
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968358799502755e-05,
+      "loss": 5.5,
+      "step": 2588
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968321085248658e-05,
+      "loss": 6.625,
+      "step": 2589
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968283348710905e-05,
+      "loss": 5.0312,
+      "step": 2590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9682455898899216e-05,
+      "loss": 5.375,
+      "step": 2591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968207808786137e-05,
+      "loss": 2.4688,
+      "step": 2592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968170005399978e-05,
+      "loss": 6.0625,
+      "step": 2593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968132179731872e-05,
+      "loss": 6.75,
+      "step": 2594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968094331782248e-05,
+      "loss": 5.3438,
+      "step": 2595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.968056461551535e-05,
+      "loss": 6.0625,
+      "step": 2596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9680185690401606e-05,
+      "loss": 5.0781,
+      "step": 2597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.967980654248555e-05,
+      "loss": 4.6562,
+      "step": 2598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.967942717177147e-05,
+      "loss": 7.5625,
+      "step": 2599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.967904757826365e-05,
+      "loss": 6.9375,
+      "step": 2600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9678667761966404e-05,
+      "loss": 6.1719,
+      "step": 2601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.967828772288401e-05,
+      "loss": 7.2969,
+      "step": 2602
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9677907461020796e-05,
+      "loss": 3.5625,
+      "step": 2603
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.967752697638106e-05,
+      "loss": 3.75,
+      "step": 2604
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.96771462689691e-05,
+      "loss": 7.3125,
+      "step": 2605
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.967676533878923e-05,
+      "loss": 3.625,
+      "step": 2606
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.967638418584576e-05,
+      "loss": 6.5,
+      "step": 2607
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.967600281014301e-05,
+      "loss": 4.0,
+      "step": 2608
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9675621211685305e-05,
+      "loss": 6.5469,
+      "step": 2609
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9675239390476946e-05,
+      "loss": 4.6562,
+      "step": 2610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9674857346522274e-05,
+      "loss": 5.875,
+      "step": 2611
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9674475079825604e-05,
+      "loss": 6.7812,
+      "step": 2612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9674092590391266e-05,
+      "loss": 4.5156,
+      "step": 2613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.967370987822359e-05,
+      "loss": 2.4375,
+      "step": 2614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.967332694332691e-05,
+      "loss": 5.4062,
+      "step": 2615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9672943785705555e-05,
+      "loss": 3.0938,
+      "step": 2616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9672560405363874e-05,
+      "loss": 5.0156,
+      "step": 2617
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9672176802306196e-05,
+      "loss": 4.8125,
+      "step": 2618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.967179297653688e-05,
+      "loss": 5.7188,
+      "step": 2619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.967140892806025e-05,
+      "loss": 7.875,
+      "step": 2620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.967102465688067e-05,
+      "loss": 8.7656,
+      "step": 2621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.967064016300248e-05,
+      "loss": 7.7188,
+      "step": 2622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9670255446430045e-05,
+      "loss": 8.3906,
+      "step": 2623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.96698705071677e-05,
+      "loss": 4.8438,
+      "step": 2624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.966948534521983e-05,
+      "loss": 4.6094,
+      "step": 2625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.966909996059078e-05,
+      "loss": 5.9062,
+      "step": 2626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9668714353284906e-05,
+      "loss": 4.0625,
+      "step": 2627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.966832852330659e-05,
+      "loss": 4.5,
+      "step": 2628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.966794247066019e-05,
+      "loss": 4.2344,
+      "step": 2629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9667556195350076e-05,
+      "loss": 4.75,
+      "step": 2630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9667169697380624e-05,
+      "loss": 3.8594,
+      "step": 2631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9666782976756214e-05,
+      "loss": 6.7812,
+      "step": 2632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.966639603348122e-05,
+      "loss": 3.6875,
+      "step": 2633
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9666008867560016e-05,
+      "loss": 4.8438,
+      "step": 2634
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9665621478996995e-05,
+      "loss": 5.0469,
+      "step": 2635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9665233867796535e-05,
+      "loss": 5.9688,
+      "step": 2636
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.966484603396303e-05,
+      "loss": 4.9688,
+      "step": 2637
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.966445797750087e-05,
+      "loss": 7.0156,
+      "step": 2638
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.966406969841445e-05,
+      "loss": 5.2188,
+      "step": 2639
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.966368119670816e-05,
+      "loss": 2.2188,
+      "step": 2640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.966329247238641e-05,
+      "loss": 4.4531,
+      "step": 2641
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.966290352545357e-05,
+      "loss": 2.6562,
+      "step": 2642
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.966251435591409e-05,
+      "loss": 6.8906,
+      "step": 2643
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.966212496377234e-05,
+      "loss": 3.7344,
+      "step": 2644
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.966173534903274e-05,
+      "loss": 4.4844,
+      "step": 2645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.96613455116997e-05,
+      "loss": 3.625,
+      "step": 2646
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9660955451777635e-05,
+      "loss": 7.0781,
+      "step": 2647
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9660565169270965e-05,
+      "loss": 6.1562,
+      "step": 2648
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.966017466418409e-05,
+      "loss": 5.375,
+      "step": 2649
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.965978393652145e-05,
+      "loss": 3.4844,
+      "step": 2650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.965939298628747e-05,
+      "loss": 3.0625,
+      "step": 2651
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.965900181348656e-05,
+      "loss": 6.7656,
+      "step": 2652
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.965861041812316e-05,
+      "loss": 5.0,
+      "step": 2653
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.96582188002017e-05,
+      "loss": 7.2812,
+      "step": 2654
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.965782695972661e-05,
+      "loss": 4.7969,
+      "step": 2655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9657434896702326e-05,
+      "loss": 7.3594,
+      "step": 2656
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9657042611133294e-05,
+      "loss": 4.0938,
+      "step": 2657
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9656650103023946e-05,
+      "loss": 6.5312,
+      "step": 2658
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.965625737237873e-05,
+      "loss": 3.6406,
+      "step": 2659
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.965586441920209e-05,
+      "loss": 4.9531,
+      "step": 2660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.965547124349848e-05,
+      "loss": 5.7031,
+      "step": 2661
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.965507784527235e-05,
+      "loss": 3.4062,
+      "step": 2662
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.965468422452814e-05,
+      "loss": 3.9219,
+      "step": 2663
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.965429038127033e-05,
+      "loss": 3.7031,
+      "step": 2664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.965389631550335e-05,
+      "loss": 6.2812,
+      "step": 2665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.965350202723169e-05,
+      "loss": 3.8125,
+      "step": 2666
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.96531075164598e-05,
+      "loss": 3.6562,
+      "step": 2667
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.965271278319215e-05,
+      "loss": 7.0469,
+      "step": 2668
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.96523178274332e-05,
+      "loss": 5.7812,
+      "step": 2669
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.965192264918743e-05,
+      "loss": 3.125,
+      "step": 2670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.965152724845931e-05,
+      "loss": 9.0469,
+      "step": 2671
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.965113162525332e-05,
+      "loss": 4.4062,
+      "step": 2672
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9650735779573934e-05,
+      "loss": 3.8594,
+      "step": 2673
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.965033971142564e-05,
+      "loss": 3.9688,
+      "step": 2674
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.964994342081291e-05,
+      "loss": 2.9688,
+      "step": 2675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9649546907740245e-05,
+      "loss": 4.6875,
+      "step": 2676
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9649150172212135e-05,
+      "loss": 6.4219,
+      "step": 2677
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9648753214233054e-05,
+      "loss": 2.3125,
+      "step": 2678
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.964835603380751e-05,
+      "loss": 3.2188,
+      "step": 2679
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.964795863093999e-05,
+      "loss": 1.9688,
+      "step": 2680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9647561005635e-05,
+      "loss": 3.9844,
+      "step": 2681
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.964716315789705e-05,
+      "loss": 5.0938,
+      "step": 2682
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.964676508773063e-05,
+      "loss": 4.125,
+      "step": 2683
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.964636679514025e-05,
+      "loss": 5.7812,
+      "step": 2684
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9645968280130416e-05,
+      "loss": 6.2188,
+      "step": 2685
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9645569542705645e-05,
+      "loss": 6.0156,
+      "step": 2686
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.964517058287046e-05,
+      "loss": 5.375,
+      "step": 2687
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9644771400629356e-05,
+      "loss": 3.6875,
+      "step": 2688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9644371995986866e-05,
+      "loss": 5.6875,
+      "step": 2689
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.964397236894751e-05,
+      "loss": 3.0781,
+      "step": 2690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.964357251951581e-05,
+      "loss": 3.8438,
+      "step": 2691
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.96431724476963e-05,
+      "loss": 4.3438,
+      "step": 2692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9642772153493506e-05,
+      "loss": 3.9688,
+      "step": 2693
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.964237163691195e-05,
+      "loss": 2.5156,
+      "step": 2694
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.964197089795618e-05,
+      "loss": 6.1562,
+      "step": 2695
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.964156993663072e-05,
+      "loss": 3.8906,
+      "step": 2696
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9641168752940125e-05,
+      "loss": 6.4688,
+      "step": 2697
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.964076734688892e-05,
+      "loss": 6.25,
+      "step": 2698
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9640365718481655e-05,
+      "loss": 8.7188,
+      "step": 2699
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963996386772289e-05,
+      "loss": 4.8594,
+      "step": 2700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963956179461716e-05,
+      "loss": 5.0938,
+      "step": 2701
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963915949916902e-05,
+      "loss": 2.4688,
+      "step": 2702
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963875698138302e-05,
+      "loss": 3.0312,
+      "step": 2703
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9638354241263726e-05,
+      "loss": 4.75,
+      "step": 2704
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9637951278815696e-05,
+      "loss": 6.5,
+      "step": 2705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9637548094043486e-05,
+      "loss": 3.7812,
+      "step": 2706
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963714468695166e-05,
+      "loss": 10.2969,
+      "step": 2707
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.96367410575448e-05,
+      "loss": 3.8125,
+      "step": 2708
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9636337205827454e-05,
+      "loss": 3.6875,
+      "step": 2709
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963593313180421e-05,
+      "loss": 7.375,
+      "step": 2710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963552883547964e-05,
+      "loss": 5.0781,
+      "step": 2711
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963512431685831e-05,
+      "loss": 5.6719,
+      "step": 2712
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963471957594482e-05,
+      "loss": 4.3438,
+      "step": 2713
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963431461274373e-05,
+      "loss": 5.7344,
+      "step": 2714
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963390942725964e-05,
+      "loss": 4.5938,
+      "step": 2715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963350401949712e-05,
+      "loss": 5.3438,
+      "step": 2716
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963309838946079e-05,
+      "loss": 5.6875,
+      "step": 2717
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9632692537155216e-05,
+      "loss": 5.8438,
+      "step": 2718
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963228646258499e-05,
+      "loss": 6.5156,
+      "step": 2719
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963188016575473e-05,
+      "loss": 5.6562,
+      "step": 2720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9631473646669026e-05,
+      "loss": 5.8906,
+      "step": 2721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963106690533248e-05,
+      "loss": 6.6875,
+      "step": 2722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.96306599417497e-05,
+      "loss": 2.9219,
+      "step": 2723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.963025275592528e-05,
+      "loss": 3.0312,
+      "step": 2724
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962984534786384e-05,
+      "loss": 4.25,
+      "step": 2725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9629437717570004e-05,
+      "loss": 2.9688,
+      "step": 2726
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962902986504836e-05,
+      "loss": 7.4766,
+      "step": 2727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9628621790303545e-05,
+      "loss": 4.1719,
+      "step": 2728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962821349334018e-05,
+      "loss": 4.6562,
+      "step": 2729
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962780497416287e-05,
+      "loss": 4.9688,
+      "step": 2730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9627396232776263e-05,
+      "loss": 4.3438,
+      "step": 2731
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962698726918497e-05,
+      "loss": 2.5625,
+      "step": 2732
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962657808339363e-05,
+      "loss": 4.5469,
+      "step": 2733
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962616867540686e-05,
+      "loss": 2.0312,
+      "step": 2734
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962575904522932e-05,
+      "loss": 6.0781,
+      "step": 2735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962534919286562e-05,
+      "loss": 4.9375,
+      "step": 2736
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962493911832043e-05,
+      "loss": 3.0625,
+      "step": 2737
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962452882159836e-05,
+      "loss": 3.7656,
+      "step": 2738
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962411830270408e-05,
+      "loss": 6.4375,
+      "step": 2739
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962370756164223e-05,
+      "loss": 4.0625,
+      "step": 2740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9623296598417454e-05,
+      "loss": 3.875,
+      "step": 2741
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962288541303441e-05,
+      "loss": 2.8594,
+      "step": 2742
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962247400549775e-05,
+      "loss": 2.8125,
+      "step": 2743
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962206237581214e-05,
+      "loss": 8.1406,
+      "step": 2744
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962165052398223e-05,
+      "loss": 5.2188,
+      "step": 2745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9621238450012684e-05,
+      "loss": 5.5156,
+      "step": 2746
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962082615390817e-05,
+      "loss": 7.8438,
+      "step": 2747
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9620413635673365e-05,
+      "loss": 2.875,
+      "step": 2748
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.962000089531293e-05,
+      "loss": 3.75,
+      "step": 2749
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.9619587932831524e-05,
+      "loss": 5.4688,
+      "step": 2750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.961917474823385e-05,
+      "loss": 4.375,
+      "step": 2751
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.961876134152456e-05,
+      "loss": 4.875,
+      "step": 2752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.961834771270835e-05,
+      "loss": 2.9062,
+      "step": 2753
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.96179338617899e-05,
+      "loss": 2.6562,
+      "step": 2754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9617519788773896e-05,
+      "loss": 3.6562,
+      "step": 2755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9617105493665016e-05,
+      "loss": 3.8594,
+      "step": 2756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961669097646796e-05,
+      "loss": 5.0625,
+      "step": 2757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9616276237187416e-05,
+      "loss": 3.8281,
+      "step": 2758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961586127582809e-05,
+      "loss": 3.8594,
+      "step": 2759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9615446092394665e-05,
+      "loss": 6.5938,
+      "step": 2760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9615030686891856e-05,
+      "loss": 3.2188,
+      "step": 2761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961461505932435e-05,
+      "loss": 5.9062,
+      "step": 2762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961419920969686e-05,
+      "loss": 4.4688,
+      "step": 2763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.96137831380141e-05,
+      "loss": 4.3125,
+      "step": 2764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961336684428077e-05,
+      "loss": 4.1875,
+      "step": 2765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9612950328501595e-05,
+      "loss": 4.4531,
+      "step": 2766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961253359068127e-05,
+      "loss": 4.1875,
+      "step": 2767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961211663082454e-05,
+      "loss": 8.25,
+      "step": 2768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.96116994489361e-05,
+      "loss": 5.5,
+      "step": 2769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961128204502069e-05,
+      "loss": 4.9062,
+      "step": 2770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961086441908303e-05,
+      "loss": 6.4062,
+      "step": 2771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9610446571127843e-05,
+      "loss": 5.375,
+      "step": 2772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961002850115987e-05,
+      "loss": 6.8125,
+      "step": 2773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9609610209183834e-05,
+      "loss": 5.7031,
+      "step": 2774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9609191695204476e-05,
+      "loss": 7.1875,
+      "step": 2775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9608772959226536e-05,
+      "loss": 5.5781,
+      "step": 2776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.960835400125475e-05,
+      "loss": 5.4375,
+      "step": 2777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.960793482129386e-05,
+      "loss": 4.4688,
+      "step": 2778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.960751541934862e-05,
+      "loss": 4.9062,
+      "step": 2779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9607095795423774e-05,
+      "loss": 2.4062,
+      "step": 2780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.960667594952406e-05,
+      "loss": 4.3125,
+      "step": 2781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.960625588165425e-05,
+      "loss": 3.625,
+      "step": 2782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.960583559181909e-05,
+      "loss": 3.3438,
+      "step": 2783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.960541508002334e-05,
+      "loss": 7.2188,
+      "step": 2784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9604994346271766e-05,
+      "loss": 7.7969,
+      "step": 2785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.960457339056912e-05,
+      "loss": 5.0938,
+      "step": 2786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9604152212920174e-05,
+      "loss": 3.1094,
+      "step": 2787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.96037308133297e-05,
+      "loss": 2.8594,
+      "step": 2788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.960330919180246e-05,
+      "loss": 9.2734,
+      "step": 2789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9602887348343226e-05,
+      "loss": 3.5469,
+      "step": 2790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.960246528295679e-05,
+      "loss": 3.5,
+      "step": 2791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.960204299564791e-05,
+      "loss": 3.5156,
+      "step": 2792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.960162048642138e-05,
+      "loss": 3.2969,
+      "step": 2793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.960119775528199e-05,
+      "loss": 3.0,
+      "step": 2794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9600774802234504e-05,
+      "loss": 2.7188,
+      "step": 2795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.960035162728372e-05,
+      "loss": 6.2344,
+      "step": 2796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.959992823043444e-05,
+      "loss": 3.4219,
+      "step": 2797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9599504611691447e-05,
+      "loss": 5.1094,
+      "step": 2798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.959908077105954e-05,
+      "loss": 4.1719,
+      "step": 2799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9598656708543505e-05,
+      "loss": 5.5781,
+      "step": 2800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9598232424148154e-05,
+      "loss": 9.0312,
+      "step": 2801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9597807917878296e-05,
+      "loss": 4.0312,
+      "step": 2802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.959738318973873e-05,
+      "loss": 7.7812,
+      "step": 2803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9596958239734264e-05,
+      "loss": 3.1094,
+      "step": 2804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9596533067869704e-05,
+      "loss": 7.3906,
+      "step": 2805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.959610767414988e-05,
+      "loss": 7.5781,
+      "step": 2806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9595682058579594e-05,
+      "loss": 4.2188,
+      "step": 2807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.959525622116366e-05,
+      "loss": 4.0156,
+      "step": 2808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.959483016190692e-05,
+      "loss": 3.2812,
+      "step": 2809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.959440388081417e-05,
+      "loss": 3.2969,
+      "step": 2810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9593977377890256e-05,
+      "loss": 7.5,
+      "step": 2811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.959355065314e-05,
+      "loss": 4.3438,
+      "step": 2812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9593123706568235e-05,
+      "loss": 7.4375,
+      "step": 2813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.95926965381798e-05,
+      "loss": 8.4844,
+      "step": 2814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9592269147979514e-05,
+      "loss": 2.4688,
+      "step": 2815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.959184153597223e-05,
+      "loss": 3.7812,
+      "step": 2816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9591413702162785e-05,
+      "loss": 3.3125,
+      "step": 2817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.959098564655602e-05,
+      "loss": 5.5,
+      "step": 2818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9590557369156775e-05,
+      "loss": 4.25,
+      "step": 2819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9590128869969924e-05,
+      "loss": 5.4531,
+      "step": 2820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.958970014900029e-05,
+      "loss": 8.2656,
+      "step": 2821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.958927120625274e-05,
+      "loss": 4.5156,
+      "step": 2822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9588842041732125e-05,
+      "loss": 2.8125,
+      "step": 2823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.958841265544331e-05,
+      "loss": 6.0,
+      "step": 2824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9587983047391156e-05,
+      "loss": 5.5938,
+      "step": 2825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9587553217580516e-05,
+      "loss": 4.0,
+      "step": 2826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.958712316601627e-05,
+      "loss": 5.5156,
+      "step": 2827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.958669289270327e-05,
+      "loss": 4.0,
+      "step": 2828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9586262397646414e-05,
+      "loss": 6.4062,
+      "step": 2829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9585831680850546e-05,
+      "loss": 4.1562,
+      "step": 2830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.958540074232057e-05,
+      "loss": 3.625,
+      "step": 2831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.958496958206133e-05,
+      "loss": 8.0312,
+      "step": 2832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.958453820007774e-05,
+      "loss": 5.9531,
+      "step": 2833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9584106596374664e-05,
+      "loss": 3.7812,
+      "step": 2834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9583674770957e-05,
+      "loss": 4.9688,
+      "step": 2835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.958324272382963e-05,
+      "loss": 6.4531,
+      "step": 2836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9582810454997446e-05,
+      "loss": 3.5625,
+      "step": 2837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9582377964465344e-05,
+      "loss": 4.5312,
+      "step": 2838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.958194525223822e-05,
+      "loss": 2.7188,
+      "step": 2839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.958151231832097e-05,
+      "loss": 4.8125,
+      "step": 2840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.95810791627185e-05,
+      "loss": 5.9219,
+      "step": 2841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.958064578543572e-05,
+      "loss": 3.7344,
+      "step": 2842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.958021218647751e-05,
+      "loss": 4.0469,
+      "step": 2843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.957977836584881e-05,
+      "loss": 3.0938,
+      "step": 2844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.9579344323554507e-05,
+      "loss": 3.2812,
+      "step": 2845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.957891005959953e-05,
+      "loss": 5.6875,
+      "step": 2846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957847557398879e-05,
+      "loss": 3.7344,
+      "step": 2847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957804086672721e-05,
+      "loss": 3.1875,
+      "step": 2848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957760593781971e-05,
+      "loss": 5.1562,
+      "step": 2849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957717078727121e-05,
+      "loss": 3.0938,
+      "step": 2850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957673541508663e-05,
+      "loss": 6.6406,
+      "step": 2851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957629982127092e-05,
+      "loss": 4.3906,
+      "step": 2852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957586400582899e-05,
+      "loss": 6.0781,
+      "step": 2853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957542796876579e-05,
+      "loss": 5.5625,
+      "step": 2854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957499171008624e-05,
+      "loss": 7.0625,
+      "step": 2855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.95745552297953e-05,
+      "loss": 5.2812,
+      "step": 2856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957411852789789e-05,
+      "loss": 2.9375,
+      "step": 2857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957368160439897e-05,
+      "loss": 3.4062,
+      "step": 2858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9573244459303476e-05,
+      "loss": 4.5,
+      "step": 2859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9572807092616365e-05,
+      "loss": 3.5312,
+      "step": 2860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9572369504342576e-05,
+      "loss": 6.6406,
+      "step": 2861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957193169448708e-05,
+      "loss": 7.2031,
+      "step": 2862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957149366305482e-05,
+      "loss": 4.4844,
+      "step": 2863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957105541005076e-05,
+      "loss": 6.7812,
+      "step": 2864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.957061693547987e-05,
+      "loss": 3.5938,
+      "step": 2865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.95701782393471e-05,
+      "loss": 4.9531,
+      "step": 2866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9569739321657416e-05,
+      "loss": 2.3438,
+      "step": 2867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9569300182415797e-05,
+      "loss": 7.5938,
+      "step": 2868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9568860821627214e-05,
+      "loss": 4.2031,
+      "step": 2869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9568421239296636e-05,
+      "loss": 4.3125,
+      "step": 2870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.956798143542904e-05,
+      "loss": 2.75,
+      "step": 2871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9567541410029414e-05,
+      "loss": 5.2188,
+      "step": 2872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9567101163102725e-05,
+      "loss": 2.7031,
+      "step": 2873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.956666069465397e-05,
+      "loss": 3.8281,
+      "step": 2874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9566220004688126e-05,
+      "loss": 4.0781,
+      "step": 2875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9565779093210185e-05,
+      "loss": 7.1875,
+      "step": 2876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.956533796022513e-05,
+      "loss": 4.5312,
+      "step": 2877
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.956489660573798e-05,
+      "loss": 7.7656,
+      "step": 2878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9564455029753713e-05,
+      "loss": 3.75,
+      "step": 2879
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9564013232277324e-05,
+      "loss": 4.7656,
+      "step": 2880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.956357121331383e-05,
+      "loss": 3.6875,
+      "step": 2881
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.956312897286823e-05,
+      "loss": 3.6562,
+      "step": 2882
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.956268651094552e-05,
+      "loss": 6.125,
+      "step": 2883
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.956224382755072e-05,
+      "loss": 3.25,
+      "step": 2884
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9561800922688844e-05,
+      "loss": 5.0625,
+      "step": 2885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.956135779636489e-05,
+      "loss": 4.9375,
+      "step": 2886
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.956091444858389e-05,
+      "loss": 7.5625,
+      "step": 2887
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9560470879350866e-05,
+      "loss": 7.6094,
+      "step": 2888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9560027088670825e-05,
+      "loss": 6.5781,
+      "step": 2889
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.955958307654881e-05,
+      "loss": 4.75,
+      "step": 2890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.955913884298982e-05,
+      "loss": 6.7344,
+      "step": 2891
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9558694387998906e-05,
+      "loss": 6.2344,
+      "step": 2892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.95582497115811e-05,
+      "loss": 4.6719,
+      "step": 2893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.955780481374142e-05,
+      "loss": 4.9062,
+      "step": 2894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.955735969448493e-05,
+      "loss": 2.6094,
+      "step": 2895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9556914353816645e-05,
+      "loss": 4.8281,
+      "step": 2896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.955646879174161e-05,
+      "loss": 5.0312,
+      "step": 2897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.955602300826488e-05,
+      "loss": 4.4062,
+      "step": 2898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9555577003391494e-05,
+      "loss": 3.5312,
+      "step": 2899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9555130777126504e-05,
+      "loss": 4.0,
+      "step": 2900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.955468432947496e-05,
+      "loss": 1.4062,
+      "step": 2901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.955423766044192e-05,
+      "loss": 5.1094,
+      "step": 2902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.955379077003243e-05,
+      "loss": 3.8125,
+      "step": 2903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.955334365825156e-05,
+      "loss": 3.1094,
+      "step": 2904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.9552896325104365e-05,
+      "loss": 5.4375,
+      "step": 2905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.955244877059593e-05,
+      "loss": 2.875,
+      "step": 2906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.955200099473129e-05,
+      "loss": 3.4531,
+      "step": 2907
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.955155299751553e-05,
+      "loss": 3.625,
+      "step": 2908
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.955110477895373e-05,
+      "loss": 4.5312,
+      "step": 2909
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9550656339050945e-05,
+      "loss": 2.4219,
+      "step": 2910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.955020767781227e-05,
+      "loss": 9.3438,
+      "step": 2911
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9549758795242775e-05,
+      "loss": 6.3125,
+      "step": 2912
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954930969134754e-05,
+      "loss": 4.9531,
+      "step": 2913
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9548860366131666e-05,
+      "loss": 5.7656,
+      "step": 2914
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9548410819600216e-05,
+      "loss": 2.9375,
+      "step": 2915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954796105175829e-05,
+      "loss": 3.0938,
+      "step": 2916
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954751106261098e-05,
+      "loss": 6.3125,
+      "step": 2917
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954706085216338e-05,
+      "loss": 3.9062,
+      "step": 2918
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9546610420420594e-05,
+      "loss": 4.4219,
+      "step": 2919
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954615976738771e-05,
+      "loss": 2.9531,
+      "step": 2920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954570889306984e-05,
+      "loss": 3.1406,
+      "step": 2921
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954525779747208e-05,
+      "loss": 1.9375,
+      "step": 2922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954480648059954e-05,
+      "loss": 4.7969,
+      "step": 2923
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954435494245733e-05,
+      "loss": 6.6719,
+      "step": 2924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9543903183050556e-05,
+      "loss": 3.8438,
+      "step": 2925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9543451202384344e-05,
+      "loss": 6.3125,
+      "step": 2926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954299900046381e-05,
+      "loss": 7.9062,
+      "step": 2927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954254657729406e-05,
+      "loss": 3.4688,
+      "step": 2928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954209393288023e-05,
+      "loss": 3.6875,
+      "step": 2929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954164106722743e-05,
+      "loss": 4.1562,
+      "step": 2930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.95411879803408e-05,
+      "loss": 4.625,
+      "step": 2931
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.954073467222547e-05,
+      "loss": 4.2188,
+      "step": 2932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9540281142886563e-05,
+      "loss": 5.5469,
+      "step": 2933
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.953982739232922e-05,
+      "loss": 4.2812,
+      "step": 2934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9539373420558576e-05,
+      "loss": 4.4688,
+      "step": 2935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.953891922757977e-05,
+      "loss": 2.75,
+      "step": 2936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9538464813397936e-05,
+      "loss": 6.5,
+      "step": 2937
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953801017801823e-05,
+      "loss": 5.625,
+      "step": 2938
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.95375553214458e-05,
+      "loss": 5.4844,
+      "step": 2939
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953710024368579e-05,
+      "loss": 3.7188,
+      "step": 2940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953664494474335e-05,
+      "loss": 3.9375,
+      "step": 2941
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9536189424623647e-05,
+      "loss": 3.4531,
+      "step": 2942
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953573368333182e-05,
+      "loss": 6.7188,
+      "step": 2943
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953527772087304e-05,
+      "loss": 4.75,
+      "step": 2944
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9534821537252463e-05,
+      "loss": 6.5156,
+      "step": 2945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953436513247526e-05,
+      "loss": 7.7344,
+      "step": 2946
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953390850654659e-05,
+      "loss": 5.9062,
+      "step": 2947
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953345165947163e-05,
+      "loss": 3.8281,
+      "step": 2948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953299459125556e-05,
+      "loss": 4.5312,
+      "step": 2949
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953253730190353e-05,
+      "loss": 4.4062,
+      "step": 2950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953207979142073e-05,
+      "loss": 7.8125,
+      "step": 2951
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953162205981235e-05,
+      "loss": 5.5156,
+      "step": 2952
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953116410708356e-05,
+      "loss": 2.4688,
+      "step": 2953
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953070593323954e-05,
+      "loss": 4.5625,
+      "step": 2954
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.953024753828549e-05,
+      "loss": 3.7656,
+      "step": 2955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.952978892222659e-05,
+      "loss": 4.2656,
+      "step": 2956
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9529330085068044e-05,
+      "loss": 6.375,
+      "step": 2957
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.952887102681503e-05,
+      "loss": 7.875,
+      "step": 2958
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.952841174747276e-05,
+      "loss": 4.7188,
+      "step": 2959
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.952795224704642e-05,
+      "loss": 5.875,
+      "step": 2960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9527492525541215e-05,
+      "loss": 3.6875,
+      "step": 2961
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9527032582962366e-05,
+      "loss": 4.1562,
+      "step": 2962
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.952657241931506e-05,
+      "loss": 7.9062,
+      "step": 2963
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9526112034604515e-05,
+      "loss": 6.5312,
+      "step": 2964
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.952565142883595e-05,
+      "loss": 4.3438,
+      "step": 2965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9525190602014564e-05,
+      "loss": 4.375,
+      "step": 2966
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.952472955414558e-05,
+      "loss": 3.0469,
+      "step": 2967
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.952426828523422e-05,
+      "loss": 3.7812,
+      "step": 2968
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.952380679528571e-05,
+      "loss": 7.4062,
+      "step": 2969
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.952334508430527e-05,
+      "loss": 4.1562,
+      "step": 2970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9522883152298124e-05,
+      "loss": 7.4688,
+      "step": 2971
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.952242099926951e-05,
+      "loss": 3.8438,
+      "step": 2972
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.952195862522465e-05,
+      "loss": 3.125,
+      "step": 2973
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.952149603016878e-05,
+      "loss": 4.625,
+      "step": 2974
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.952103321410715e-05,
+      "loss": 4.9375,
+      "step": 2975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9520570177044984e-05,
+      "loss": 3.8125,
+      "step": 2976
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.952010691898753e-05,
+      "loss": 2.8125,
+      "step": 2977
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9519643439940034e-05,
+      "loss": 6.1875,
+      "step": 2978
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9519179739907744e-05,
+      "loss": 3.9688,
+      "step": 2979
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.951871581889591e-05,
+      "loss": 5.7656,
+      "step": 2980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9518251676909785e-05,
+      "loss": 3.6875,
+      "step": 2981
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.951778731395461e-05,
+      "loss": 3.9062,
+      "step": 2982
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9517322730035656e-05,
+      "loss": 3.5,
+      "step": 2983
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.951685792515818e-05,
+      "loss": 5.9219,
+      "step": 2984
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.951639289932745e-05,
+      "loss": 2.6406,
+      "step": 2985
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9515927652548714e-05,
+      "loss": 6.75,
+      "step": 2986
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.951546218482726e-05,
+      "loss": 5.75,
+      "step": 2987
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.951499649616834e-05,
+      "loss": 6.0938,
+      "step": 2988
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9514530586577226e-05,
+      "loss": 4.7188,
+      "step": 2989
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.951406445605921e-05,
+      "loss": 6.9688,
+      "step": 2990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9513598104619555e-05,
+      "loss": 3.5625,
+      "step": 2991
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.951313153226354e-05,
+      "loss": 4.6875,
+      "step": 2992
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.951266473899646e-05,
+      "loss": 6.8906,
+      "step": 2993
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.951219772482359e-05,
+      "loss": 5.8047,
+      "step": 2994
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9511730489750213e-05,
+      "loss": 4.3438,
+      "step": 2995
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9511263033781627e-05,
+      "loss": 5.6094,
+      "step": 2996
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9510795356923124e-05,
+      "loss": 4.5312,
+      "step": 2997
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.951032745917999e-05,
+      "loss": 6.2812,
+      "step": 2998
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950985934055753e-05,
+      "loss": 7.4531,
+      "step": 2999
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950939100106104e-05,
+      "loss": 6.375,
+      "step": 3000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950892244069583e-05,
+      "loss": 6.9688,
+      "step": 3001
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9508453659467185e-05,
+      "loss": 3.5625,
+      "step": 3002
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950798465738044e-05,
+      "loss": 6.4688,
+      "step": 3003
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950751543444088e-05,
+      "loss": 4.0938,
+      "step": 3004
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950704599065384e-05,
+      "loss": 4.8906,
+      "step": 3005
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950657632602461e-05,
+      "loss": 4.0625,
+      "step": 3006
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950610644055852e-05,
+      "loss": 5.125,
+      "step": 3007
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950563633426089e-05,
+      "loss": 5.9062,
+      "step": 3008
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950516600713704e-05,
+      "loss": 6.8125,
+      "step": 3009
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950469545919229e-05,
+      "loss": 7.2188,
+      "step": 3010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950422469043198e-05,
+      "loss": 5.7344,
+      "step": 3011
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950375370086144e-05,
+      "loss": 4.4688,
+      "step": 3012
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950328249048598e-05,
+      "loss": 5.5312,
+      "step": 3013
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9502811059310955e-05,
+      "loss": 4.2969,
+      "step": 3014
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950233940734169e-05,
+      "loss": 6.4688,
+      "step": 3015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9501867534583545e-05,
+      "loss": 5.2188,
+      "step": 3016
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950139544104184e-05,
+      "loss": 3.9375,
+      "step": 3017
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950092312672192e-05,
+      "loss": 11.4219,
+      "step": 3018
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9500450591629145e-05,
+      "loss": 4.4062,
+      "step": 3019
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.949997783576886e-05,
+      "loss": 8.5625,
+      "step": 3020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9499504859146414e-05,
+      "loss": 4.1562,
+      "step": 3021
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.949903166176717e-05,
+      "loss": 6.7812,
+      "step": 3022
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.949855824363647e-05,
+      "loss": 7.375,
+      "step": 3023
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9498084604759695e-05,
+      "loss": 2.9688,
+      "step": 3024
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.949761074514219e-05,
+      "loss": 8.2812,
+      "step": 3025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.949713666478932e-05,
+      "loss": 3.7656,
+      "step": 3026
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.949666236370646e-05,
+      "loss": 2.8438,
+      "step": 3027
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.949618784189897e-05,
+      "loss": 9.5312,
+      "step": 3028
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.949571309937224e-05,
+      "loss": 5.4062,
+      "step": 3029
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.949523813613163e-05,
+      "loss": 3.6875,
+      "step": 3030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9494762952182516e-05,
+      "loss": 4.6406,
+      "step": 3031
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9494287547530286e-05,
+      "loss": 4.7812,
+      "step": 3032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9493811922180316e-05,
+      "loss": 2.8125,
+      "step": 3033
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.949333607613799e-05,
+      "loss": 6.2969,
+      "step": 3034
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9492860009408706e-05,
+      "loss": 4.0,
+      "step": 3035
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9492383721997835e-05,
+      "loss": 5.3281,
+      "step": 3036
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9491907213910784e-05,
+      "loss": 5.0938,
+      "step": 3037
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.949143048515295e-05,
+      "loss": 3.0625,
+      "step": 3038
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9490953535729715e-05,
+      "loss": 4.2812,
+      "step": 3039
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.949047636564648e-05,
+      "loss": 5.0312,
+      "step": 3040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9489998974908675e-05,
+      "loss": 4.875,
+      "step": 3041
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9489521363521665e-05,
+      "loss": 3.1562,
+      "step": 3042
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9489043531490886e-05,
+      "loss": 6.0,
+      "step": 3043
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.948856547882173e-05,
+      "loss": 4.75,
+      "step": 3044
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.948808720551962e-05,
+      "loss": 5.5625,
+      "step": 3045
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9487608711589966e-05,
+      "loss": 3.7656,
+      "step": 3046
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.948712999703819e-05,
+      "loss": 5.9531,
+      "step": 3047
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.948665106186971e-05,
+      "loss": 3.125,
+      "step": 3048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9486171906089936e-05,
+      "loss": 5.125,
+      "step": 3049
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.948569252970431e-05,
+      "loss": 4.2812,
+      "step": 3050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.948521293271824e-05,
+      "loss": 5.7656,
+      "step": 3051
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9484733115137184e-05,
+      "loss": 6.1875,
+      "step": 3052
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9484253076966545e-05,
+      "loss": 5.75,
+      "step": 3053
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.948377281821177e-05,
+      "loss": 1.7656,
+      "step": 3054
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9483292338878306e-05,
+      "loss": 4.3125,
+      "step": 3055
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.948281163897157e-05,
+      "loss": 3.9062,
+      "step": 3056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.948233071849703e-05,
+      "loss": 3.8281,
+      "step": 3057
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.94818495774601e-05,
+      "loss": 2.1719,
+      "step": 3058
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9481368215866255e-05,
+      "loss": 2.4219,
+      "step": 3059
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.948088663372093e-05,
+      "loss": 4.4688,
+      "step": 3060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.948040483102959e-05,
+      "loss": 4.875,
+      "step": 3061
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.947992280779767e-05,
+      "loss": 2.2812,
+      "step": 3062
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9479440564030643e-05,
+      "loss": 6.0312,
+      "step": 3063
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.947895809973396e-05,
+      "loss": 2.6875,
+      "step": 3064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.947847541491308e-05,
+      "loss": 3.9062,
+      "step": 3065
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9477992509573485e-05,
+      "loss": 3.2188,
+      "step": 3066
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.947750938372063e-05,
+      "loss": 3.2969,
+      "step": 3067
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.947702603735998e-05,
+      "loss": 3.9062,
+      "step": 3068
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.947654247049701e-05,
+      "loss": 4.5312,
+      "step": 3069
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.94760586831372e-05,
+      "loss": 4.6719,
+      "step": 3070
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.947557467528603e-05,
+      "loss": 3.75,
+      "step": 3071
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.947509044694896e-05,
+      "loss": 4.0469,
+      "step": 3072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.947460599813149e-05,
+      "loss": 5.375,
+      "step": 3073
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.94741213288391e-05,
+      "loss": 2.3438,
+      "step": 3074
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.947363643907729e-05,
+      "loss": 3.1875,
+      "step": 3075
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9473151328851515e-05,
+      "loss": 9.2656,
+      "step": 3076
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.94726659981673e-05,
+      "loss": 4.625,
+      "step": 3077
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9472180447030125e-05,
+      "loss": 5.6406,
+      "step": 3078
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9471694675445486e-05,
+      "loss": 2.6875,
+      "step": 3079
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9471208683418895e-05,
+      "loss": 3.8125,
+      "step": 3080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.947072247095584e-05,
+      "loss": 2.8438,
+      "step": 3081
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.947023603806184e-05,
+      "loss": 5.4219,
+      "step": 3082
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.946974938474238e-05,
+      "loss": 3.6406,
+      "step": 3083
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9469262511002984e-05,
+      "loss": 3.4531,
+      "step": 3084
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.946877541684917e-05,
+      "loss": 3.9062,
+      "step": 3085
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9468288102286435e-05,
+      "loss": 5.4688,
+      "step": 3086
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9467800567320315e-05,
+      "loss": 3.5938,
+      "step": 3087
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.946731281195631e-05,
+      "loss": 3.9062,
+      "step": 3088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.946682483619996e-05,
+      "loss": 6.4062,
+      "step": 3089
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9466336640056784e-05,
+      "loss": 3.7812,
+      "step": 3090
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.94658482235323e-05,
+      "loss": 6.25,
+      "step": 3091
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.946535958663204e-05,
+      "loss": 4.6875,
+      "step": 3092
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.946487072936155e-05,
+      "loss": 5.2656,
+      "step": 3093
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.946438165172635e-05,
+      "loss": 3.7812,
+      "step": 3094
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.946389235373198e-05,
+      "loss": 2.6875,
+      "step": 3095
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.946340283538398e-05,
+      "loss": 7.375,
+      "step": 3096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.946291309668789e-05,
+      "loss": 6.0938,
+      "step": 3097
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9462423137649264e-05,
+      "loss": 6.0,
+      "step": 3098
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.946193295827364e-05,
+      "loss": 2.4062,
+      "step": 3099
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.946144255856656e-05,
+      "loss": 11.5312,
+      "step": 3100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.94609519385336e-05,
+      "loss": 7.625,
+      "step": 3101
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.946046109818029e-05,
+      "loss": 9.5938,
+      "step": 3102
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.94599700375122e-05,
+      "loss": 3.7969,
+      "step": 3103
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.945947875653488e-05,
+      "loss": 4.9375,
+      "step": 3104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.94589872552539e-05,
+      "loss": 6.0938,
+      "step": 3105
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.945849553367483e-05,
+      "loss": 4.5312,
+      "step": 3106
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.945800359180322e-05,
+      "loss": 7.7188,
+      "step": 3107
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.945751142964465e-05,
+      "loss": 7.3906,
+      "step": 3108
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.945701904720469e-05,
+      "loss": 3.9375,
+      "step": 3109
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9456526444488916e-05,
+      "loss": 3.5,
+      "step": 3110
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9456033621502897e-05,
+      "loss": 4.0156,
+      "step": 3111
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.945554057825223e-05,
+      "loss": 4.9062,
+      "step": 3112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.945504731474248e-05,
+      "loss": 4.4375,
+      "step": 3113
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9454553830979225e-05,
+      "loss": 4.6562,
+      "step": 3114
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.945406012696808e-05,
+      "loss": 7.2812,
+      "step": 3115
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9453566202714606e-05,
+      "loss": 5.5312,
+      "step": 3116
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9453072058224415e-05,
+      "loss": 5.4531,
+      "step": 3117
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.945257769350309e-05,
+      "loss": 4.1797,
+      "step": 3118
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9452083108556225e-05,
+      "loss": 6.5,
+      "step": 3119
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.945158830338943e-05,
+      "loss": 4.5312,
+      "step": 3120
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.94510932780083e-05,
+      "loss": 5.4062,
+      "step": 3121
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9450598032418444e-05,
+      "loss": 9.3594,
+      "step": 3122
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9450102566625464e-05,
+      "loss": 5.2812,
+      "step": 3123
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.944960688063497e-05,
+      "loss": 4.4062,
+      "step": 3124
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.944911097445257e-05,
+      "loss": 2.3125,
+      "step": 3125
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.944861484808389e-05,
+      "loss": 3.7969,
+      "step": 3126
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.944811850153453e-05,
+      "loss": 6.4219,
+      "step": 3127
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.944762193481013e-05,
+      "loss": 9.1562,
+      "step": 3128
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.94471251479163e-05,
+      "loss": 8.25,
+      "step": 3129
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.944662814085866e-05,
+      "loss": 5.0938,
+      "step": 3130
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.944613091364284e-05,
+      "loss": 4.6094,
+      "step": 3131
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9445633466274464e-05,
+      "loss": 4.25,
+      "step": 3132
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9445135798759186e-05,
+      "loss": 5.2969,
+      "step": 3133
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9444637911102614e-05,
+      "loss": 5.0469,
+      "step": 3134
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.94441398033104e-05,
+      "loss": 3.0,
+      "step": 3135
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.944364147538817e-05,
+      "loss": 3.7188,
+      "step": 3136
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.944314292734159e-05,
+      "loss": 5.5625,
+      "step": 3137
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.944264415917627e-05,
+      "loss": 6.6875,
+      "step": 3138
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.944214517089788e-05,
+      "loss": 12.3125,
+      "step": 3139
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.944164596251207e-05,
+      "loss": 5.3125,
+      "step": 3140
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.944114653402448e-05,
+      "loss": 4.625,
+      "step": 3141
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.944064688544077e-05,
+      "loss": 3.2188,
+      "step": 3142
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9440147016766594e-05,
+      "loss": 3.6875,
+      "step": 3143
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.943964692800762e-05,
+      "loss": 3.125,
+      "step": 3144
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9439146619169496e-05,
+      "loss": 4.4844,
+      "step": 3145
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.943864609025789e-05,
+      "loss": 5.6562,
+      "step": 3146
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.943814534127848e-05,
+      "loss": 2.0,
+      "step": 3147
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.943764437223692e-05,
+      "loss": 3.5,
+      "step": 3148
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.943714318313889e-05,
+      "loss": 5.0,
+      "step": 3149
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9436641773990064e-05,
+      "loss": 6.7188,
+      "step": 3150
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.943614014479611e-05,
+      "loss": 3.1406,
+      "step": 3151
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.943563829556272e-05,
+      "loss": 6.375,
+      "step": 3152
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.943513622629557e-05,
+      "loss": 5.4219,
+      "step": 3153
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9434633937000335e-05,
+      "loss": 5.0,
+      "step": 3154
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.943413142768271e-05,
+      "loss": 8.9375,
+      "step": 3155
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9433628698348387e-05,
+      "loss": 4.0625,
+      "step": 3156
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9433125749003055e-05,
+      "loss": 4.4375,
+      "step": 3157
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.943262257965241e-05,
+      "loss": 3.375,
+      "step": 3158
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.943211919030214e-05,
+      "loss": 5.5625,
+      "step": 3159
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.943161558095795e-05,
+      "loss": 7.5469,
+      "step": 3160
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.943111175162554e-05,
+      "loss": 4.3438,
+      "step": 3161
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.943060770231061e-05,
+      "loss": 4.4062,
+      "step": 3162
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9430103433018884e-05,
+      "loss": 5.3594,
+      "step": 3163
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9429598943756053e-05,
+      "loss": 4.0156,
+      "step": 3164
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.942909423452783e-05,
+      "loss": 4.0312,
+      "step": 3165
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9428589305339935e-05,
+      "loss": 3.7188,
+      "step": 3166
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.942808415619808e-05,
+      "loss": 5.6719,
+      "step": 3167
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9427578787107983e-05,
+      "loss": 3.9375,
+      "step": 3168
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9427073198075375e-05,
+      "loss": 8.0781,
+      "step": 3169
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9426567389105966e-05,
+      "loss": 10.8438,
+      "step": 3170
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9426061360205496e-05,
+      "loss": 3.2188,
+      "step": 3171
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.942555511137968e-05,
+      "loss": 3.75,
+      "step": 3172
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.942504864263426e-05,
+      "loss": 4.0625,
+      "step": 3173
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9424541953974964e-05,
+      "loss": 5.6719,
+      "step": 3174
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.942403504540753e-05,
+      "loss": 7.6562,
+      "step": 3175
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9423527916937696e-05,
+      "loss": 6.3438,
+      "step": 3176
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.942302056857121e-05,
+      "loss": 4.9375,
+      "step": 3177
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.94225130003138e-05,
+      "loss": 6.25,
+      "step": 3178
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.942200521217123e-05,
+      "loss": 4.4062,
+      "step": 3179
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.942149720414924e-05,
+      "loss": 4.9375,
+      "step": 3180
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9420988976253585e-05,
+      "loss": 5.4531,
+      "step": 3181
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.942048052849001e-05,
+      "loss": 1.4062,
+      "step": 3182
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.941997186086428e-05,
+      "loss": 4.0156,
+      "step": 3183
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9419462973382155e-05,
+      "loss": 5.875,
+      "step": 3184
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9418953866049394e-05,
+      "loss": 6.4375,
+      "step": 3185
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.941844453887175e-05,
+      "loss": 3.7344,
+      "step": 3186
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9417934991855e-05,
+      "loss": 5.5312,
+      "step": 3187
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.941742522500491e-05,
+      "loss": 7.3281,
+      "step": 3188
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9416915238327254e-05,
+      "loss": 7.125,
+      "step": 3189
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.941640503182781e-05,
+      "loss": 4.7188,
+      "step": 3190
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.941589460551234e-05,
+      "loss": 8.1094,
+      "step": 3191
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.941538395938663e-05,
+      "loss": 5.7812,
+      "step": 3192
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9414873093456466e-05,
+      "loss": 6.1562,
+      "step": 3193
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.941436200772762e-05,
+      "loss": 1.9688,
+      "step": 3194
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9413850702205894e-05,
+      "loss": 5.2656,
+      "step": 3195
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.941333917689706e-05,
+      "loss": 7.625,
+      "step": 3196
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.941282743180692e-05,
+      "loss": 2.125,
+      "step": 3197
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.941231546694126e-05,
+      "loss": 3.0,
+      "step": 3198
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9411803282305886e-05,
+      "loss": 6.75,
+      "step": 3199
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9411290877906586e-05,
+      "loss": 5.9062,
+      "step": 3200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.941077825374917e-05,
+      "loss": 10.1562,
+      "step": 3201
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9410265409839435e-05,
+      "loss": 6.6562,
+      "step": 3202
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.940975234618319e-05,
+      "loss": 9.6094,
+      "step": 3203
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.940923906278624e-05,
+      "loss": 8.0625,
+      "step": 3204
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.94087255596544e-05,
+      "loss": 4.125,
+      "step": 3205
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.940821183679348e-05,
+      "loss": 7.7812,
+      "step": 3206
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9407697894209305e-05,
+      "loss": 4.5312,
+      "step": 3207
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9407183731907676e-05,
+      "loss": 9.0938,
+      "step": 3208
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9406669349894435e-05,
+      "loss": 11.375,
+      "step": 3209
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.940615474817539e-05,
+      "loss": 6.125,
+      "step": 3210
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9405639926756374e-05,
+      "loss": 5.5469,
+      "step": 3211
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.940512488564321e-05,
+      "loss": 4.75,
+      "step": 3212
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.940460962484173e-05,
+      "loss": 5.7188,
+      "step": 3213
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9404094144357767e-05,
+      "loss": 4.9531,
+      "step": 3214
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9403578444197166e-05,
+      "loss": 3.2812,
+      "step": 3215
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.940306252436575e-05,
+      "loss": 6.75,
+      "step": 3216
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9402546384869376e-05,
+      "loss": 6.0,
+      "step": 3217
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.940203002571388e-05,
+      "loss": 6.8906,
+      "step": 3218
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.940151344690511e-05,
+      "loss": 4.2031,
+      "step": 3219
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.94009966484489e-05,
+      "loss": 5.5938,
+      "step": 3220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9400479630351124e-05,
+      "loss": 2.8438,
+      "step": 3221
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939996239261762e-05,
+      "loss": 2.9688,
+      "step": 3222
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939944493525424e-05,
+      "loss": 5.0938,
+      "step": 3223
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939892725826685e-05,
+      "loss": 5.2812,
+      "step": 3224
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939840936166132e-05,
+      "loss": 5.5625,
+      "step": 3225
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.93978912454435e-05,
+      "loss": 4.8125,
+      "step": 3226
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939737290961925e-05,
+      "loss": 6.9688,
+      "step": 3227
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939685435419445e-05,
+      "loss": 2.1875,
+      "step": 3228
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9396335579174976e-05,
+      "loss": 4.5938,
+      "step": 3229
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939581658456669e-05,
+      "loss": 6.3125,
+      "step": 3230
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939529737037547e-05,
+      "loss": 4.2812,
+      "step": 3231
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939477793660719e-05,
+      "loss": 4.8438,
+      "step": 3232
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9394258283267735e-05,
+      "loss": 4.2969,
+      "step": 3233
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939373841036299e-05,
+      "loss": 5.3438,
+      "step": 3234
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939321831789885e-05,
+      "loss": 4.625,
+      "step": 3235
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939269800588118e-05,
+      "loss": 4.1094,
+      "step": 3236
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939217747431588e-05,
+      "loss": 3.4062,
+      "step": 3237
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939165672320886e-05,
+      "loss": 5.8438,
+      "step": 3238
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9391135752565984e-05,
+      "loss": 3.9375,
+      "step": 3239
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.939061456239318e-05,
+      "loss": 3.1406,
+      "step": 3240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9390093152696325e-05,
+      "loss": 2.8125,
+      "step": 3241
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.938957152348134e-05,
+      "loss": 5.875,
+      "step": 3242
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9389049674754116e-05,
+      "loss": 6.3125,
+      "step": 3243
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.938852760652057e-05,
+      "loss": 5.8125,
+      "step": 3244
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.938800531878661e-05,
+      "loss": 5.2812,
+      "step": 3245
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.938748281155815e-05,
+      "loss": 4.9375,
+      "step": 3246
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.938696008484111e-05,
+      "loss": 7.3906,
+      "step": 3247
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.93864371386414e-05,
+      "loss": 4.25,
+      "step": 3248
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9385913972964934e-05,
+      "loss": 4.4375,
+      "step": 3249
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9385390587817647e-05,
+      "loss": 4.625,
+      "step": 3250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9384866983205464e-05,
+      "loss": 6.7344,
+      "step": 3251
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.938434315913431e-05,
+      "loss": 8.2031,
+      "step": 3252
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.938381911561011e-05,
+      "loss": 4.9688,
+      "step": 3253
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9383294852638804e-05,
+      "loss": 3.875,
+      "step": 3254
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.938277037022633e-05,
+      "loss": 4.5312,
+      "step": 3255
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9382245668378615e-05,
+      "loss": 4.9844,
+      "step": 3256
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9381720747101604e-05,
+      "loss": 5.0625,
+      "step": 3257
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.938119560640124e-05,
+      "loss": 3.7812,
+      "step": 3258
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9380670246283474e-05,
+      "loss": 5.4375,
+      "step": 3259
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.938014466675425e-05,
+      "loss": 6.8125,
+      "step": 3260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937961886781951e-05,
+      "loss": 3.1094,
+      "step": 3261
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937909284948521e-05,
+      "loss": 1.5625,
+      "step": 3262
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937856661175732e-05,
+      "loss": 5.0938,
+      "step": 3263
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937804015464178e-05,
+      "loss": 8.1094,
+      "step": 3264
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937751347814455e-05,
+      "loss": 5.625,
+      "step": 3265
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937698658227161e-05,
+      "loss": 5.6875,
+      "step": 3266
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937645946702891e-05,
+      "loss": 4.9375,
+      "step": 3267
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937593213242242e-05,
+      "loss": 3.2188,
+      "step": 3268
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9375404578458105e-05,
+      "loss": 3.0938,
+      "step": 3269
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937487680514195e-05,
+      "loss": 5.6719,
+      "step": 3270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937434881247993e-05,
+      "loss": 3.4688,
+      "step": 3271
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.937382060047801e-05,
+      "loss": 3.25,
+      "step": 3272
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9373292169142176e-05,
+      "loss": 7.4375,
+      "step": 3273
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.937276351847841e-05,
+      "loss": 6.4062,
+      "step": 3274
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.93722346484927e-05,
+      "loss": 5.25,
+      "step": 3275
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9371705559191035e-05,
+      "loss": 5.1875,
+      "step": 3276
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.937117625057939e-05,
+      "loss": 1.6562,
+      "step": 3277
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9370646722663775e-05,
+      "loss": 4.5156,
+      "step": 3278
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.937011697545018e-05,
+      "loss": 4.9531,
+      "step": 3279
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.93695870089446e-05,
+      "loss": 3.25,
+      "step": 3280
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.936905682315303e-05,
+      "loss": 2.5156,
+      "step": 3281
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.936852641808148e-05,
+      "loss": 2.7812,
+      "step": 3282
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.936799579373595e-05,
+      "loss": 5.375,
+      "step": 3283
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.936746495012246e-05,
+      "loss": 3.375,
+      "step": 3284
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9366933887247e-05,
+      "loss": 3.875,
+      "step": 3285
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9366402605115596e-05,
+      "loss": 3.0938,
+      "step": 3286
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9365871103734264e-05,
+      "loss": 5.9062,
+      "step": 3287
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9365339383109005e-05,
+      "loss": 5.6875,
+      "step": 3288
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9364807443245855e-05,
+      "loss": 4.4219,
+      "step": 3289
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9364275284150826e-05,
+      "loss": 3.3125,
+      "step": 3290
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9363742905829955e-05,
+      "loss": 4.625,
+      "step": 3291
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.936321030828926e-05,
+      "loss": 3.4375,
+      "step": 3292
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.936267749153476e-05,
+      "loss": 3.7812,
+      "step": 3293
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.936214445557252e-05,
+      "loss": 6.0859,
+      "step": 3294
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9361611200408533e-05,
+      "loss": 2.1875,
+      "step": 3295
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.936107772604887e-05,
+      "loss": 5.1875,
+      "step": 3296
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.936054403249955e-05,
+      "loss": 9.0312,
+      "step": 3297
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.936001011976663e-05,
+      "loss": 10.5625,
+      "step": 3298
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.935947598785614e-05,
+      "loss": 4.2188,
+      "step": 3299
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9358941636774134e-05,
+      "loss": 5.4688,
+      "step": 3300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9358407066526664e-05,
+      "loss": 4.9531,
+      "step": 3301
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.935787227711978e-05,
+      "loss": 3.6094,
+      "step": 3302
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.935733726855953e-05,
+      "loss": 6.0938,
+      "step": 3303
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.935680204085199e-05,
+      "loss": 4.5625,
+      "step": 3304
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9356266594003194e-05,
+      "loss": 10.125,
+      "step": 3305
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.935573092801922e-05,
+      "loss": 7.375,
+      "step": 3306
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.935519504290613e-05,
+      "loss": 3.9062,
+      "step": 3307
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.935465893866998e-05,
+      "loss": 2.1094,
+      "step": 3308
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9354122615316856e-05,
+      "loss": 6.2188,
+      "step": 3309
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.935358607285282e-05,
+      "loss": 5.0781,
+      "step": 3310
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9353049311283946e-05,
+      "loss": 5.2812,
+      "step": 3311
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.935251233061632e-05,
+      "loss": 3.4688,
+      "step": 3312
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9351975130856e-05,
+      "loss": 2.5156,
+      "step": 3313
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.93514377120091e-05,
+      "loss": 5.4219,
+      "step": 3314
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.935090007408168e-05,
+      "loss": 5.0312,
+      "step": 3315
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.935036221707983e-05,
+      "loss": 7.7031,
+      "step": 3316
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.934982414100964e-05,
+      "loss": 3.6562,
+      "step": 3317
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9349285845877207e-05,
+      "loss": 8.7969,
+      "step": 3318
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.934874733168863e-05,
+      "loss": 4.1719,
+      "step": 3319
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9348208598449986e-05,
+      "loss": 2.625,
+      "step": 3320
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9347669646167396e-05,
+      "loss": 4.0156,
+      "step": 3321
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9347130474846945e-05,
+      "loss": 3.5469,
+      "step": 3322
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.934659108449474e-05,
+      "loss": 4.7031,
+      "step": 3323
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.934605147511689e-05,
+      "loss": 4.7656,
+      "step": 3324
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.934551164671951e-05,
+      "loss": 4.5625,
+      "step": 3325
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9344971599308706e-05,
+      "loss": 7.2344,
+      "step": 3326
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.934443133289059e-05,
+      "loss": 4.0312,
+      "step": 3327
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9343890847471284e-05,
+      "loss": 3.875,
+      "step": 3328
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.93433501430569e-05,
+      "loss": 6.5938,
+      "step": 3329
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.934280921965356e-05,
+      "loss": 2.5938,
+      "step": 3330
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.93422680772674e-05,
+      "loss": 3.75,
+      "step": 3331
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.934172671590453e-05,
+      "loss": 4.5156,
+      "step": 3332
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.934118513557109e-05,
+      "loss": 3.0312,
+      "step": 3333
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.93406433362732e-05,
+      "loss": 8.7344,
+      "step": 3334
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.934010131801701e-05,
+      "loss": 4.4062,
+      "step": 3335
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9339559080808644e-05,
+      "loss": 4.0938,
+      "step": 3336
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.933901662465424e-05,
+      "loss": 4.25,
+      "step": 3337
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.933847394955995e-05,
+      "loss": 5.1094,
+      "step": 3338
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.933793105553191e-05,
+      "loss": 4.1562,
+      "step": 3339
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.933738794257626e-05,
+      "loss": 3.5625,
+      "step": 3340
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.933684461069916e-05,
+      "loss": 3.4062,
+      "step": 3341
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.933630105990676e-05,
+      "loss": 3.9062,
+      "step": 3342
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.933575729020521e-05,
+      "loss": 4.7969,
+      "step": 3343
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.933521330160067e-05,
+      "loss": 2.8438,
+      "step": 3344
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9334669094099285e-05,
+      "loss": 3.7188,
+      "step": 3345
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9334124667707235e-05,
+      "loss": 4.4688,
+      "step": 3346
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.933358002243067e-05,
+      "loss": 4.6562,
+      "step": 3347
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9333035158275763e-05,
+      "loss": 4.3438,
+      "step": 3348
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.933249007524868e-05,
+      "loss": 5.4062,
+      "step": 3349
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.933194477335559e-05,
+      "loss": 6.0469,
+      "step": 3350
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.933139925260268e-05,
+      "loss": 3.4688,
+      "step": 3351
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.93308535129961e-05,
+      "loss": 3.3438,
+      "step": 3352
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9330307554542046e-05,
+      "loss": 3.6562,
+      "step": 3353
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.93297613772467e-05,
+      "loss": 3.8281,
+      "step": 3354
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9329214981116245e-05,
+      "loss": 5.0938,
+      "step": 3355
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.932866836615686e-05,
+      "loss": 6.6094,
+      "step": 3356
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.932812153237473e-05,
+      "loss": 4.0938,
+      "step": 3357
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.932757447977606e-05,
+      "loss": 4.1094,
+      "step": 3358
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9327027208367034e-05,
+      "loss": 3.7344,
+      "step": 3359
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.932647971815384e-05,
+      "loss": 4.0,
+      "step": 3360
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.93259320091427e-05,
+      "loss": 10.6406,
+      "step": 3361
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.932538408133979e-05,
+      "loss": 4.8438,
+      "step": 3362
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.932483593475133e-05,
+      "loss": 5.5312,
+      "step": 3363
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.932428756938352e-05,
+      "loss": 4.8125,
+      "step": 3364
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.932373898524256e-05,
+      "loss": 2.5938,
+      "step": 3365
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.932319018233467e-05,
+      "loss": 6.6562,
+      "step": 3366
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9322641160666064e-05,
+      "loss": 4.5625,
+      "step": 3367
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.932209192024295e-05,
+      "loss": 3.7031,
+      "step": 3368
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.932154246107156e-05,
+      "loss": 2.625,
+      "step": 3369
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.932099278315809e-05,
+      "loss": 3.25,
+      "step": 3370
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.932044288650879e-05,
+      "loss": 2.4688,
+      "step": 3371
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.931989277112987e-05,
+      "loss": 2.9375,
+      "step": 3372
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9319342437027553e-05,
+      "loss": 3.5938,
+      "step": 3373
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.931879188420809e-05,
+      "loss": 6.2031,
+      "step": 3374
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9318241112677694e-05,
+      "loss": 3.1875,
+      "step": 3375
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.931769012244261e-05,
+      "loss": 4.8125,
+      "step": 3376
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9317138913509076e-05,
+      "loss": 6.4062,
+      "step": 3377
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9316587485883333e-05,
+      "loss": 3.4375,
+      "step": 3378
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9316035839571614e-05,
+      "loss": 5.0312,
+      "step": 3379
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.931548397458017e-05,
+      "loss": 2.4375,
+      "step": 3380
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9314931890915254e-05,
+      "loss": 4.5625,
+      "step": 3381
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.931437958858312e-05,
+      "loss": 3.4375,
+      "step": 3382
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.931382706759e-05,
+      "loss": 4.3125,
+      "step": 3383
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9313274327942166e-05,
+      "loss": 6.0156,
+      "step": 3384
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9312721369645874e-05,
+      "loss": 3.125,
+      "step": 3385
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.931216819270738e-05,
+      "loss": 3.5938,
+      "step": 3386
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9311614797132945e-05,
+      "loss": 3.4375,
+      "step": 3387
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9311061182928844e-05,
+      "loss": 3.4062,
+      "step": 3388
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.931050735010134e-05,
+      "loss": 4.6406,
+      "step": 3389
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.930995329865669e-05,
+      "loss": 6.6562,
+      "step": 3390
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.930939902860119e-05,
+      "loss": 4.5625,
+      "step": 3391
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.930884453994109e-05,
+      "loss": 4.125,
+      "step": 3392
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.930828983268269e-05,
+      "loss": 3.9531,
+      "step": 3393
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9307734906832256e-05,
+      "loss": 5.25,
+      "step": 3394
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.930717976239607e-05,
+      "loss": 4.4531,
+      "step": 3395
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9306624399380425e-05,
+      "loss": 2.8906,
+      "step": 3396
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9306068817791614e-05,
+      "loss": 2.5625,
+      "step": 3397
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.930551301763591e-05,
+      "loss": 4.2344,
+      "step": 3398
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9304956998919606e-05,
+      "loss": 4.0938,
+      "step": 3399
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.930440076164901e-05,
+      "loss": 4.1875,
+      "step": 3400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.930384430583041e-05,
+      "loss": 5.8438,
+      "step": 3401
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.930328763147011e-05,
+      "loss": 3.6875,
+      "step": 3402
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.930273073857441e-05,
+      "loss": 3.2188,
+      "step": 3403
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9302173627149615e-05,
+      "loss": 4.0312,
+      "step": 3404
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9301616297202035e-05,
+      "loss": 4.25,
+      "step": 3405
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.930105874873797e-05,
+      "loss": 3.2812,
+      "step": 3406
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9300500981763743e-05,
+      "loss": 3.25,
+      "step": 3407
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9299942996285676e-05,
+      "loss": 3.1562,
+      "step": 3408
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.929938479231006e-05,
+      "loss": 8.0,
+      "step": 3409
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9298826369843237e-05,
+      "loss": 2.3438,
+      "step": 3410
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9298267728891515e-05,
+      "loss": 3.9531,
+      "step": 3411
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9297708869461224e-05,
+      "loss": 3.7188,
+      "step": 3412
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.92971497915587e-05,
+      "loss": 4.8438,
+      "step": 3413
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.929659049519026e-05,
+      "loss": 4.4062,
+      "step": 3414
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9296030980362235e-05,
+      "loss": 2.9375,
+      "step": 3415
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.929547124708097e-05,
+      "loss": 5.4531,
+      "step": 3416
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.929491129535279e-05,
+      "loss": 3.875,
+      "step": 3417
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.929435112518404e-05,
+      "loss": 2.5312,
+      "step": 3418
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9293790736581077e-05,
+      "loss": 4.4219,
+      "step": 3419
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9293230129550214e-05,
+      "loss": 6.0156,
+      "step": 3420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9292669304097824e-05,
+      "loss": 4.5625,
+      "step": 3421
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.929210826023024e-05,
+      "loss": 4.5469,
+      "step": 3422
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.929154699795382e-05,
+      "loss": 5.6875,
+      "step": 3423
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.929098551727492e-05,
+      "loss": 6.5469,
+      "step": 3424
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9290423818199894e-05,
+      "loss": 3.5938,
+      "step": 3425
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9289861900735105e-05,
+      "loss": 9.0156,
+      "step": 3426
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.928929976488691e-05,
+      "loss": 3.0781,
+      "step": 3427
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9288737410661675e-05,
+      "loss": 4.7188,
+      "step": 3428
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.928817483806576e-05,
+      "loss": 5.2656,
+      "step": 3429
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.928761204710554e-05,
+      "loss": 1.9375,
+      "step": 3430
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.928704903778739e-05,
+      "loss": 3.4062,
+      "step": 3431
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.928648581011768e-05,
+      "loss": 5.4688,
+      "step": 3432
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9285922364102794e-05,
+      "loss": 7.6875,
+      "step": 3433
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.92853586997491e-05,
+      "loss": 4.2656,
+      "step": 3434
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.928479481706297e-05,
+      "loss": 6.4688,
+      "step": 3435
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.928423071605081e-05,
+      "loss": 3.2344,
+      "step": 3436
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.928366639671899e-05,
+      "loss": 6.0625,
+      "step": 3437
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.928310185907392e-05,
+      "loss": 5.7188,
+      "step": 3438
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.928253710312196e-05,
+      "loss": 5.5781,
+      "step": 3439
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9281972128869535e-05,
+      "loss": 2.3125,
+      "step": 3440
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9281406936323016e-05,
+      "loss": 3.0,
+      "step": 3441
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9280841525488816e-05,
+      "loss": 4.8281,
+      "step": 3442
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.928027589637333e-05,
+      "loss": 6.6719,
+      "step": 3443
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.927971004898296e-05,
+      "loss": 4.4688,
+      "step": 3444
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.927914398332413e-05,
+      "loss": 4.2188,
+      "step": 3445
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.927857769940323e-05,
+      "loss": 6.3438,
+      "step": 3446
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.927801119722667e-05,
+      "loss": 8.8281,
+      "step": 3447
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.927744447680087e-05,
+      "loss": 7.5312,
+      "step": 3448
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.927687753813224e-05,
+      "loss": 7.4531,
+      "step": 3449
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.927631038122721e-05,
+      "loss": 4.2812,
+      "step": 3450
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.92757430060922e-05,
+      "loss": 3.7188,
+      "step": 3451
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.927517541273361e-05,
+      "loss": 3.7188,
+      "step": 3452
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.92746076011579e-05,
+      "loss": 5.5938,
+      "step": 3453
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.927403957137148e-05,
+      "loss": 4.9062,
+      "step": 3454
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.927347132338077e-05,
+      "loss": 3.6562,
+      "step": 3455
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.927290285719223e-05,
+      "loss": 9.875,
+      "step": 3456
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.927233417281227e-05,
+      "loss": 7.4062,
+      "step": 3457
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.927176527024734e-05,
+      "loss": 3.4375,
+      "step": 3458
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9271196149503883e-05,
+      "loss": 9.0625,
+      "step": 3459
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.927062681058834e-05,
+      "loss": 5.7188,
+      "step": 3460
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9270057253507154e-05,
+      "loss": 4.5938,
+      "step": 3461
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.926948747826678e-05,
+      "loss": 4.9688,
+      "step": 3462
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.926891748487365e-05,
+      "loss": 3.875,
+      "step": 3463
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9268347273334246e-05,
+      "loss": 3.5156,
+      "step": 3464
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9267776843655e-05,
+      "loss": 5.0,
+      "step": 3465
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.926720619584237e-05,
+      "loss": 4.125,
+      "step": 3466
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9266635329902836e-05,
+      "loss": 2.0,
+      "step": 3467
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9266064245842844e-05,
+      "loss": 2.5938,
+      "step": 3468
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9265492943668866e-05,
+      "loss": 7.4062,
+      "step": 3469
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.926492142338736e-05,
+      "loss": 7.3438,
+      "step": 3470
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.926434968500481e-05,
+      "loss": 10.5312,
+      "step": 3471
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.926377772852769e-05,
+      "loss": 7.9688,
+      "step": 3472
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9263205553962455e-05,
+      "loss": 6.3125,
+      "step": 3473
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.926263316131561e-05,
+      "loss": 4.875,
+      "step": 3474
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.92620605505936e-05,
+      "loss": 9.9844,
+      "step": 3475
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.926148772180295e-05,
+      "loss": 10.5,
+      "step": 3476
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9260914674950114e-05,
+      "loss": 4.2188,
+      "step": 3477
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9260341410041595e-05,
+      "loss": 5.875,
+      "step": 3478
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9259767927083866e-05,
+      "loss": 4.4375,
+      "step": 3479
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9259194226083436e-05,
+      "loss": 5.9844,
+      "step": 3480
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.925862030704679e-05,
+      "loss": 2.0156,
+      "step": 3481
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9258046169980446e-05,
+      "loss": 5.4062,
+      "step": 3482
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9257471814890875e-05,
+      "loss": 5.3438,
+      "step": 3483
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.92568972417846e-05,
+      "loss": 5.5,
+      "step": 3484
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.925632245066811e-05,
+      "loss": 6.1719,
+      "step": 3485
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.925574744154792e-05,
+      "loss": 8.9922,
+      "step": 3486
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9255172214430544e-05,
+      "loss": 7.5312,
+      "step": 3487
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.925459676932249e-05,
+      "loss": 8.125,
+      "step": 3488
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.925402110623027e-05,
+      "loss": 5.9531,
+      "step": 3489
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.925344522516041e-05,
+      "loss": 7.1094,
+      "step": 3490
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.925286912611942e-05,
+      "loss": 7.5938,
+      "step": 3491
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9252292809113825e-05,
+      "loss": 4.7812,
+      "step": 3492
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9251716274150155e-05,
+      "loss": 2.2656,
+      "step": 3493
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.925113952123493e-05,
+      "loss": 5.5312,
+      "step": 3494
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.925056255037468e-05,
+      "loss": 5.8438,
+      "step": 3495
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.924998536157595e-05,
+      "loss": 4.25,
+      "step": 3496
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.924940795484525e-05,
+      "loss": 3.8438,
+      "step": 3497
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9248830330189135e-05,
+      "loss": 4.375,
+      "step": 3498
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.924825248761414e-05,
+      "loss": 5.875,
+      "step": 3499
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.92476744271268e-05,
+      "loss": 5.0469,
+      "step": 3500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9247096148733674e-05,
+      "loss": 3.5312,
+      "step": 3501
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.924651765244131e-05,
+      "loss": 3.7188,
+      "step": 3502
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.924593893825623e-05,
+      "loss": 3.875,
+      "step": 3503
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.924536000618501e-05,
+      "loss": 4.7812,
+      "step": 3504
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9244780856234205e-05,
+      "loss": 3.9375,
+      "step": 3505
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.924420148841035e-05,
+      "loss": 3.5781,
+      "step": 3506
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.924362190272003e-05,
+      "loss": 6.1094,
+      "step": 3507
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9243042099169796e-05,
+      "loss": 2.9531,
+      "step": 3508
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.92424620777662e-05,
+      "loss": 7.6562,
+      "step": 3509
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.924188183851583e-05,
+      "loss": 3.9219,
+      "step": 3510
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.924130138142524e-05,
+      "loss": 4.125,
+      "step": 3511
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.924072070650101e-05,
+      "loss": 3.5312,
+      "step": 3512
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9240139813749706e-05,
+      "loss": 5.2812,
+      "step": 3513
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9239558703177914e-05,
+      "loss": 4.625,
+      "step": 3514
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.92389773747922e-05,
+      "loss": 3.9375,
+      "step": 3515
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9238395828599166e-05,
+      "loss": 2.7656,
+      "step": 3516
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.923781406460537e-05,
+      "loss": 5.2031,
+      "step": 3517
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.923723208281741e-05,
+      "loss": 4.625,
+      "step": 3518
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9236649883241875e-05,
+      "loss": 2.4844,
+      "step": 3519
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9236067465885365e-05,
+      "loss": 3.0312,
+      "step": 3520
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.923548483075446e-05,
+      "loss": 5.9688,
+      "step": 3521
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.923490197785576e-05,
+      "loss": 4.3125,
+      "step": 3522
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.923431890719587e-05,
+      "loss": 2.0,
+      "step": 3523
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.923373561878138e-05,
+      "loss": 4.2188,
+      "step": 3524
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.92331521126189e-05,
+      "loss": 5.4688,
+      "step": 3525
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9232568388715034e-05,
+      "loss": 4.2656,
+      "step": 3526
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.92319844470764e-05,
+      "loss": 2.9688,
+      "step": 3527
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.923140028770959e-05,
+      "loss": 5.4062,
+      "step": 3528
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.923081591062124e-05,
+      "loss": 3.125,
+      "step": 3529
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.923023131581794e-05,
+      "loss": 4.3125,
+      "step": 3530
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.922964650330633e-05,
+      "loss": 4.7188,
+      "step": 3531
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.922906147309302e-05,
+      "loss": 2.4062,
+      "step": 3532
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.922847622518464e-05,
+      "loss": 5.0312,
+      "step": 3533
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.922789075958781e-05,
+      "loss": 5.1875,
+      "step": 3534
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.922730507630916e-05,
+      "loss": 5.9375,
+      "step": 3535
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9226719175355316e-05,
+      "loss": 5.9062,
+      "step": 3536
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9226133056732916e-05,
+      "loss": 5.5156,
+      "step": 3537
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.92255467204486e-05,
+      "loss": 9.1562,
+      "step": 3538
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9224960166509e-05,
+      "loss": 5.875,
+      "step": 3539
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.922437339492075e-05,
+      "loss": 5.4062,
+      "step": 3540
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.922378640569051e-05,
+      "loss": 4.6875,
+      "step": 3541
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9223199198824916e-05,
+      "loss": 4.2969,
+      "step": 3542
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9222611774330615e-05,
+      "loss": 2.4844,
+      "step": 3543
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.922202413221425e-05,
+      "loss": 4.0938,
+      "step": 3544
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9221436272482485e-05,
+      "loss": 4.9375,
+      "step": 3545
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.922084819514197e-05,
+      "loss": 3.5156,
+      "step": 3546
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.922025990019937e-05,
+      "loss": 1.75,
+      "step": 3547
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9219671387661346e-05,
+      "loss": 3.8594,
+      "step": 3548
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.921908265753454e-05,
+      "loss": 3.6094,
+      "step": 3549
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.921849370982564e-05,
+      "loss": 3.5,
+      "step": 3550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.92179045445413e-05,
+      "loss": 5.0312,
+      "step": 3551
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9217315161688204e-05,
+      "loss": 5.6875,
+      "step": 3552
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.921672556127301e-05,
+      "loss": 5.3281,
+      "step": 3553
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.92161357433024e-05,
+      "loss": 6.5312,
+      "step": 3554
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9215545707783056e-05,
+      "loss": 4.25,
+      "step": 3555
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.921495545472164e-05,
+      "loss": 4.0312,
+      "step": 3556
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9214364984124854e-05,
+      "loss": 3.9062,
+      "step": 3557
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.921377429599937e-05,
+      "loss": 3.875,
+      "step": 3558
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.921318339035189e-05,
+      "loss": 3.0156,
+      "step": 3559
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.921259226718908e-05,
+      "loss": 3.875,
+      "step": 3560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9212000926517655e-05,
+      "loss": 4.0312,
+      "step": 3561
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9211409368344305e-05,
+      "loss": 4.6875,
+      "step": 3562
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9210817592675716e-05,
+      "loss": 3.2188,
+      "step": 3563
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.921022559951859e-05,
+      "loss": 4.3125,
+      "step": 3564
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9209633388879645e-05,
+      "loss": 2.3594,
+      "step": 3565
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.920904096076557e-05,
+      "loss": 5.0312,
+      "step": 3566
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9208448315183074e-05,
+      "loss": 4.7812,
+      "step": 3567
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9207855452138874e-05,
+      "loss": 4.625,
+      "step": 3568
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.920726237163967e-05,
+      "loss": 2.8594,
+      "step": 3569
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.920666907369219e-05,
+      "loss": 5.25,
+      "step": 3570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9206075558303135e-05,
+      "loss": 4.5156,
+      "step": 3571
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9205481825479236e-05,
+      "loss": 6.2812,
+      "step": 3572
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.920488787522722e-05,
+      "loss": 4.9531,
+      "step": 3573
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.920429370755379e-05,
+      "loss": 3.6875,
+      "step": 3574
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.920369932246569e-05,
+      "loss": 5.1875,
+      "step": 3575
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.920310471996965e-05,
+      "loss": 2.7031,
+      "step": 3576
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9202509900072396e-05,
+      "loss": 3.0938,
+      "step": 3577
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.920191486278066e-05,
+      "loss": 10.75,
+      "step": 3578
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9201319608101174e-05,
+      "loss": 3.1562,
+      "step": 3579
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9200724136040686e-05,
+      "loss": 7.0781,
+      "step": 3580
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.920012844660594e-05,
+      "loss": 5.2812,
+      "step": 3581
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.919953253980368e-05,
+      "loss": 4.1172,
+      "step": 3582
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.919893641564063e-05,
+      "loss": 6.7812,
+      "step": 3583
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.919834007412358e-05,
+      "loss": 3.9219,
+      "step": 3584
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.919774351525924e-05,
+      "loss": 6.0,
+      "step": 3585
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.919714673905439e-05,
+      "loss": 4.2812,
+      "step": 3586
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.919654974551577e-05,
+      "loss": 5.8438,
+      "step": 3587
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9195952534650155e-05,
+      "loss": 4.125,
+      "step": 3588
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.919535510646429e-05,
+      "loss": 10.5,
+      "step": 3589
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.919475746096495e-05,
+      "loss": 5.0,
+      "step": 3590
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.919415959815889e-05,
+      "loss": 5.7344,
+      "step": 3591
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.919356151805289e-05,
+      "loss": 3.875,
+      "step": 3592
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.919296322065372e-05,
+      "loss": 2.9375,
+      "step": 3593
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.919236470596814e-05,
+      "loss": 6.4531,
+      "step": 3594
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.919176597400294e-05,
+      "loss": 5.1562,
+      "step": 3595
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.91911670247649e-05,
+      "loss": 4.8438,
+      "step": 3596
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.919056785826079e-05,
+      "loss": 6.3438,
+      "step": 3597
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.91899684744974e-05,
+      "loss": 4.5,
+      "step": 3598
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.918936887348151e-05,
+      "loss": 4.9844,
+      "step": 3599
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.918876905521991e-05,
+      "loss": 3.4531,
+      "step": 3600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9188169019719395e-05,
+      "loss": 3.375,
+      "step": 3601
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.918756876698676e-05,
+      "loss": 5.2031,
+      "step": 3602
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9186968297028795e-05,
+      "loss": 3.0312,
+      "step": 3603
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9186367609852295e-05,
+      "loss": 6.2812,
+      "step": 3604
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.918576670546407e-05,
+      "loss": 8.7344,
+      "step": 3605
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.918516558387091e-05,
+      "loss": 4.2344,
+      "step": 3606
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.918456424507963e-05,
+      "loss": 4.6406,
+      "step": 3607
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.918396268909704e-05,
+      "loss": 5.3125,
+      "step": 3608
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.918336091592994e-05,
+      "loss": 2.625,
+      "step": 3609
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9182758925585155e-05,
+      "loss": 7.4375,
+      "step": 3610
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.918215671806949e-05,
+      "loss": 4.875,
+      "step": 3611
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9181554293389775e-05,
+      "loss": 4.125,
+      "step": 3612
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.918095165155281e-05,
+      "loss": 5.5938,
+      "step": 3613
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9180348792565435e-05,
+      "loss": 3.125,
+      "step": 3614
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.917974571643447e-05,
+      "loss": 5.0312,
+      "step": 3615
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9179142423166744e-05,
+      "loss": 4.7969,
+      "step": 3616
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9178538912769075e-05,
+      "loss": 3.9062,
+      "step": 3617
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.917793518524832e-05,
+      "loss": 2.5781,
+      "step": 3618
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.917733124061129e-05,
+      "loss": 3.625,
+      "step": 3619
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.917672707886483e-05,
+      "loss": 4.3125,
+      "step": 3620
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.917612270001579e-05,
+      "loss": 4.6562,
+      "step": 3621
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9175518104070995e-05,
+      "loss": 3.0625,
+      "step": 3622
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.917491329103731e-05,
+      "loss": 5.6875,
+      "step": 3623
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.917430826092155e-05,
+      "loss": 6.375,
+      "step": 3624
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.91737030137306e-05,
+      "loss": 2.7969,
+      "step": 3625
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9173097549471294e-05,
+      "loss": 4.4062,
+      "step": 3626
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9172491868150495e-05,
+      "loss": 6.25,
+      "step": 3627
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.917188596977506e-05,
+      "loss": 7.75,
+      "step": 3628
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9171279854351825e-05,
+      "loss": 3.25,
+      "step": 3629
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.917067352188768e-05,
+      "loss": 6.7188,
+      "step": 3630
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.917006697238948e-05,
+      "loss": 4.125,
+      "step": 3631
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.916946020586409e-05,
+      "loss": 5.0625,
+      "step": 3632
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.916885322231839e-05,
+      "loss": 6.3594,
+      "step": 3633
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.916824602175923e-05,
+      "loss": 3.4688,
+      "step": 3634
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.91676386041935e-05,
+      "loss": 5.6875,
+      "step": 3635
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.916703096962807e-05,
+      "loss": 3.9375,
+      "step": 3636
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9166423118069835e-05,
+      "loss": 3.7812,
+      "step": 3637
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9165815049525655e-05,
+      "loss": 6.0938,
+      "step": 3638
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.916520676400242e-05,
+      "loss": 3.375,
+      "step": 3639
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9164598261507025e-05,
+      "loss": 3.4688,
+      "step": 3640
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.916398954204636e-05,
+      "loss": 6.0469,
+      "step": 3641
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.91633806056273e-05,
+      "loss": 4.2812,
+      "step": 3642
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9162771452256747e-05,
+      "loss": 2.8438,
+      "step": 3643
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.916216208194161e-05,
+      "loss": 3.5625,
+      "step": 3644
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9161552494688766e-05,
+      "loss": 3.9844,
+      "step": 3645
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9160942690505136e-05,
+      "loss": 3.6562,
+      "step": 3646
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.916033266939761e-05,
+      "loss": 4.8438,
+      "step": 3647
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9159722431373095e-05,
+      "loss": 4.1562,
+      "step": 3648
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9159111976438513e-05,
+      "loss": 3.25,
+      "step": 3649
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.915850130460076e-05,
+      "loss": 3.9531,
+      "step": 3650
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.915789041586676e-05,
+      "loss": 4.6562,
+      "step": 3651
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9157279310243416e-05,
+      "loss": 4.5938,
+      "step": 3652
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.915666798773766e-05,
+      "loss": 4.1406,
+      "step": 3653
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.915605644835641e-05,
+      "loss": 3.625,
+      "step": 3654
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9155444692106584e-05,
+      "loss": 3.2812,
+      "step": 3655
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9154832718995116e-05,
+      "loss": 6.0,
+      "step": 3656
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9154220529028915e-05,
+      "loss": 5.0938,
+      "step": 3657
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9153608122214934e-05,
+      "loss": 7.0,
+      "step": 3658
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.91529954985601e-05,
+      "loss": 4.3125,
+      "step": 3659
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.915238265807134e-05,
+      "loss": 3.4375,
+      "step": 3660
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9151769600755606e-05,
+      "loss": 3.1562,
+      "step": 3661
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9151156326619826e-05,
+      "loss": 3.0,
+      "step": 3662
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.915054283567095e-05,
+      "loss": 6.1719,
+      "step": 3663
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.9149929127915916e-05,
+      "loss": 3.9375,
+      "step": 3664
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.914931520336168e-05,
+      "loss": 4.1562,
+      "step": 3665
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.914870106201519e-05,
+      "loss": 4.2812,
+      "step": 3666
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.91480867038834e-05,
+      "loss": 2.9531,
+      "step": 3667
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.914747212897326e-05,
+      "loss": 4.125,
+      "step": 3668
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.914685733729173e-05,
+      "loss": 8.0156,
+      "step": 3669
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.914624232884578e-05,
+      "loss": 4.0,
+      "step": 3670
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9145627103642356e-05,
+      "loss": 6.7188,
+      "step": 3671
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.914501166168843e-05,
+      "loss": 3.0312,
+      "step": 3672
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9144396002990975e-05,
+      "loss": 6.375,
+      "step": 3673
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.914378012755695e-05,
+      "loss": 4.5781,
+      "step": 3674
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.914316403539334e-05,
+      "loss": 4.4219,
+      "step": 3675
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.914254772650711e-05,
+      "loss": 2.5,
+      "step": 3676
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.914193120090525e-05,
+      "loss": 4.5938,
+      "step": 3677
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.914131445859472e-05,
+      "loss": 4.125,
+      "step": 3678
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.914069749958252e-05,
+      "loss": 3.8594,
+      "step": 3679
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9140080323875624e-05,
+      "loss": 3.7656,
+      "step": 3680
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.913946293148102e-05,
+      "loss": 3.2188,
+      "step": 3681
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9138845322405705e-05,
+      "loss": 3.8594,
+      "step": 3682
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.913822749665666e-05,
+      "loss": 4.4375,
+      "step": 3683
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9137609454240894e-05,
+      "loss": 3.4062,
+      "step": 3684
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9136991195165394e-05,
+      "loss": 4.2812,
+      "step": 3685
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.913637271943716e-05,
+      "loss": 5.25,
+      "step": 3686
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.91357540270632e-05,
+      "loss": 4.5781,
+      "step": 3687
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9135135118050514e-05,
+      "loss": 6.6719,
+      "step": 3688
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.91345159924061e-05,
+      "loss": 3.2812,
+      "step": 3689
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.913389665013698e-05,
+      "loss": 2.5,
+      "step": 3690
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.913327709125016e-05,
+      "loss": 2.0156,
+      "step": 3691
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9132657315752654e-05,
+      "loss": 2.7188,
+      "step": 3692
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.913203732365148e-05,
+      "loss": 2.75,
+      "step": 3693
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9131417114953656e-05,
+      "loss": 7.7344,
+      "step": 3694
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.913079668966621e-05,
+      "loss": 5.75,
+      "step": 3695
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.913017604779615e-05,
+      "loss": 4.1875,
+      "step": 3696
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9129555189350525e-05,
+      "loss": 3.375,
+      "step": 3697
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.9128934114336346e-05,
+      "loss": 4.7812,
+      "step": 3698
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.912831282276065e-05,
+      "loss": 2.6562,
+      "step": 3699
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9127691314630466e-05,
+      "loss": 4.5312,
+      "step": 3700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9127069589952846e-05,
+      "loss": 4.1562,
+      "step": 3701
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.912644764873481e-05,
+      "loss": 6.0625,
+      "step": 3702
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9125825490983406e-05,
+      "loss": 4.125,
+      "step": 3703
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9125203116705674e-05,
+      "loss": 2.8438,
+      "step": 3704
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.912458052590867e-05,
+      "loss": 2.9688,
+      "step": 3705
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.912395771859944e-05,
+      "loss": 4.9688,
+      "step": 3706
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.912333469478502e-05,
+      "loss": 8.2344,
+      "step": 3707
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.912271145447248e-05,
+      "loss": 3.5469,
+      "step": 3708
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9122087997668874e-05,
+      "loss": 3.7188,
+      "step": 3709
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.912146432438126e-05,
+      "loss": 3.6719,
+      "step": 3710
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.912084043461669e-05,
+      "loss": 2.7812,
+      "step": 3711
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.912021632838222e-05,
+      "loss": 6.4844,
+      "step": 3712
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.911959200568495e-05,
+      "loss": 7.2188,
+      "step": 3713
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.911896746653191e-05,
+      "loss": 6.625,
+      "step": 3714
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.911834271093019e-05,
+      "loss": 5.9375,
+      "step": 3715
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9117717738886863e-05,
+      "loss": 7.1406,
+      "step": 3716
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9117092550409e-05,
+      "loss": 4.0625,
+      "step": 3717
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9116467145503686e-05,
+      "loss": 4.375,
+      "step": 3718
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9115841524177986e-05,
+      "loss": 5.25,
+      "step": 3719
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.911521568643899e-05,
+      "loss": 3.2188,
+      "step": 3720
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9114589632293795e-05,
+      "loss": 6.7188,
+      "step": 3721
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.911396336174947e-05,
+      "loss": 4.9062,
+      "step": 3722
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.911333687481312e-05,
+      "loss": 4.8906,
+      "step": 3723
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.911271017149183e-05,
+      "loss": 4.6562,
+      "step": 3724
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.91120832517927e-05,
+      "loss": 5.2031,
+      "step": 3725
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.911145611572282e-05,
+      "loss": 3.8906,
+      "step": 3726
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.9110828763289295e-05,
+      "loss": 7.25,
+      "step": 3727
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.911020119449922e-05,
+      "loss": 1.25,
+      "step": 3728
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.910957340935971e-05,
+      "loss": 3.7812,
+      "step": 3729
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.910894540787787e-05,
+      "loss": 7.0,
+      "step": 3730
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.910831719006081e-05,
+      "loss": 6.5625,
+      "step": 3731
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.910768875591563e-05,
+      "loss": 4.4688,
+      "step": 3732
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.910706010544947e-05,
+      "loss": 5.9219,
+      "step": 3733
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9106431238669416e-05,
+      "loss": 3.7812,
+      "step": 3734
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.910580215558261e-05,
+      "loss": 3.1562,
+      "step": 3735
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9105172856196166e-05,
+      "loss": 1.9531,
+      "step": 3736
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9104543340517215e-05,
+      "loss": 4.2188,
+      "step": 3737
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.910391360855287e-05,
+      "loss": 5.2812,
+      "step": 3738
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9103283660310276e-05,
+      "loss": 5.4375,
+      "step": 3739
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.910265349579655e-05,
+      "loss": 7.25,
+      "step": 3740
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.910202311501884e-05,
+      "loss": 5.125,
+      "step": 3741
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.910139251798428e-05,
+      "loss": 2.9688,
+      "step": 3742
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9100761704699994e-05,
+      "loss": 4.6719,
+      "step": 3743
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.910013067517314e-05,
+      "loss": 8.25,
+      "step": 3744
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.909949942941086e-05,
+      "loss": 5.2812,
+      "step": 3745
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.909886796742029e-05,
+      "loss": 6.2656,
+      "step": 3746
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.909823628920859e-05,
+      "loss": 3.6562,
+      "step": 3747
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.909760439478291e-05,
+      "loss": 3.3438,
+      "step": 3748
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9096972284150394e-05,
+      "loss": 3.125,
+      "step": 3749
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.909633995731821e-05,
+      "loss": 2.9375,
+      "step": 3750
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9095707414293516e-05,
+      "loss": 3.8438,
+      "step": 3751
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9095074655083464e-05,
+      "loss": 2.7188,
+      "step": 3752
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.909444167969523e-05,
+      "loss": 3.4062,
+      "step": 3753
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9093808488135963e-05,
+      "loss": 3.3438,
+      "step": 3754
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.909317508041285e-05,
+      "loss": 4.9062,
+      "step": 3755
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.909254145653304e-05,
+      "loss": 3.8438,
+      "step": 3756
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.909190761650373e-05,
+      "loss": 2.9844,
+      "step": 3757
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.9091273560332084e-05,
+      "loss": 3.7188,
+      "step": 3758
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.909063928802528e-05,
+      "loss": 4.125,
+      "step": 3759
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.90900047995905e-05,
+      "loss": 2.6875,
+      "step": 3760
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.908937009503493e-05,
+      "loss": 6.2344,
+      "step": 3761
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.908873517436575e-05,
+      "loss": 6.8438,
+      "step": 3762
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.908810003759015e-05,
+      "loss": 4.4844,
+      "step": 3763
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.908746468471533e-05,
+      "loss": 3.4062,
+      "step": 3764
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.908682911574846e-05,
+      "loss": 3.125,
+      "step": 3765
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.908619333069676e-05,
+      "loss": 4.0625,
+      "step": 3766
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.9085557329567414e-05,
+      "loss": 5.1406,
+      "step": 3767
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.908492111236763e-05,
+      "loss": 6.2812,
+      "step": 3768
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.90842846791046e-05,
+      "loss": 5.4375,
+      "step": 3769
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.908364802978554e-05,
+      "loss": 5.625,
+      "step": 3770
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.9083011164417655e-05,
+      "loss": 6.3438,
+      "step": 3771
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.9082374083008144e-05,
+      "loss": 5.1094,
+      "step": 3772
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.908173678556424e-05,
+      "loss": 4.9062,
+      "step": 3773
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.9081099272093144e-05,
+      "loss": 7.0156,
+      "step": 3774
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.9080461542602075e-05,
+      "loss": 5.0938,
+      "step": 3775
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.907982359709825e-05,
+      "loss": 4.0,
+      "step": 3776
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.90791854355889e-05,
+      "loss": 5.9219,
+      "step": 3777
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.907854705808124e-05,
+      "loss": 5.5156,
+      "step": 3778
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.907790846458251e-05,
+      "loss": 3.6875,
+      "step": 3779
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.9077269655099924e-05,
+      "loss": 3.2188,
+      "step": 3780
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.907663062964072e-05,
+      "loss": 6.4062,
+      "step": 3781
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.9075991388212146e-05,
+      "loss": 3.5781,
+      "step": 3782
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.907535193082142e-05,
+      "loss": 1.75,
+      "step": 3783
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.907471225747579e-05,
+      "loss": 5.0312,
+      "step": 3784
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.90740723681825e-05,
+      "loss": 4.4062,
+      "step": 3785
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.907343226294878e-05,
+      "loss": 2.2188,
+      "step": 3786
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.9072791941781895e-05,
+      "loss": 3.875,
+      "step": 3787
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.907215140468909e-05,
+      "loss": 6.4375,
+      "step": 3788
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.90715106516776e-05,
+      "loss": 6.4844,
+      "step": 3789
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.90708696827547e-05,
+      "loss": 6.8281,
+      "step": 3790
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.907022849792765e-05,
+      "loss": 4.4219,
+      "step": 3791
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9069587097203686e-05,
+      "loss": 5.7969,
+      "step": 3792
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.906894548059008e-05,
+      "loss": 3.25,
+      "step": 3793
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.90683036480941e-05,
+      "loss": 5.375,
+      "step": 3794
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.906766159972301e-05,
+      "loss": 4.0781,
+      "step": 3795
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.906701933548408e-05,
+      "loss": 5.1406,
+      "step": 3796
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9066376855384574e-05,
+      "loss": 3.5781,
+      "step": 3797
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9065734159431774e-05,
+      "loss": 3.2344,
+      "step": 3798
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.906509124763295e-05,
+      "loss": 3.0312,
+      "step": 3799
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9064448119995386e-05,
+      "loss": 3.2969,
+      "step": 3800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9063804776526366e-05,
+      "loss": 2.9062,
+      "step": 3801
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9063161217233155e-05,
+      "loss": 4.0469,
+      "step": 3802
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9062517442123064e-05,
+      "loss": 7.3906,
+      "step": 3803
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9061873451203364e-05,
+      "loss": 8.2188,
+      "step": 3804
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9061229244481346e-05,
+      "loss": 6.2031,
+      "step": 3805
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9060584821964316e-05,
+      "loss": 4.4688,
+      "step": 3806
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.905994018365955e-05,
+      "loss": 3.9844,
+      "step": 3807
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.905929532957437e-05,
+      "loss": 2.625,
+      "step": 3808
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.905865025971605e-05,
+      "loss": 4.7031,
+      "step": 3809
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.905800497409192e-05,
+      "loss": 6.125,
+      "step": 3810
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.905735947270927e-05,
+      "loss": 3.2969,
+      "step": 3811
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.90567137555754e-05,
+      "loss": 6.0469,
+      "step": 3812
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.905606782269764e-05,
+      "loss": 5.7188,
+      "step": 3813
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.905542167408329e-05,
+      "loss": 3.375,
+      "step": 3814
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.905477530973967e-05,
+      "loss": 3.6875,
+      "step": 3815
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.905412872967409e-05,
+      "loss": 5.1875,
+      "step": 3816
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.905348193389388e-05,
+      "loss": 8.1562,
+      "step": 3817
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9052834922406357e-05,
+      "loss": 2.2188,
+      "step": 3818
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9052187695218856e-05,
+      "loss": 4.9688,
+      "step": 3819
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.9051540252338684e-05,
+      "loss": 3.625,
+      "step": 3820
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.9050892593773187e-05,
+      "loss": 6.2188,
+      "step": 3821
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.905024471952969e-05,
+      "loss": 3.8125,
+      "step": 3822
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.904959662961552e-05,
+      "loss": 4.9062,
+      "step": 3823
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.904894832403804e-05,
+      "loss": 5.125,
+      "step": 3824
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.904829980280457e-05,
+      "loss": 5.6562,
+      "step": 3825
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.9047651065922455e-05,
+      "loss": 4.4375,
+      "step": 3826
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.904700211339904e-05,
+      "loss": 2.9375,
+      "step": 3827
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.904635294524167e-05,
+      "loss": 5.7969,
+      "step": 3828
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.90457035614577e-05,
+      "loss": 4.9219,
+      "step": 3829
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.904505396205448e-05,
+      "loss": 3.0938,
+      "step": 3830
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.9044404147039356e-05,
+      "loss": 4.4219,
+      "step": 3831
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.90437541164197e-05,
+      "loss": 4.25,
+      "step": 3832
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.904310387020285e-05,
+      "loss": 3.2188,
+      "step": 3833
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.9042453408396186e-05,
+      "loss": 4.1875,
+      "step": 3834
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.904180273100707e-05,
+      "loss": 3.8125,
+      "step": 3835
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.904115183804286e-05,
+      "loss": 2.8125,
+      "step": 3836
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.904050072951093e-05,
+      "loss": 2.9844,
+      "step": 3837
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.9039849405418645e-05,
+      "loss": 2.2188,
+      "step": 3838
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.903919786577339e-05,
+      "loss": 2.5625,
+      "step": 3839
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.903854611058253e-05,
+      "loss": 3.7344,
+      "step": 3840
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.903789413985346e-05,
+      "loss": 3.2344,
+      "step": 3841
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.903724195359354e-05,
+      "loss": 4.4688,
+      "step": 3842
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.9036589551810166e-05,
+      "loss": 3.6094,
+      "step": 3843
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.903593693451072e-05,
+      "loss": 5.375,
+      "step": 3844
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.9035284101702594e-05,
+      "loss": 5.1406,
+      "step": 3845
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.9034631053393175e-05,
+      "loss": 3.8594,
+      "step": 3846
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.903397778958986e-05,
+      "loss": 4.9062,
+      "step": 3847
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.903332431030004e-05,
+      "loss": 7.625,
+      "step": 3848
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.903267061553111e-05,
+      "loss": 5.5625,
+      "step": 3849
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.9032016705290486e-05,
+      "loss": 4.6875,
+      "step": 3850
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.9031362579585563e-05,
+      "loss": 4.3125,
+      "step": 3851
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.9030708238423736e-05,
+      "loss": 5.4375,
+      "step": 3852
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.9030053681812424e-05,
+      "loss": 5.7344,
+      "step": 3853
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.9029398909759034e-05,
+      "loss": 3.7188,
+      "step": 3854
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.9028743922270986e-05,
+      "loss": 2.3125,
+      "step": 3855
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.902808871935568e-05,
+      "loss": 4.0469,
+      "step": 3856
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.902743330102054e-05,
+      "loss": 2.7969,
+      "step": 3857
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.9026777667273e-05,
+      "loss": 5.25,
+      "step": 3858
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.9026121818120473e-05,
+      "loss": 7.0938,
+      "step": 3859
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.902546575357037e-05,
+      "loss": 5.8906,
+      "step": 3860
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.902480947363014e-05,
+      "loss": 5.1406,
+      "step": 3861
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.90241529783072e-05,
+      "loss": 3.625,
+      "step": 3862
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.9023496267608986e-05,
+      "loss": 2.9688,
+      "step": 3863
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.902283934154293e-05,
+      "loss": 3.75,
+      "step": 3864
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.902218220011647e-05,
+      "loss": 3.8906,
+      "step": 3865
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.902152484333705e-05,
+      "loss": 4.0,
+      "step": 3866
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.902086727121211e-05,
+      "loss": 2.75,
+      "step": 3867
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.9020209483749096e-05,
+      "loss": 3.8125,
+      "step": 3868
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.901955148095545e-05,
+      "loss": 5.1406,
+      "step": 3869
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.901889326283862e-05,
+      "loss": 3.625,
+      "step": 3870
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.901823482940607e-05,
+      "loss": 6.1562,
+      "step": 3871
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.9017576180665234e-05,
+      "loss": 6.9531,
+      "step": 3872
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.901691731662359e-05,
+      "loss": 5.4531,
+      "step": 3873
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.901625823728859e-05,
+      "loss": 5.5312,
+      "step": 3874
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.9015598942667684e-05,
+      "loss": 2.125,
+      "step": 3875
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.901493943276835e-05,
+      "loss": 5.25,
+      "step": 3876
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.901427970759805e-05,
+      "loss": 4.9375,
+      "step": 3877
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.901361976716425e-05,
+      "loss": 6.0781,
+      "step": 3878
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.901295961147443e-05,
+      "loss": 3.2812,
+      "step": 3879
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.901229924053604e-05,
+      "loss": 4.4531,
+      "step": 3880
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.9011638654356596e-05,
+      "loss": 4.6562,
+      "step": 3881
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.901097785294354e-05,
+      "loss": 5.0625,
+      "step": 3882
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.901031683630437e-05,
+      "loss": 4.3125,
+      "step": 3883
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.900965560444657e-05,
+      "loss": 3.0156,
+      "step": 3884
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.900899415737762e-05,
+      "loss": 4.3438,
+      "step": 3885
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.900833249510501e-05,
+      "loss": 3.4062,
+      "step": 3886
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.900767061763623e-05,
+      "loss": 7.7656,
+      "step": 3887
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.9007008524978786e-05,
+      "loss": 1.9688,
+      "step": 3888
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.900634621714015e-05,
+      "loss": 3.5,
+      "step": 3889
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.900568369412784e-05,
+      "loss": 3.125,
+      "step": 3890
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.900502095594934e-05,
+      "loss": 5.6875,
+      "step": 3891
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.9004358002612176e-05,
+      "loss": 3.3594,
+      "step": 3892
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.900369483412383e-05,
+      "loss": 4.3125,
+      "step": 3893
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.900303145049182e-05,
+      "loss": 3.5156,
+      "step": 3894
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.9002367851723654e-05,
+      "loss": 3.2031,
+      "step": 3895
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.900170403782685e-05,
+      "loss": 4.375,
+      "step": 3896
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.9001040008808917e-05,
+      "loss": 3.3125,
+      "step": 3897
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.900037576467738e-05,
+      "loss": 3.7656,
+      "step": 3898
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.899971130543975e-05,
+      "loss": 4.8906,
+      "step": 3899
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.899904663110356e-05,
+      "loss": 3.1406,
+      "step": 3900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.8998381741676317e-05,
+      "loss": 2.4062,
+      "step": 3901
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.8997716637165564e-05,
+      "loss": 2.4375,
+      "step": 3902
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.8997051317578834e-05,
+      "loss": 5.4062,
+      "step": 3903
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.899638578292364e-05,
+      "loss": 3.0938,
+      "step": 3904
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.899572003320753e-05,
+      "loss": 4.2969,
+      "step": 3905
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.899505406843805e-05,
+      "loss": 7.625,
+      "step": 3906
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.899438788862272e-05,
+      "loss": 5.6094,
+      "step": 3907
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.899372149376909e-05,
+      "loss": 4.625,
+      "step": 3908
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.8993054883884707e-05,
+      "loss": 5.625,
+      "step": 3909
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.899238805897712e-05,
+      "loss": 5.5312,
+      "step": 3910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.899172101905387e-05,
+      "loss": 4.0,
+      "step": 3911
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.899105376412251e-05,
+      "loss": 4.7969,
+      "step": 3912
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8990386294190596e-05,
+      "loss": 6.5,
+      "step": 3913
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.898971860926569e-05,
+      "loss": 2.7812,
+      "step": 3914
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8989050709355344e-05,
+      "loss": 3.0469,
+      "step": 3915
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.898838259446712e-05,
+      "loss": 3.8125,
+      "step": 3916
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.898771426460859e-05,
+      "loss": 4.4375,
+      "step": 3917
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.898704571978731e-05,
+      "loss": 3.7812,
+      "step": 3918
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.898637696001085e-05,
+      "loss": 2.6406,
+      "step": 3919
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8985707985286784e-05,
+      "loss": 5.5,
+      "step": 3920
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.898503879562269e-05,
+      "loss": 6.125,
+      "step": 3921
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8984369391026134e-05,
+      "loss": 5.4062,
+      "step": 3922
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.89836997715047e-05,
+      "loss": 3.625,
+      "step": 3923
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.898302993706597e-05,
+      "loss": 9.0625,
+      "step": 3924
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.898235988771753e-05,
+      "loss": 5.1719,
+      "step": 3925
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.898168962346695e-05,
+      "loss": 4.1719,
+      "step": 3926
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.898101914432184e-05,
+      "loss": 5.2031,
+      "step": 3927
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.898034845028978e-05,
+      "loss": 4.5,
+      "step": 3928
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8979677541378356e-05,
+      "loss": 3.0938,
+      "step": 3929
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8979006417595175e-05,
+      "loss": 5.0938,
+      "step": 3930
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.897833507894783e-05,
+      "loss": 6.1562,
+      "step": 3931
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.897766352544392e-05,
+      "loss": 2.375,
+      "step": 3932
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.897699175709106e-05,
+      "loss": 5.7188,
+      "step": 3933
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.897631977389683e-05,
+      "loss": 4.5312,
+      "step": 3934
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.897564757586886e-05,
+      "loss": 4.7656,
+      "step": 3935
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8974975163014754e-05,
+      "loss": 5.4688,
+      "step": 3936
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.897430253534212e-05,
+      "loss": 4.0,
+      "step": 3937
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.897362969285857e-05,
+      "loss": 3.6719,
+      "step": 3938
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8972956635571735e-05,
+      "loss": 4.8906,
+      "step": 3939
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.897228336348923e-05,
+      "loss": 3.9219,
+      "step": 3940
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.8971609876618666e-05,
+      "loss": 3.5625,
+      "step": 3941
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.897093617496769e-05,
+      "loss": 6.5938,
+      "step": 3942
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.89702622585439e-05,
+      "loss": 3.8125,
+      "step": 3943
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.896958812735494e-05,
+      "loss": 4.2344,
+      "step": 3944
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8968913781408454e-05,
+      "loss": 6.0625,
+      "step": 3945
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.896823922071206e-05,
+      "loss": 3.4531,
+      "step": 3946
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.896756444527339e-05,
+      "loss": 6.1875,
+      "step": 3947
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8966889455100114e-05,
+      "loss": 6.0781,
+      "step": 3948
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8966214250199836e-05,
+      "loss": 3.6562,
+      "step": 3949
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8965538830580215e-05,
+      "loss": 3.6562,
+      "step": 3950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.896486319624891e-05,
+      "loss": 4.2344,
+      "step": 3951
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.896418734721355e-05,
+      "loss": 2.8281,
+      "step": 3952
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8963511283481806e-05,
+      "loss": 3.5312,
+      "step": 3953
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.896283500506131e-05,
+      "loss": 3.0781,
+      "step": 3954
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8962158511959733e-05,
+      "loss": 3.6562,
+      "step": 3955
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.896148180418473e-05,
+      "loss": 2.875,
+      "step": 3956
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.896080488174397e-05,
+      "loss": 3.1875,
+      "step": 3957
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.89601277446451e-05,
+      "loss": 5.5781,
+      "step": 3958
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8959450392895794e-05,
+      "loss": 3.5625,
+      "step": 3959
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.895877282650373e-05,
+      "loss": 5.9062,
+      "step": 3960
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8958095045476564e-05,
+      "loss": 6.5312,
+      "step": 3961
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.895741704982198e-05,
+      "loss": 7.25,
+      "step": 3962
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8956738839547646e-05,
+      "loss": 4.9688,
+      "step": 3963
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8956060414661234e-05,
+      "loss": 7.625,
+      "step": 3964
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.895538177517044e-05,
+      "loss": 6.5625,
+      "step": 3965
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8954702921082946e-05,
+      "loss": 5.9062,
+      "step": 3966
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.895402385240643e-05,
+      "loss": 6.4375,
+      "step": 3967
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.895334456914858e-05,
+      "loss": 3.9062,
+      "step": 3968
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8952665071317086e-05,
+      "loss": 7.8281,
+      "step": 3969
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.895198535891965e-05,
+      "loss": 7.9375,
+      "step": 3970
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.8951305431963955e-05,
+      "loss": 6.9062,
+      "step": 3971
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.89506252904577e-05,
+      "loss": 5.6562,
+      "step": 3972
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.8949944934408595e-05,
+      "loss": 5.9688,
+      "step": 3973
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.894926436382433e-05,
+      "loss": 5.0625,
+      "step": 3974
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.894858357871263e-05,
+      "loss": 2.625,
+      "step": 3975
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.8947902579081175e-05,
+      "loss": 6.4375,
+      "step": 3976
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.894722136493769e-05,
+      "loss": 3.3438,
+      "step": 3977
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.894653993628989e-05,
+      "loss": 4.25,
+      "step": 3978
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.894585829314548e-05,
+      "loss": 3.875,
+      "step": 3979
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.894517643551218e-05,
+      "loss": 2.8125,
+      "step": 3980
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.894449436339771e-05,
+      "loss": 5.5625,
+      "step": 3981
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.894381207680979e-05,
+      "loss": 3.6562,
+      "step": 3982
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.894312957575616e-05,
+      "loss": 4.375,
+      "step": 3983
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.894244686024452e-05,
+      "loss": 5.6875,
+      "step": 3984
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.8941763930282624e-05,
+      "loss": 7.3594,
+      "step": 3985
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.894108078587818e-05,
+      "loss": 5.4062,
+      "step": 3986
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.894039742703895e-05,
+      "loss": 6.9375,
+      "step": 3987
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.8939713853772644e-05,
+      "loss": 2.8125,
+      "step": 3988
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.893903006608701e-05,
+      "loss": 5.25,
+      "step": 3989
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.8938346063989794e-05,
+      "loss": 4.4688,
+      "step": 3990
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.893766184748874e-05,
+      "loss": 2.7812,
+      "step": 3991
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.893697741659159e-05,
+      "loss": 5.2656,
+      "step": 3992
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.893629277130608e-05,
+      "loss": 5.5625,
+      "step": 3993
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.893560791163999e-05,
+      "loss": 4.1875,
+      "step": 3994
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.893492283760105e-05,
+      "loss": 3.4688,
+      "step": 3995
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.8934237549197025e-05,
+      "loss": 3.1562,
+      "step": 3996
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.893355204643567e-05,
+      "loss": 2.625,
+      "step": 3997
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.893286632932475e-05,
+      "loss": 3.0,
+      "step": 3998
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.893218039787203e-05,
+      "loss": 3.6094,
+      "step": 3999
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.893149425208527e-05,
+      "loss": 3.1562,
+      "step": 4000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.893080789197223e-05,
+      "loss": 4.7031,
+      "step": 4001
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.893012131754069e-05,
+      "loss": 4.8125,
+      "step": 4002
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8929434528798434e-05,
+      "loss": 3.9688,
+      "step": 4003
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8928747525753216e-05,
+      "loss": 5.4531,
+      "step": 4004
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.892806030841282e-05,
+      "loss": 4.0938,
+      "step": 4005
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.892737287678503e-05,
+      "loss": 3.8438,
+      "step": 4006
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8926685230877634e-05,
+      "loss": 3.3438,
+      "step": 4007
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8925997370698404e-05,
+      "loss": 5.125,
+      "step": 4008
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.892530929625514e-05,
+      "loss": 9.0781,
+      "step": 4009
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.892462100755562e-05,
+      "loss": 2.7031,
+      "step": 4010
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.892393250460765e-05,
+      "loss": 1.5938,
+      "step": 4011
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.892324378741902e-05,
+      "loss": 5.2188,
+      "step": 4012
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.89225548559975e-05,
+      "loss": 2.375,
+      "step": 4013
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.892186571035094e-05,
+      "loss": 7.1875,
+      "step": 4014
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.89211763504871e-05,
+      "loss": 4.7812,
+      "step": 4015
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.892048677641381e-05,
+      "loss": 5.5625,
+      "step": 4016
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8919796988138853e-05,
+      "loss": 4.1562,
+      "step": 4017
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8919106985670056e-05,
+      "loss": 3.9062,
+      "step": 4018
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.891841676901523e-05,
+      "loss": 6.1406,
+      "step": 4019
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.891772633818218e-05,
+      "loss": 2.4688,
+      "step": 4020
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8917035693178726e-05,
+      "loss": 4.5938,
+      "step": 4021
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8916344834012695e-05,
+      "loss": 3.4375,
+      "step": 4022
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.89156537606919e-05,
+      "loss": 4.4531,
+      "step": 4023
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.891496247322417e-05,
+      "loss": 3.1562,
+      "step": 4024
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8914270971617324e-05,
+      "loss": 8.2969,
+      "step": 4025
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.891357925587919e-05,
+      "loss": 5.5781,
+      "step": 4026
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8912887326017604e-05,
+      "loss": 3.7969,
+      "step": 4027
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.891219518204041e-05,
+      "loss": 6.5,
+      "step": 4028
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.891150282395541e-05,
+      "loss": 7.0312,
+      "step": 4029
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.891081025177049e-05,
+      "loss": 3.0,
+      "step": 4030
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.891011746549345e-05,
+      "loss": 3.8281,
+      "step": 4031
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.8909424465132154e-05,
+      "loss": 5.4844,
+      "step": 4032
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.890873125069444e-05,
+      "loss": 4.7969,
+      "step": 4033
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8908037822188165e-05,
+      "loss": 3.9219,
+      "step": 4034
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.890734417962116e-05,
+      "loss": 2.9688,
+      "step": 4035
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.89066503230013e-05,
+      "loss": 2.2188,
+      "step": 4036
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.890595625233643e-05,
+      "loss": 3.3906,
+      "step": 4037
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.89052619676344e-05,
+      "loss": 5.1094,
+      "step": 4038
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.890456746890309e-05,
+      "loss": 5.1562,
+      "step": 4039
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8903872756150346e-05,
+      "loss": 3.0,
+      "step": 4040
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.890317782938404e-05,
+      "loss": 2.4062,
+      "step": 4041
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8902482688612036e-05,
+      "loss": 3.3438,
+      "step": 4042
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.89017873338422e-05,
+      "loss": 1.8125,
+      "step": 4043
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8901091765082416e-05,
+      "loss": 3.7031,
+      "step": 4044
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8900395982340556e-05,
+      "loss": 4.5,
+      "step": 4045
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.889969998562448e-05,
+      "loss": 4.6562,
+      "step": 4046
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8899003774942085e-05,
+      "loss": 11.75,
+      "step": 4047
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8898307350301255e-05,
+      "loss": 2.9531,
+      "step": 4048
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8897610711709865e-05,
+      "loss": 5.625,
+      "step": 4049
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.88969138591758e-05,
+      "loss": 3.875,
+      "step": 4050
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.889621679270696e-05,
+      "loss": 7.0938,
+      "step": 4051
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.889551951231123e-05,
+      "loss": 4.75,
+      "step": 4052
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8894822017996495e-05,
+      "loss": 4.4062,
+      "step": 4053
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.889412430977067e-05,
+      "loss": 5.0312,
+      "step": 4054
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.889342638764164e-05,
+      "loss": 4.9219,
+      "step": 4055
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8892728251617315e-05,
+      "loss": 3.7031,
+      "step": 4056
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8892029901705596e-05,
+      "loss": 4.75,
+      "step": 4057
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.889133133791438e-05,
+      "loss": 3.7812,
+      "step": 4058
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.889063256025159e-05,
+      "loss": 3.3125,
+      "step": 4059
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8889933568725126e-05,
+      "loss": 3.6562,
+      "step": 4060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.888923436334291e-05,
+      "loss": 2.4062,
+      "step": 4061
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.888853494411285e-05,
+      "loss": 3.375,
+      "step": 4062
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.8887835311042865e-05,
+      "loss": 4.1875,
+      "step": 4063
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.888713546414088e-05,
+      "loss": 4.75,
+      "step": 4064
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.888643540341482e-05,
+      "loss": 3.5625,
+      "step": 4065
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.88857351288726e-05,
+      "loss": 2.5625,
+      "step": 4066
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.888503464052216e-05,
+      "loss": 2.75,
+      "step": 4067
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.888433393837143e-05,
+      "loss": 4.5469,
+      "step": 4068
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8883633022428326e-05,
+      "loss": 4.6719,
+      "step": 4069
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.88829318927008e-05,
+      "loss": 4.375,
+      "step": 4070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.888223054919679e-05,
+      "loss": 4.4844,
+      "step": 4071
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.888152899192422e-05,
+      "loss": 2.9062,
+      "step": 4072
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.888082722089106e-05,
+      "loss": 3.6875,
+      "step": 4073
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.888012523610522e-05,
+      "loss": 3.8906,
+      "step": 4074
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8879423037574675e-05,
+      "loss": 6.0625,
+      "step": 4075
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.887872062530736e-05,
+      "loss": 3.2656,
+      "step": 4076
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.887801799931124e-05,
+      "loss": 3.9688,
+      "step": 4077
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.887731515959425e-05,
+      "loss": 7.0938,
+      "step": 4078
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8876612106164364e-05,
+      "loss": 3.2969,
+      "step": 4079
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.887590883902954e-05,
+      "loss": 3.75,
+      "step": 4080
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8875205358197735e-05,
+      "loss": 4.625,
+      "step": 4081
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.887450166367691e-05,
+      "loss": 4.0938,
+      "step": 4082
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8873797755475034e-05,
+      "loss": 5.5312,
+      "step": 4083
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8873093633600085e-05,
+      "loss": 3.1719,
+      "step": 4084
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.887238929806001e-05,
+      "loss": 2.4531,
+      "step": 4085
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.887168474886282e-05,
+      "loss": 6.0938,
+      "step": 4086
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.887097998601646e-05,
+      "loss": 3.4531,
+      "step": 4087
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.887027500952893e-05,
+      "loss": 5.0625,
+      "step": 4088
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.886956981940819e-05,
+      "loss": 4.1562,
+      "step": 4089
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.886886441566223e-05,
+      "loss": 5.4688,
+      "step": 4090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8868158798299056e-05,
+      "loss": 3.7344,
+      "step": 4091
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.886745296732664e-05,
+      "loss": 3.6875,
+      "step": 4092
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.886674692275296e-05,
+      "loss": 5.4688,
+      "step": 4093
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.8866040664586035e-05,
+      "loss": 2.75,
+      "step": 4094
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.886533419283384e-05,
+      "loss": 5.7188,
+      "step": 4095
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8864627507504386e-05,
+      "loss": 5.9062,
+      "step": 4096
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8863920608605675e-05,
+      "loss": 2.4219,
+      "step": 4097
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.88632134961457e-05,
+      "loss": 3.3438,
+      "step": 4098
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.886250617013247e-05,
+      "loss": 7.7969,
+      "step": 4099
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.886179863057399e-05,
+      "loss": 6.375,
+      "step": 4100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.886109087747828e-05,
+      "loss": 5.0312,
+      "step": 4101
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.886038291085334e-05,
+      "loss": 3.2656,
+      "step": 4102
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8859674730707194e-05,
+      "loss": 3.5938,
+      "step": 4103
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8858966337047856e-05,
+      "loss": 4.6562,
+      "step": 4104
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.885825772988335e-05,
+      "loss": 3.1875,
+      "step": 4105
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.885754890922169e-05,
+      "loss": 3.5,
+      "step": 4106
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.88568398750709e-05,
+      "loss": 3.0156,
+      "step": 4107
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.885613062743902e-05,
+      "loss": 5.2812,
+      "step": 4108
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8855421166334074e-05,
+      "loss": 6.6094,
+      "step": 4109
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.885471149176408e-05,
+      "loss": 4.8438,
+      "step": 4110
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.885400160373709e-05,
+      "loss": 5.7812,
+      "step": 4111
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.885329150226114e-05,
+      "loss": 8.0312,
+      "step": 4112
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.885258118734426e-05,
+      "loss": 6.75,
+      "step": 4113
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.88518706589945e-05,
+      "loss": 6.375,
+      "step": 4114
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.885115991721989e-05,
+      "loss": 6.5938,
+      "step": 4115
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.885044896202849e-05,
+      "loss": 2.5625,
+      "step": 4116
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.884973779342835e-05,
+      "loss": 5.2188,
+      "step": 4117
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.884902641142751e-05,
+      "loss": 4.0,
+      "step": 4118
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.884831481603403e-05,
+      "loss": 4.625,
+      "step": 4119
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.884760300725597e-05,
+      "loss": 5.0312,
+      "step": 4120
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.884689098510138e-05,
+      "loss": 4.7656,
+      "step": 4121
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8846178749578324e-05,
+      "loss": 5.1406,
+      "step": 4122
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8845466300694866e-05,
+      "loss": 4.9375,
+      "step": 4123
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.8844753638459076e-05,
+      "loss": 6.9062,
+      "step": 4124
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.884404076287902e-05,
+      "loss": 5.3125,
+      "step": 4125
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8843327673962766e-05,
+      "loss": 4.1562,
+      "step": 4126
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.884261437171838e-05,
+      "loss": 5.5,
+      "step": 4127
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.884190085615395e-05,
+      "loss": 7.7656,
+      "step": 4128
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.884118712727755e-05,
+      "loss": 2.4062,
+      "step": 4129
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.884047318509726e-05,
+      "loss": 6.375,
+      "step": 4130
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8839759029621164e-05,
+      "loss": 3.8438,
+      "step": 4131
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.883904466085734e-05,
+      "loss": 5.0781,
+      "step": 4132
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.883833007881388e-05,
+      "loss": 4.4219,
+      "step": 4133
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.883761528349888e-05,
+      "loss": 5.3125,
+      "step": 4134
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.883690027492043e-05,
+      "loss": 4.25,
+      "step": 4135
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.883618505308661e-05,
+      "loss": 6.8906,
+      "step": 4136
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.883546961800554e-05,
+      "loss": 3.4062,
+      "step": 4137
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8834753969685305e-05,
+      "loss": 4.6562,
+      "step": 4138
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8834038108134e-05,
+      "loss": 3.6719,
+      "step": 4139
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.883332203335975e-05,
+      "loss": 3.4688,
+      "step": 4140
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.883260574537064e-05,
+      "loss": 2.9688,
+      "step": 4141
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8831889244174804e-05,
+      "loss": 4.5625,
+      "step": 4142
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.883117252978033e-05,
+      "loss": 3.2656,
+      "step": 4143
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8830455602195354e-05,
+      "loss": 4.9062,
+      "step": 4144
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.882973846142797e-05,
+      "loss": 3.5625,
+      "step": 4145
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8829021107486314e-05,
+      "loss": 3.8438,
+      "step": 4146
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8828303540378496e-05,
+      "loss": 5.2891,
+      "step": 4147
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.882758576011264e-05,
+      "loss": 4.7188,
+      "step": 4148
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.882686776669688e-05,
+      "loss": 3.7812,
+      "step": 4149
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.882614956013935e-05,
+      "loss": 2.5938,
+      "step": 4150
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.882543114044816e-05,
+      "loss": 3.5938,
+      "step": 4151
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.882471250763146e-05,
+      "loss": 5.7344,
+      "step": 4152
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.882399366169738e-05,
+      "loss": 2.9688,
+      "step": 4153
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.882327460265407e-05,
+      "loss": 6.5156,
+      "step": 4154
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.882255533050964e-05,
+      "loss": 3.7188,
+      "step": 4155
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.882183584527226e-05,
+      "loss": 5.2188,
+      "step": 4156
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.8821116146950076e-05,
+      "loss": 5.4375,
+      "step": 4157
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.8820396235551224e-05,
+      "loss": 6.9531,
+      "step": 4158
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.8819676111083856e-05,
+      "loss": 3.8906,
+      "step": 4159
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.8818955773556125e-05,
+      "loss": 4.0781,
+      "step": 4160
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.881823522297619e-05,
+      "loss": 4.25,
+      "step": 4161
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.881751445935221e-05,
+      "loss": 2.5312,
+      "step": 4162
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.8816793482692335e-05,
+      "loss": 2.4375,
+      "step": 4163
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.8816072293004735e-05,
+      "loss": 2.6406,
+      "step": 4164
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.881535089029758e-05,
+      "loss": 3.2188,
+      "step": 4165
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.881462927457902e-05,
+      "loss": 5.5156,
+      "step": 4166
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.8813907445857236e-05,
+      "loss": 2.6875,
+      "step": 4167
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.88131854041404e-05,
+      "loss": 5.5781,
+      "step": 4168
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.881246314943669e-05,
+      "loss": 3.6875,
+      "step": 4169
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.881174068175428e-05,
+      "loss": 2.9844,
+      "step": 4170
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.8811018001101333e-05,
+      "loss": 5.2344,
+      "step": 4171
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.881029510748605e-05,
+      "loss": 3.125,
+      "step": 4172
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.880957200091662e-05,
+      "loss": 6.5,
+      "step": 4173
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.880884868140121e-05,
+      "loss": 4.4688,
+      "step": 4174
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.880812514894801e-05,
+      "loss": 6.5625,
+      "step": 4175
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.8807401403565236e-05,
+      "loss": 6.6562,
+      "step": 4176
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.8806677445261056e-05,
+      "loss": 3.2812,
+      "step": 4177
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.880595327404367e-05,
+      "loss": 2.5625,
+      "step": 4178
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.880522888992129e-05,
+      "loss": 5.8125,
+      "step": 4179
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.8804504292902104e-05,
+      "loss": 3.6719,
+      "step": 4180
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.880377948299432e-05,
+      "loss": 2.7969,
+      "step": 4181
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.880305446020614e-05,
+      "loss": 5.5781,
+      "step": 4182
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.880232922454578e-05,
+      "loss": 4.0938,
+      "step": 4183
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.880160377602144e-05,
+      "loss": 5.0938,
+      "step": 4184
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.880087811464134e-05,
+      "loss": 5.7969,
+      "step": 4185
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.88001522404137e-05,
+      "loss": 2.5781,
+      "step": 4186
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.879942615334672e-05,
+      "loss": 2.6875,
+      "step": 4187
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.879869985344864e-05,
+      "loss": 4.6562,
+      "step": 4188
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.879797334072767e-05,
+      "loss": 3.8438,
+      "step": 4189
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.879724661519204e-05,
+      "loss": 5.125,
+      "step": 4190
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.879651967684998e-05,
+      "loss": 3.5312,
+      "step": 4191
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.8795792525709715e-05,
+      "loss": 5.0781,
+      "step": 4192
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.879506516177947e-05,
+      "loss": 2.625,
+      "step": 4193
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.87943375850675e-05,
+      "loss": 5.8438,
+      "step": 4194
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.879360979558202e-05,
+      "loss": 4.6562,
+      "step": 4195
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.879288179333128e-05,
+      "loss": 5.9375,
+      "step": 4196
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.879215357832353e-05,
+      "loss": 7.2812,
+      "step": 4197
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.8791425150567e-05,
+      "loss": 6.25,
+      "step": 4198
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.879069651006994e-05,
+      "loss": 4.125,
+      "step": 4199
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.8789967656840606e-05,
+      "loss": 2.6875,
+      "step": 4200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.8789238590887245e-05,
+      "loss": 8.375,
+      "step": 4201
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.878850931221811e-05,
+      "loss": 4.75,
+      "step": 4202
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.878777982084145e-05,
+      "loss": 3.1875,
+      "step": 4203
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.878705011676554e-05,
+      "loss": 5.7188,
+      "step": 4204
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.878632019999863e-05,
+      "loss": 5.125,
+      "step": 4205
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.8785590070548984e-05,
+      "loss": 5.0625,
+      "step": 4206
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.878485972842487e-05,
+      "loss": 3.4062,
+      "step": 4207
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.878412917363456e-05,
+      "loss": 4.0,
+      "step": 4208
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.8783398406186315e-05,
+      "loss": 5.4531,
+      "step": 4209
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.878266742608842e-05,
+      "loss": 4.0938,
+      "step": 4210
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.878193623334914e-05,
+      "loss": 2.9688,
+      "step": 4211
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.878120482797676e-05,
+      "loss": 4.2656,
+      "step": 4212
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.878047320997956e-05,
+      "loss": 6.4141,
+      "step": 4213
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.877974137936582e-05,
+      "loss": 4.625,
+      "step": 4214
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.877900933614382e-05,
+      "loss": 4.7188,
+      "step": 4215
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.877827708032186e-05,
+      "loss": 8.9688,
+      "step": 4216
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.8777544611908215e-05,
+      "loss": 4.0938,
+      "step": 4217
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.877681193091119e-05,
+      "loss": 5.9375,
+      "step": 4218
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.877607903733908e-05,
+      "loss": 4.7031,
+      "step": 4219
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.8775345931200176e-05,
+      "loss": 2.4375,
+      "step": 4220
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.877461261250278e-05,
+      "loss": 4.375,
+      "step": 4221
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.877387908125519e-05,
+      "loss": 6.5469,
+      "step": 4222
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.877314533746572e-05,
+      "loss": 5.75,
+      "step": 4223
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.877241138114266e-05,
+      "loss": 3.3438,
+      "step": 4224
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.877167721229434e-05,
+      "loss": 4.9688,
+      "step": 4225
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.8770942830929055e-05,
+      "loss": 4.5,
+      "step": 4226
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.877020823705513e-05,
+      "loss": 3.3438,
+      "step": 4227
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.876947343068087e-05,
+      "loss": 3.25,
+      "step": 4228
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.87687384118146e-05,
+      "loss": 1.8906,
+      "step": 4229
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.876800318046465e-05,
+      "loss": 3.2812,
+      "step": 4230
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.876726773663934e-05,
+      "loss": 4.3125,
+      "step": 4231
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.876653208034698e-05,
+      "loss": 3.9375,
+      "step": 4232
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.8765796211595915e-05,
+      "loss": 5.1719,
+      "step": 4233
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.876506013039447e-05,
+      "loss": 4.1406,
+      "step": 4234
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.8764323836750984e-05,
+      "loss": 3.9375,
+      "step": 4235
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.876358733067378e-05,
+      "loss": 3.2344,
+      "step": 4236
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.8762850612171214e-05,
+      "loss": 4.5469,
+      "step": 4237
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.876211368125161e-05,
+      "loss": 3.0,
+      "step": 4238
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.876137653792332e-05,
+      "loss": 3.8281,
+      "step": 4239
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.876063918219469e-05,
+      "loss": 4.6406,
+      "step": 4240
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.8759901614074056e-05,
+      "loss": 2.9531,
+      "step": 4241
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.8759163833569784e-05,
+      "loss": 2.9375,
+      "step": 4242
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.8758425840690215e-05,
+      "loss": 4.375,
+      "step": 4243
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.875768763544371e-05,
+      "loss": 4.5,
+      "step": 4244
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.875694921783862e-05,
+      "loss": 4.375,
+      "step": 4245
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.875621058788332e-05,
+      "loss": 4.0312,
+      "step": 4246
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8755471745586155e-05,
+      "loss": 4.1719,
+      "step": 4247
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.875473269095549e-05,
+      "loss": 4.1875,
+      "step": 4248
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8753993423999705e-05,
+      "loss": 3.375,
+      "step": 4249
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8753253944727164e-05,
+      "loss": 4.3594,
+      "step": 4250
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8752514253146235e-05,
+      "loss": 4.875,
+      "step": 4251
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8751774349265294e-05,
+      "loss": 3.5469,
+      "step": 4252
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8751034233092715e-05,
+      "loss": 5.75,
+      "step": 4253
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.875029390463688e-05,
+      "loss": 5.2812,
+      "step": 4254
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.874955336390618e-05,
+      "loss": 5.25,
+      "step": 4255
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.874881261090898e-05,
+      "loss": 4.8594,
+      "step": 4256
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.874807164565367e-05,
+      "loss": 2.2031,
+      "step": 4257
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8747330468148655e-05,
+      "loss": 5.375,
+      "step": 4258
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.874658907840231e-05,
+      "loss": 6.7344,
+      "step": 4259
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8745847476423025e-05,
+      "loss": 2.375,
+      "step": 4260
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.874510566221922e-05,
+      "loss": 5.125,
+      "step": 4261
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.874436363579926e-05,
+      "loss": 5.5625,
+      "step": 4262
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.874362139717157e-05,
+      "loss": 5.8125,
+      "step": 4263
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.874287894634455e-05,
+      "loss": 4.0781,
+      "step": 4264
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.874213628332659e-05,
+      "loss": 5.2812,
+      "step": 4265
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.874139340812612e-05,
+      "loss": 5.5,
+      "step": 4266
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.874065032075153e-05,
+      "loss": 4.7031,
+      "step": 4267
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8739907021211254e-05,
+      "loss": 4.5156,
+      "step": 4268
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8739163509513685e-05,
+      "loss": 5.2031,
+      "step": 4269
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.873841978566725e-05,
+      "loss": 2.4688,
+      "step": 4270
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.873767584968037e-05,
+      "loss": 3.9062,
+      "step": 4271
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8736931701561466e-05,
+      "loss": 5.9062,
+      "step": 4272
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.873618734131896e-05,
+      "loss": 4.8438,
+      "step": 4273
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.873544276896129e-05,
+      "loss": 5.5625,
+      "step": 4274
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.8734697984496866e-05,
+      "loss": 6.8438,
+      "step": 4275
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.873395298793414e-05,
+      "loss": 5.5781,
+      "step": 4276
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.873320777928153e-05,
+      "loss": 2.4688,
+      "step": 4277
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8732462358547486e-05,
+      "loss": 4.5938,
+      "step": 4278
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.873171672574043e-05,
+      "loss": 4.4688,
+      "step": 4279
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.873097088086883e-05,
+      "loss": 2.2812,
+      "step": 4280
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.87302248239411e-05,
+      "loss": 3.625,
+      "step": 4281
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.872947855496571e-05,
+      "loss": 4.0938,
+      "step": 4282
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.872873207395109e-05,
+      "loss": 3.7188,
+      "step": 4283
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.87279853809057e-05,
+      "loss": 3.0312,
+      "step": 4284
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8727238475838e-05,
+      "loss": 6.0312,
+      "step": 4285
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.872649135875643e-05,
+      "loss": 2.9375,
+      "step": 4286
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8725744029669465e-05,
+      "loss": 3.4844,
+      "step": 4287
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.872499648858556e-05,
+      "loss": 2.9062,
+      "step": 4288
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.872424873551317e-05,
+      "loss": 3.9219,
+      "step": 4289
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.872350077046077e-05,
+      "loss": 3.0625,
+      "step": 4290
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.872275259343682e-05,
+      "loss": 4.1406,
+      "step": 4291
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8722004204449797e-05,
+      "loss": 2.7344,
+      "step": 4292
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.872125560350817e-05,
+      "loss": 2.75,
+      "step": 4293
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.872050679062042e-05,
+      "loss": 3.5625,
+      "step": 4294
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.871975776579502e-05,
+      "loss": 2.5781,
+      "step": 4295
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8719008529040436e-05,
+      "loss": 5.4375,
+      "step": 4296
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.871825908036517e-05,
+      "loss": 4.7812,
+      "step": 4297
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8717509419777705e-05,
+      "loss": 8.7656,
+      "step": 4298
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8716759547286516e-05,
+      "loss": 4.375,
+      "step": 4299
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.871600946290011e-05,
+      "loss": 5.5312,
+      "step": 4300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.871525916662696e-05,
+      "loss": 3.4219,
+      "step": 4301
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.871450865847557e-05,
+      "loss": 4.5469,
+      "step": 4302
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.871375793845444e-05,
+      "loss": 4.5312,
+      "step": 4303
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.8713007006572053e-05,
+      "loss": 3.1875,
+      "step": 4304
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.871225586283693e-05,
+      "loss": 3.2031,
+      "step": 4305
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.871150450725757e-05,
+      "loss": 2.7188,
+      "step": 4306
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.871075293984247e-05,
+      "loss": 3.2188,
+      "step": 4307
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.871000116060014e-05,
+      "loss": 4.7188,
+      "step": 4308
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.87092491695391e-05,
+      "loss": 3.625,
+      "step": 4309
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.870849696666786e-05,
+      "loss": 4.9062,
+      "step": 4310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.8707744551994935e-05,
+      "loss": 4.8125,
+      "step": 4311
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.870699192552884e-05,
+      "loss": 4.375,
+      "step": 4312
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.87062390872781e-05,
+      "loss": 3.75,
+      "step": 4313
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.8705486037251234e-05,
+      "loss": 4.5312,
+      "step": 4314
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.870473277545677e-05,
+      "loss": 4.4844,
+      "step": 4315
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.870397930190324e-05,
+      "loss": 4.5,
+      "step": 4316
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.870322561659916e-05,
+      "loss": 4.8125,
+      "step": 4317
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.8702471719553084e-05,
+      "loss": 4.7969,
+      "step": 4318
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.8701717610773526e-05,
+      "loss": 7.0,
+      "step": 4319
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.8700963290269036e-05,
+      "loss": 3.7656,
+      "step": 4320
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.870020875804815e-05,
+      "loss": 3.0938,
+      "step": 4321
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.869945401411941e-05,
+      "loss": 5.5781,
+      "step": 4322
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.869869905849136e-05,
+      "loss": 4.7344,
+      "step": 4323
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.869794389117255e-05,
+      "loss": 6.2188,
+      "step": 4324
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.869718851217153e-05,
+      "loss": 3.4688,
+      "step": 4325
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.8696432921496845e-05,
+      "loss": 4.0,
+      "step": 4326
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.869567711915705e-05,
+      "loss": 6.4375,
+      "step": 4327
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.869492110516071e-05,
+      "loss": 5.6562,
+      "step": 4328
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.869416487951638e-05,
+      "loss": 4.4531,
+      "step": 4329
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.8693408442232615e-05,
+      "loss": 6.8125,
+      "step": 4330
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.869265179331799e-05,
+      "loss": 4.0938,
+      "step": 4331
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.8691894932781066e-05,
+      "loss": 5.75,
+      "step": 4332
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.86911378606304e-05,
+      "loss": 5.3438,
+      "step": 4333
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.8690380576874584e-05,
+      "loss": 4.6875,
+      "step": 4334
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.868962308152218e-05,
+      "loss": 1.1719,
+      "step": 4335
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.868886537458175e-05,
+      "loss": 5.9375,
+      "step": 4336
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.8688107456061904e-05,
+      "loss": 3.625,
+      "step": 4337
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8687349325971196e-05,
+      "loss": 3.9688,
+      "step": 4338
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.868659098431822e-05,
+      "loss": 3.4688,
+      "step": 4339
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.868583243111156e-05,
+      "loss": 4.1875,
+      "step": 4340
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.868507366635981e-05,
+      "loss": 4.5,
+      "step": 4341
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8684314690071545e-05,
+      "loss": 7.1719,
+      "step": 4342
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.868355550225537e-05,
+      "loss": 3.3125,
+      "step": 4343
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.868279610291987e-05,
+      "loss": 5.8281,
+      "step": 4344
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.868203649207365e-05,
+      "loss": 4.5625,
+      "step": 4345
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.86812766697253e-05,
+      "loss": 5.9688,
+      "step": 4346
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.868051663588344e-05,
+      "loss": 5.75,
+      "step": 4347
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.867975639055666e-05,
+      "loss": 1.8125,
+      "step": 4348
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8678995933753576e-05,
+      "loss": 3.5781,
+      "step": 4349
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8678235265482786e-05,
+      "loss": 5.5625,
+      "step": 4350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.867747438575291e-05,
+      "loss": 5.0938,
+      "step": 4351
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8676713294572564e-05,
+      "loss": 3.8438,
+      "step": 4352
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.867595199195036e-05,
+      "loss": 5.3906,
+      "step": 4353
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.867519047789491e-05,
+      "loss": 4.7344,
+      "step": 4354
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8674428752414834e-05,
+      "loss": 5.6875,
+      "step": 4355
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.867366681551878e-05,
+      "loss": 5.1562,
+      "step": 4356
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.867290466721535e-05,
+      "loss": 3.5625,
+      "step": 4357
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8672142307513184e-05,
+      "loss": 3.1406,
+      "step": 4358
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8671379736420907e-05,
+      "loss": 2.6562,
+      "step": 4359
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.867061695394715e-05,
+      "loss": 3.7812,
+      "step": 4360
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.866985396010056e-05,
+      "loss": 7.3125,
+      "step": 4361
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.866909075488977e-05,
+      "loss": 2.9531,
+      "step": 4362
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.866832733832341e-05,
+      "loss": 2.7812,
+      "step": 4363
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.866756371041013e-05,
+      "loss": 2.375,
+      "step": 4364
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8666799871158585e-05,
+      "loss": 3.5,
+      "step": 4365
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8666035820577406e-05,
+      "loss": 4.9688,
+      "step": 4366
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.866527155867525e-05,
+      "loss": 2.9688,
+      "step": 4367
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.8664507085460774e-05,
+      "loss": 5.125,
+      "step": 4368
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.866374240094263e-05,
+      "loss": 6.1094,
+      "step": 4369
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.866297750512947e-05,
+      "loss": 4.3125,
+      "step": 4370
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8662212398029954e-05,
+      "loss": 3.7656,
+      "step": 4371
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8661447079652746e-05,
+      "loss": 6.4375,
+      "step": 4372
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.866068155000652e-05,
+      "loss": 5.2031,
+      "step": 4373
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8659915809099925e-05,
+      "loss": 9.9062,
+      "step": 4374
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.865914985694165e-05,
+      "loss": 4.7812,
+      "step": 4375
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.865838369354035e-05,
+      "loss": 5.6094,
+      "step": 4376
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.86576173189047e-05,
+      "loss": 3.2188,
+      "step": 4377
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.865685073304338e-05,
+      "loss": 3.75,
+      "step": 4378
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8656083935965075e-05,
+      "loss": 5.4375,
+      "step": 4379
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.865531692767846e-05,
+      "loss": 4.75,
+      "step": 4380
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8654549708192214e-05,
+      "loss": 5.7188,
+      "step": 4381
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.865378227751503e-05,
+      "loss": 4.6875,
+      "step": 4382
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.865301463565559e-05,
+      "loss": 6.0156,
+      "step": 4383
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.86522467826226e-05,
+      "loss": 2.2656,
+      "step": 4384
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8651478718424724e-05,
+      "loss": 5.5,
+      "step": 4385
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.865071044307069e-05,
+      "loss": 4.7812,
+      "step": 4386
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.864994195656917e-05,
+      "loss": 2.5,
+      "step": 4387
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.864917325892888e-05,
+      "loss": 5.3438,
+      "step": 4388
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.864840435015851e-05,
+      "loss": 4.9062,
+      "step": 4389
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.864763523026678e-05,
+      "loss": 4.7188,
+      "step": 4390
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.864686589926239e-05,
+      "loss": 4.2812,
+      "step": 4391
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.864609635715404e-05,
+      "loss": 2.375,
+      "step": 4392
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8645326603950455e-05,
+      "loss": 2.5625,
+      "step": 4393
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.864455663966035e-05,
+      "loss": 1.6875,
+      "step": 4394
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.864378646429243e-05,
+      "loss": 4.4844,
+      "step": 4395
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8643016077855426e-05,
+      "loss": 5.1875,
+      "step": 4396
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.864224548035805e-05,
+      "loss": 6.4688,
+      "step": 4397
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8641474671809045e-05,
+      "loss": 4.5,
+      "step": 4398
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.864070365221712e-05,
+      "loss": 5.2344,
+      "step": 4399
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.8639932421591e-05,
+      "loss": 5.0469,
+      "step": 4400
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.863916097993943e-05,
+      "loss": 3.7188,
+      "step": 4401
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.863838932727113e-05,
+      "loss": 3.125,
+      "step": 4402
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.863761746359485e-05,
+      "loss": 3.4688,
+      "step": 4403
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.863684538891932e-05,
+      "loss": 4.0781,
+      "step": 4404
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.8636073103253276e-05,
+      "loss": 2.0938,
+      "step": 4405
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.863530060660548e-05,
+      "loss": 5.6406,
+      "step": 4406
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.8634527898984654e-05,
+      "loss": 6.7812,
+      "step": 4407
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.863375498039957e-05,
+      "loss": 3.9844,
+      "step": 4408
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.863298185085896e-05,
+      "loss": 4.3906,
+      "step": 4409
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.863220851037158e-05,
+      "loss": 4.25,
+      "step": 4410
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.863143495894618e-05,
+      "loss": 2.9688,
+      "step": 4411
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.863066119659153e-05,
+      "loss": 3.4219,
+      "step": 4412
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.862988722331639e-05,
+      "loss": 6.375,
+      "step": 4413
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.8629113039129515e-05,
+      "loss": 4.8438,
+      "step": 4414
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.862833864403967e-05,
+      "loss": 3.2031,
+      "step": 4415
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.862756403805563e-05,
+      "loss": 3.0938,
+      "step": 4416
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.862678922118615e-05,
+      "loss": 3.0469,
+      "step": 4417
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.8626014193440015e-05,
+      "loss": 3.0,
+      "step": 4418
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.8625238954825996e-05,
+      "loss": 4.3125,
+      "step": 4419
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.862446350535287e-05,
+      "loss": 5.7344,
+      "step": 4420
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.86236878450294e-05,
+      "loss": 4.5312,
+      "step": 4421
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.862291197386439e-05,
+      "loss": 3.4219,
+      "step": 4422
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.862213589186662e-05,
+      "loss": 2.3438,
+      "step": 4423
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.8621359599044865e-05,
+      "loss": 4.4375,
+      "step": 4424
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.8620583095407915e-05,
+      "loss": 2.125,
+      "step": 4425
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.861980638096457e-05,
+      "loss": 5.3438,
+      "step": 4426
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.8619029455723624e-05,
+      "loss": 2.3438,
+      "step": 4427
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.861825231969386e-05,
+      "loss": 2.4531,
+      "step": 4428
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.861747497288409e-05,
+      "loss": 2.0625,
+      "step": 4429
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.8616697415303104e-05,
+      "loss": 5.0938,
+      "step": 4430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.861591964695971e-05,
+      "loss": 2.3594,
+      "step": 4431
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.861514166786271e-05,
+      "loss": 2.4688,
+      "step": 4432
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.8614363478020916e-05,
+      "loss": 2.7812,
+      "step": 4433
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.8613585077443136e-05,
+      "loss": 4.1406,
+      "step": 4434
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.8612806466138176e-05,
+      "loss": 4.0,
+      "step": 4435
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.8612027644114864e-05,
+      "loss": 4.1406,
+      "step": 4436
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.861124861138201e-05,
+      "loss": 3.9688,
+      "step": 4437
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.861046936794843e-05,
+      "loss": 3.0156,
+      "step": 4438
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.860968991382295e-05,
+      "loss": 3.4688,
+      "step": 4439
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.8608910249014394e-05,
+      "loss": 3.1406,
+      "step": 4440
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.860813037353159e-05,
+      "loss": 4.6875,
+      "step": 4441
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.860735028738337e-05,
+      "loss": 3.8125,
+      "step": 4442
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.860656999057855e-05,
+      "loss": 3.875,
+      "step": 4443
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.860578948312598e-05,
+      "loss": 4.6016,
+      "step": 4444
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.8605008765034495e-05,
+      "loss": 3.2969,
+      "step": 4445
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.8604227836312925e-05,
+      "loss": 3.7188,
+      "step": 4446
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.860344669697012e-05,
+      "loss": 3.2344,
+      "step": 4447
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.860266534701492e-05,
+      "loss": 5.125,
+      "step": 4448
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.860188378645617e-05,
+      "loss": 2.8906,
+      "step": 4449
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.8601102015302705e-05,
+      "loss": 3.5781,
+      "step": 4450
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.86003200335634e-05,
+      "loss": 3.1875,
+      "step": 4451
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.859953784124709e-05,
+      "loss": 4.4375,
+      "step": 4452
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.859875543836265e-05,
+      "loss": 4.1562,
+      "step": 4453
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.859797282491892e-05,
+      "loss": 5.4531,
+      "step": 4454
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.859719000092476e-05,
+      "loss": 4.9062,
+      "step": 4455
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.859640696638904e-05,
+      "loss": 2.7031,
+      "step": 4456
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.8595623721320617e-05,
+      "loss": 1.75,
+      "step": 4457
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.8594840265728365e-05,
+      "loss": 5.25,
+      "step": 4458
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.859405659962116e-05,
+      "loss": 2.6562,
+      "step": 4459
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.8593272723007854e-05,
+      "loss": 4.9844,
+      "step": 4460
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.859248863589734e-05,
+      "loss": 5.6562,
+      "step": 4461
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.859170433829849e-05,
+      "loss": 2.3594,
+      "step": 4462
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.859091983022018e-05,
+      "loss": 5.0781,
+      "step": 4463
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.859013511167129e-05,
+      "loss": 6.375,
+      "step": 4464
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.858935018266071e-05,
+      "loss": 3.6406,
+      "step": 4465
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.858856504319732e-05,
+      "loss": 5.0781,
+      "step": 4466
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.8587779693290016e-05,
+      "loss": 2.7188,
+      "step": 4467
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.8586994132947674e-05,
+      "loss": 6.0312,
+      "step": 4468
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.8586208362179215e-05,
+      "loss": 3.2188,
+      "step": 4469
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.85854223809935e-05,
+      "loss": 3.2656,
+      "step": 4470
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.8584636189399455e-05,
+      "loss": 7.5469,
+      "step": 4471
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.858384978740597e-05,
+      "loss": 5.9375,
+      "step": 4472
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.858306317502195e-05,
+      "loss": 4.8438,
+      "step": 4473
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.8582276352256296e-05,
+      "loss": 5.2031,
+      "step": 4474
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.858148931911792e-05,
+      "loss": 4.7188,
+      "step": 4475
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.858070207561573e-05,
+      "loss": 4.625,
+      "step": 4476
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.8579914621758644e-05,
+      "loss": 4.2969,
+      "step": 4477
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.8579126957555564e-05,
+      "loss": 4.5938,
+      "step": 4478
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.857833908301542e-05,
+      "loss": 5.8125,
+      "step": 4479
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.857755099814713e-05,
+      "loss": 5.2188,
+      "step": 4480
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.857676270295961e-05,
+      "loss": 4.4531,
+      "step": 4481
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.8575974197461785e-05,
+      "loss": 4.3516,
+      "step": 4482
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.857518548166259e-05,
+      "loss": 3.0,
+      "step": 4483
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.857439655557095e-05,
+      "loss": 2.5,
+      "step": 4484
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.857360741919579e-05,
+      "loss": 4.0312,
+      "step": 4485
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.8572818072546055e-05,
+      "loss": 6.0938,
+      "step": 4486
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.857202851563067e-05,
+      "loss": 3.7031,
+      "step": 4487
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.857123874845858e-05,
+      "loss": 4.4375,
+      "step": 4488
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.857044877103873e-05,
+      "loss": 3.7188,
+      "step": 4489
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.8569658583380054e-05,
+      "loss": 7.6875,
+      "step": 4490
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.8568868185491496e-05,
+      "loss": 3.4219,
+      "step": 4491
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.856807757738202e-05,
+      "loss": 5.4062,
+      "step": 4492
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.856728675906057e-05,
+      "loss": 1.4375,
+      "step": 4493
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.856649573053609e-05,
+      "loss": 4.4062,
+      "step": 4494
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.856570449181754e-05,
+      "loss": 5.5,
+      "step": 4495
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.8564913042913884e-05,
+      "loss": 2.0625,
+      "step": 4496
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.856412138383408e-05,
+      "loss": 3.9375,
+      "step": 4497
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.856332951458708e-05,
+      "loss": 2.0312,
+      "step": 4498
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.856253743518186e-05,
+      "loss": 2.3438,
+      "step": 4499
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.856174514562739e-05,
+      "loss": 6.625,
+      "step": 4500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.856095264593262e-05,
+      "loss": 7.5,
+      "step": 4501
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.8560159936106544e-05,
+      "loss": 3.5625,
+      "step": 4502
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.8559367016158135e-05,
+      "loss": 4.5,
+      "step": 4503
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.855857388609635e-05,
+      "loss": 3.8906,
+      "step": 4504
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.8557780545930194e-05,
+      "loss": 4.875,
+      "step": 4505
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.855698699566862e-05,
+      "loss": 6.6875,
+      "step": 4506
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.855619323532064e-05,
+      "loss": 4.3594,
+      "step": 4507
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.855539926489522e-05,
+      "loss": 3.7188,
+      "step": 4508
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.8554605084401364e-05,
+      "loss": 5.0625,
+      "step": 4509
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.855381069384805e-05,
+      "loss": 8.1875,
+      "step": 4510
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.855301609324428e-05,
+      "loss": 3.0312,
+      "step": 4511
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.855222128259905e-05,
+      "loss": 5.4844,
+      "step": 4512
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.855142626192134e-05,
+      "loss": 5.25,
+      "step": 4513
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.855063103122018e-05,
+      "loss": 8.4375,
+      "step": 4514
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.854983559050455e-05,
+      "loss": 5.3906,
+      "step": 4515
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.854903993978347e-05,
+      "loss": 2.7656,
+      "step": 4516
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.8548244079065936e-05,
+      "loss": 3.5938,
+      "step": 4517
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.854744800836097e-05,
+      "loss": 3.3125,
+      "step": 4518
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.854665172767757e-05,
+      "loss": 3.8125,
+      "step": 4519
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.854585523702476e-05,
+      "loss": 4.8281,
+      "step": 4520
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.854505853641156e-05,
+      "loss": 2.625,
+      "step": 4521
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.854426162584698e-05,
+      "loss": 2.8438,
+      "step": 4522
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.854346450534005e-05,
+      "loss": 4.8594,
+      "step": 4523
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.854266717489979e-05,
+      "loss": 5.0625,
+      "step": 4524
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.854186963453523e-05,
+      "loss": 3.875,
+      "step": 4525
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.8541071884255393e-05,
+      "loss": 4.8125,
+      "step": 4526
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.854027392406933e-05,
+      "loss": 2.1094,
+      "step": 4527
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.8539475753986043e-05,
+      "loss": 4.3438,
+      "step": 4528
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.853867737401459e-05,
+      "loss": 5.3906,
+      "step": 4529
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.8537878784164e-05,
+      "loss": 2.6875,
+      "step": 4530
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.853707998444333e-05,
+      "loss": 3.1406,
+      "step": 4531
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.85362809748616e-05,
+      "loss": 3.8281,
+      "step": 4532
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.853548175542787e-05,
+      "loss": 3.4062,
+      "step": 4533
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.853468232615119e-05,
+      "loss": 5.625,
+      "step": 4534
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.85338826870406e-05,
+      "loss": 4.5938,
+      "step": 4535
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.853308283810516e-05,
+      "loss": 6.2344,
+      "step": 4536
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.853228277935393e-05,
+      "loss": 3.9375,
+      "step": 4537
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.8531482510795954e-05,
+      "loss": 3.3281,
+      "step": 4538
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.85306820324403e-05,
+      "loss": 2.9062,
+      "step": 4539
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.8529881344296037e-05,
+      "loss": 2.4844,
+      "step": 4540
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.852908044637222e-05,
+      "loss": 3.4688,
+      "step": 4541
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.8528279338677915e-05,
+      "loss": 4.9688,
+      "step": 4542
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.85274780212222e-05,
+      "loss": 4.3125,
+      "step": 4543
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.852667649401414e-05,
+      "loss": 3.5,
+      "step": 4544
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.852587475706281e-05,
+      "loss": 2.9219,
+      "step": 4545
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.852507281037729e-05,
+      "loss": 5.125,
+      "step": 4546
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.852427065396665e-05,
+      "loss": 6.5156,
+      "step": 4547
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.852346828783998e-05,
+      "loss": 4.7188,
+      "step": 4548
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.852266571200637e-05,
+      "loss": 2.875,
+      "step": 4549
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.852186292647489e-05,
+      "loss": 4.7656,
+      "step": 4550
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.852105993125464e-05,
+      "loss": 4.2969,
+      "step": 4551
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.8520256726354704e-05,
+      "loss": 6.4531,
+      "step": 4552
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.851945331178418e-05,
+      "loss": 6.8594,
+      "step": 4553
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.8518649687552156e-05,
+      "loss": 4.6094,
+      "step": 4554
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.8517845853667745e-05,
+      "loss": 6.9375,
+      "step": 4555
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.851704181014004e-05,
+      "loss": 3.0312,
+      "step": 4556
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.851623755697813e-05,
+      "loss": 3.5312,
+      "step": 4557
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.851543309419114e-05,
+      "loss": 4.4062,
+      "step": 4558
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.851462842178816e-05,
+      "loss": 5.9219,
+      "step": 4559
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.8513823539778325e-05,
+      "loss": 4.8281,
+      "step": 4560
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.851301844817072e-05,
+      "loss": 5.6875,
+      "step": 4561
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.8512213146974476e-05,
+      "loss": 3.125,
+      "step": 4562
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.85114076361987e-05,
+      "loss": 4.0312,
+      "step": 4563
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.8510601915852524e-05,
+      "loss": 5.7969,
+      "step": 4564
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.850979598594506e-05,
+      "loss": 1.8125,
+      "step": 4565
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.850898984648542e-05,
+      "loss": 2.75,
+      "step": 4566
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.850818349748276e-05,
+      "loss": 3.8281,
+      "step": 4567
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.8507376938946184e-05,
+      "loss": 4.7188,
+      "step": 4568
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.850657017088484e-05,
+      "loss": 2.9844,
+      "step": 4569
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.850576319330785e-05,
+      "loss": 3.1719,
+      "step": 4570
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.8504956006224355e-05,
+      "loss": 4.7969,
+      "step": 4571
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.850414860964349e-05,
+      "loss": 2.0156,
+      "step": 4572
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.85033410035744e-05,
+      "loss": 4.8438,
+      "step": 4573
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.850253318802622e-05,
+      "loss": 3.9688,
+      "step": 4574
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.850172516300811e-05,
+      "loss": 3.8906,
+      "step": 4575
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.85009169285292e-05,
+      "loss": 3.8125,
+      "step": 4576
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.850010848459866e-05,
+      "loss": 8.5625,
+      "step": 4577
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.849929983122562e-05,
+      "loss": 3.6406,
+      "step": 4578
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.849849096841925e-05,
+      "loss": 2.875,
+      "step": 4579
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.8497681896188706e-05,
+      "loss": 3.4688,
+      "step": 4580
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.849687261454314e-05,
+      "loss": 3.5312,
+      "step": 4581
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.849606312349173e-05,
+      "loss": 4.0625,
+      "step": 4582
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.849525342304362e-05,
+      "loss": 6.2031,
+      "step": 4583
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.849444351320799e-05,
+      "loss": 3.0625,
+      "step": 4584
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.8493633393994005e-05,
+      "loss": 9.1562,
+      "step": 4585
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.849282306541084e-05,
+      "loss": 3.6094,
+      "step": 4586
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.849201252746766e-05,
+      "loss": 4.8281,
+      "step": 4587
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.8491201780173646e-05,
+      "loss": 4.1562,
+      "step": 4588
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.849039082353799e-05,
+      "loss": 3.8438,
+      "step": 4589
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.848957965756985e-05,
+      "loss": 4.375,
+      "step": 4590
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.848876828227842e-05,
+      "loss": 4.0469,
+      "step": 4591
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.848795669767289e-05,
+      "loss": 2.3125,
+      "step": 4592
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.8487144903762443e-05,
+      "loss": 2.8125,
+      "step": 4593
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.848633290055627e-05,
+      "loss": 5.8438,
+      "step": 4594
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.8485520688063556e-05,
+      "loss": 4.0312,
+      "step": 4595
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.8484708266293514e-05,
+      "loss": 8.5625,
+      "step": 4596
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.848389563525533e-05,
+      "loss": 4.6875,
+      "step": 4597
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.84830827949582e-05,
+      "loss": 3.6719,
+      "step": 4598
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.848226974541134e-05,
+      "loss": 4.2031,
+      "step": 4599
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.848145648662394e-05,
+      "loss": 2.1875,
+      "step": 4600
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.848064301860522e-05,
+      "loss": 7.0781,
+      "step": 4601
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.847982934136438e-05,
+      "loss": 3.8438,
+      "step": 4602
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.847901545491063e-05,
+      "loss": 3.9062,
+      "step": 4603
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.847820135925319e-05,
+      "loss": 2.9375,
+      "step": 4604
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.8477387054401284e-05,
+      "loss": 6.0938,
+      "step": 4605
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.847657254036411e-05,
+      "loss": 8.9844,
+      "step": 4606
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.847575781715091e-05,
+      "loss": 3.4375,
+      "step": 4607
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.8474942884770896e-05,
+      "loss": 5.7656,
+      "step": 4608
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.847412774323331e-05,
+      "loss": 5.8125,
+      "step": 4609
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.847331239254735e-05,
+      "loss": 4.8203,
+      "step": 4610
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.847249683272227e-05,
+      "loss": 7.4375,
+      "step": 4611
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.84716810637673e-05,
+      "loss": 4.9688,
+      "step": 4612
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.847086508569167e-05,
+      "loss": 4.7188,
+      "step": 4613
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.847004889850461e-05,
+      "loss": 4.0625,
+      "step": 4614
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.8469232502215384e-05,
+      "loss": 4.6094,
+      "step": 4615
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.846841589683321e-05,
+      "loss": 5.5469,
+      "step": 4616
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.8467599082367355e-05,
+      "loss": 4.375,
+      "step": 4617
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.8466782058827055e-05,
+      "loss": 5.7188,
+      "step": 4618
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.846596482622156e-05,
+      "loss": 1.8438,
+      "step": 4619
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.8465147384560114e-05,
+      "loss": 4.8438,
+      "step": 4620
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.846432973385199e-05,
+      "loss": 2.6719,
+      "step": 4621
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.846351187410642e-05,
+      "loss": 3.75,
+      "step": 4622
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.846269380533269e-05,
+      "loss": 5.9062,
+      "step": 4623
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.8461875527540044e-05,
+      "loss": 4.5938,
+      "step": 4624
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.846105704073775e-05,
+      "loss": 4.9531,
+      "step": 4625
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.846023834493508e-05,
+      "loss": 5.0938,
+      "step": 4626
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.845941944014129e-05,
+      "loss": 3.4375,
+      "step": 4627
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.8458600326365655e-05,
+      "loss": 5.125,
+      "step": 4628
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.845778100361746e-05,
+      "loss": 2.5625,
+      "step": 4629
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.845696147190597e-05,
+      "loss": 3.5469,
+      "step": 4630
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.8456141731240465e-05,
+      "loss": 2.7188,
+      "step": 4631
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.845532178163023e-05,
+      "loss": 4.8594,
+      "step": 4632
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.8454501623084536e-05,
+      "loss": 3.5781,
+      "step": 4633
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.845368125561268e-05,
+      "loss": 5.2812,
+      "step": 4634
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.8452860679223946e-05,
+      "loss": 2.0469,
+      "step": 4635
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.845203989392762e-05,
+      "loss": 2.7188,
+      "step": 4636
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.8451218899733e-05,
+      "loss": 7.0781,
+      "step": 4637
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.8450397696649375e-05,
+      "loss": 9.6875,
+      "step": 4638
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.8449576284686045e-05,
+      "loss": 3.9375,
+      "step": 4639
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.844875466385231e-05,
+      "loss": 3.2656,
+      "step": 4640
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.844793283415747e-05,
+      "loss": 3.2969,
+      "step": 4641
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.844711079561084e-05,
+      "loss": 4.4219,
+      "step": 4642
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.84462885482217e-05,
+      "loss": 6.375,
+      "step": 4643
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.844546609199938e-05,
+      "loss": 5.3906,
+      "step": 4644
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.844464342695319e-05,
+      "loss": 3.625,
+      "step": 4645
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.8443820553092444e-05,
+      "loss": 2.375,
+      "step": 4646
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.844299747042644e-05,
+      "loss": 5.625,
+      "step": 4647
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.844217417896452e-05,
+      "loss": 3.9062,
+      "step": 4648
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.844135067871599e-05,
+      "loss": 4.0781,
+      "step": 4649
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.8440526969690186e-05,
+      "loss": 3.6719,
+      "step": 4650
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.843970305189641e-05,
+      "loss": 5.6406,
+      "step": 4651
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.843887892534402e-05,
+      "loss": 5.9375,
+      "step": 4652
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.843805459004232e-05,
+      "loss": 3.7031,
+      "step": 4653
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.843723004600066e-05,
+      "loss": 5.9062,
+      "step": 4654
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.843640529322836e-05,
+      "loss": 6.9688,
+      "step": 4655
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.843558033173477e-05,
+      "loss": 4.6875,
+      "step": 4656
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.843475516152922e-05,
+      "loss": 4.3125,
+      "step": 4657
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.843392978262106e-05,
+      "loss": 3.8438,
+      "step": 4658
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.843310419501963e-05,
+      "loss": 10.1562,
+      "step": 4659
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.8432278398734276e-05,
+      "loss": 6.8281,
+      "step": 4660
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.843145239377436e-05,
+      "loss": 3.875,
+      "step": 4661
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.84306261801492e-05,
+      "loss": 8.75,
+      "step": 4662
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.842979975786819e-05,
+      "loss": 6.4062,
+      "step": 4663
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.8428973126940654e-05,
+      "loss": 4.0781,
+      "step": 4664
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.842814628737596e-05,
+      "loss": 4.5625,
+      "step": 4665
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.8427319239183486e-05,
+      "loss": 2.75,
+      "step": 4666
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.842649198237258e-05,
+      "loss": 3.6719,
+      "step": 4667
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.84256645169526e-05,
+      "loss": 4.9375,
+      "step": 4668
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.842483684293293e-05,
+      "loss": 4.9375,
+      "step": 4669
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.8424008960322925e-05,
+      "loss": 4.4375,
+      "step": 4670
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.8423180869131976e-05,
+      "loss": 3.7188,
+      "step": 4671
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.8422352569369444e-05,
+      "loss": 4.6562,
+      "step": 4672
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.8421524061044706e-05,
+      "loss": 6.5781,
+      "step": 4673
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.842069534416715e-05,
+      "loss": 3.3125,
+      "step": 4674
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.841986641874615e-05,
+      "loss": 2.875,
+      "step": 4675
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.84190372847911e-05,
+      "loss": 4.0938,
+      "step": 4676
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.841820794231138e-05,
+      "loss": 5.6406,
+      "step": 4677
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.841737839131637e-05,
+      "loss": 3.3125,
+      "step": 4678
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.8416548631815474e-05,
+      "loss": 3.3438,
+      "step": 4679
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.841571866381808e-05,
+      "loss": 7.2656,
+      "step": 4680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.84148884873336e-05,
+      "loss": 3.1562,
+      "step": 4681
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.841405810237141e-05,
+      "loss": 2.4375,
+      "step": 4682
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.8413227508940926e-05,
+      "loss": 3.0938,
+      "step": 4683
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.8412396707051545e-05,
+      "loss": 5.625,
+      "step": 4684
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.8411565696712674e-05,
+      "loss": 5.4375,
+      "step": 4685
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.841073447793372e-05,
+      "loss": 5.6875,
+      "step": 4686
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.840990305072409e-05,
+      "loss": 4.2188,
+      "step": 4687
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.8409071415093204e-05,
+      "loss": 2.9375,
+      "step": 4688
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.840823957105048e-05,
+      "loss": 3.7188,
+      "step": 4689
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.840740751860532e-05,
+      "loss": 2.1562,
+      "step": 4690
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.840657525776716e-05,
+      "loss": 6.4531,
+      "step": 4691
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.840574278854541e-05,
+      "loss": 3.8438,
+      "step": 4692
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.84049101109495e-05,
+      "loss": 4.5469,
+      "step": 4693
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.8404077224988855e-05,
+      "loss": 11.75,
+      "step": 4694
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.840324413067291e-05,
+      "loss": 5.5938,
+      "step": 4695
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.840241082801109e-05,
+      "loss": 4.75,
+      "step": 4696
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.840157731701283e-05,
+      "loss": 4.375,
+      "step": 4697
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.840074359768757e-05,
+      "loss": 3.3281,
+      "step": 4698
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.839990967004475e-05,
+      "loss": 3.875,
+      "step": 4699
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.83990755340938e-05,
+      "loss": 2.9375,
+      "step": 4700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.839824118984417e-05,
+      "loss": 6.8125,
+      "step": 4701
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.8397406637305315e-05,
+      "loss": 5.75,
+      "step": 4702
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.839657187648667e-05,
+      "loss": 4.4062,
+      "step": 4703
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8395736907397686e-05,
+      "loss": 3.25,
+      "step": 4704
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.839490173004783e-05,
+      "loss": 5.9062,
+      "step": 4705
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.839406634444654e-05,
+      "loss": 5.0312,
+      "step": 4706
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.839323075060328e-05,
+      "loss": 5.6562,
+      "step": 4707
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.839239494852751e-05,
+      "loss": 2.9219,
+      "step": 4708
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8391558938228694e-05,
+      "loss": 3.3438,
+      "step": 4709
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.839072271971629e-05,
+      "loss": 4.0781,
+      "step": 4710
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8389886292999774e-05,
+      "loss": 3.2188,
+      "step": 4711
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8389049658088616e-05,
+      "loss": 1.9062,
+      "step": 4712
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.838821281499228e-05,
+      "loss": 2.4688,
+      "step": 4713
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.838737576372024e-05,
+      "loss": 3.4375,
+      "step": 4714
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.838653850428198e-05,
+      "loss": 4.4375,
+      "step": 4715
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8385701036686966e-05,
+      "loss": 3.375,
+      "step": 4716
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.83848633609447e-05,
+      "loss": 3.8281,
+      "step": 4717
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.838402547706464e-05,
+      "loss": 4.9062,
+      "step": 4718
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.838318738505629e-05,
+      "loss": 4.6562,
+      "step": 4719
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.838234908492914e-05,
+      "loss": 7.0938,
+      "step": 4720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8381510576692664e-05,
+      "loss": 2.2031,
+      "step": 4721
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8380671860356366e-05,
+      "loss": 4.2188,
+      "step": 4722
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.837983293592974e-05,
+      "loss": 2.5,
+      "step": 4723
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8378993803422284e-05,
+      "loss": 3.4062,
+      "step": 4724
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.83781544628435e-05,
+      "loss": 3.875,
+      "step": 4725
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.837731491420288e-05,
+      "loss": 4.1719,
+      "step": 4726
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8376475157509936e-05,
+      "loss": 2.1875,
+      "step": 4727
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8375635192774183e-05,
+      "loss": 5.0938,
+      "step": 4728
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.837479502000512e-05,
+      "loss": 4.375,
+      "step": 4729
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.837395463921225e-05,
+      "loss": 4.6094,
+      "step": 4730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.8373114050405106e-05,
+      "loss": 4.875,
+      "step": 4731
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.83722732535932e-05,
+      "loss": 1.8594,
+      "step": 4732
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.837143224878604e-05,
+      "loss": 2.4062,
+      "step": 4733
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8370591035993164e-05,
+      "loss": 4.2656,
+      "step": 4734
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.836974961522408e-05,
+      "loss": 2.4375,
+      "step": 4735
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.836890798648832e-05,
+      "loss": 2.1875,
+      "step": 4736
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.836806614979541e-05,
+      "loss": 4.6094,
+      "step": 4737
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8367224105154885e-05,
+      "loss": 2.9688,
+      "step": 4738
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.836638185257628e-05,
+      "loss": 5.9688,
+      "step": 4739
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.836553939206911e-05,
+      "loss": 4.0469,
+      "step": 4740
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8364696723642945e-05,
+      "loss": 5.4062,
+      "step": 4741
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.83638538473073e-05,
+      "loss": 3.7031,
+      "step": 4742
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8363010763071734e-05,
+      "loss": 4.25,
+      "step": 4743
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.836216747094577e-05,
+      "loss": 3.9219,
+      "step": 4744
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.836132397093898e-05,
+      "loss": 5.9062,
+      "step": 4745
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.836048026306089e-05,
+      "loss": 2.4531,
+      "step": 4746
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.835963634732108e-05,
+      "loss": 1.9375,
+      "step": 4747
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8358792223729077e-05,
+      "loss": 3.8438,
+      "step": 4748
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.835794789229445e-05,
+      "loss": 4.0,
+      "step": 4749
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.835710335302675e-05,
+      "loss": 7.375,
+      "step": 4750
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.835625860593555e-05,
+      "loss": 2.5938,
+      "step": 4751
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.835541365103041e-05,
+      "loss": 4.25,
+      "step": 4752
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8354568488320884e-05,
+      "loss": 3.9375,
+      "step": 4753
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.835372311781655e-05,
+      "loss": 3.125,
+      "step": 4754
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.835287753952699e-05,
+      "loss": 3.4375,
+      "step": 4755
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.835203175346176e-05,
+      "loss": 4.3906,
+      "step": 4756
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8351185759630435e-05,
+      "loss": 4.1094,
+      "step": 4757
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.83503395580426e-05,
+      "loss": 3.9688,
+      "step": 4758
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.834949314870784e-05,
+      "loss": 3.5312,
+      "step": 4759
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.834864653163572e-05,
+      "loss": 5.8125,
+      "step": 4760
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8347799706835845e-05,
+      "loss": 2.625,
+      "step": 4761
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.8346952674317785e-05,
+      "loss": 5.5781,
+      "step": 4762
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.834610543409114e-05,
+      "loss": 5.0938,
+      "step": 4763
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.834525798616549e-05,
+      "loss": 5.6406,
+      "step": 4764
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8344410330550446e-05,
+      "loss": 5.2812,
+      "step": 4765
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8343562467255594e-05,
+      "loss": 7.3438,
+      "step": 4766
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.834271439629053e-05,
+      "loss": 2.2812,
+      "step": 4767
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8341866117664864e-05,
+      "loss": 2.7969,
+      "step": 4768
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8341017631388195e-05,
+      "loss": 2.6719,
+      "step": 4769
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.834016893747013e-05,
+      "loss": 1.2188,
+      "step": 4770
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.833932003592027e-05,
+      "loss": 2.0312,
+      "step": 4771
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.833847092674823e-05,
+      "loss": 8.125,
+      "step": 4772
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.833762160996363e-05,
+      "loss": 3.3906,
+      "step": 4773
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.833677208557608e-05,
+      "loss": 5.8281,
+      "step": 4774
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.83359223535952e-05,
+      "loss": 4.0,
+      "step": 4775
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.833507241403061e-05,
+      "loss": 4.0625,
+      "step": 4776
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.833422226689192e-05,
+      "loss": 4.5312,
+      "step": 4777
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.833337191218876e-05,
+      "loss": 3.4219,
+      "step": 4778
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8332521349930774e-05,
+      "loss": 4.0312,
+      "step": 4779
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.833167058012757e-05,
+      "loss": 2.5938,
+      "step": 4780
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.833081960278879e-05,
+      "loss": 7.3125,
+      "step": 4781
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.832996841792407e-05,
+      "loss": 2.625,
+      "step": 4782
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.832911702554304e-05,
+      "loss": 1.5312,
+      "step": 4783
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.832826542565534e-05,
+      "loss": 7.1562,
+      "step": 4784
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8327413618270616e-05,
+      "loss": 3.9219,
+      "step": 4785
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.83265616033985e-05,
+      "loss": 3.9375,
+      "step": 4786
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.832570938104866e-05,
+      "loss": 3.8125,
+      "step": 4787
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.8324856951230715e-05,
+      "loss": 2.4375,
+      "step": 4788
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.832400431395434e-05,
+      "loss": 6.3906,
+      "step": 4789
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.832315146922917e-05,
+      "loss": 2.9375,
+      "step": 4790
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.832229841706487e-05,
+      "loss": 2.3438,
+      "step": 4791
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.83214451574711e-05,
+      "loss": 3.4688,
+      "step": 4792
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.832059169045752e-05,
+      "loss": 3.8438,
+      "step": 4793
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.831973801603378e-05,
+      "loss": 3.625,
+      "step": 4794
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.831888413420956e-05,
+      "loss": 5.375,
+      "step": 4795
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.831803004499451e-05,
+      "loss": 3.3594,
+      "step": 4796
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.8317175748398317e-05,
+      "loss": 1.8125,
+      "step": 4797
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.831632124443063e-05,
+      "loss": 5.4062,
+      "step": 4798
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.8315466533101154e-05,
+      "loss": 5.2812,
+      "step": 4799
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.8314611614419544e-05,
+      "loss": 5.6562,
+      "step": 4800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.8313756488395476e-05,
+      "loss": 3.1406,
+      "step": 4801
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.831290115503865e-05,
+      "loss": 8.0469,
+      "step": 4802
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.831204561435873e-05,
+      "loss": 7.2969,
+      "step": 4803
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.831118986636541e-05,
+      "loss": 1.875,
+      "step": 4804
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.8310333911068376e-05,
+      "loss": 3.9688,
+      "step": 4805
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.830947774847732e-05,
+      "loss": 5.0781,
+      "step": 4806
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.8308621378601933e-05,
+      "loss": 4.2031,
+      "step": 4807
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.830776480145192e-05,
+      "loss": 4.4062,
+      "step": 4808
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.830690801703696e-05,
+      "loss": 5.2188,
+      "step": 4809
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.8306051025366775e-05,
+      "loss": 5.1328,
+      "step": 4810
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.8305193826451044e-05,
+      "loss": 2.2188,
+      "step": 4811
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.830433642029949e-05,
+      "loss": 5.5625,
+      "step": 4812
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.830347880692181e-05,
+      "loss": 5.8281,
+      "step": 4813
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.830262098632771e-05,
+      "loss": 6.75,
+      "step": 4814
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.8301762958526914e-05,
+      "loss": 4.25,
+      "step": 4815
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.830090472352912e-05,
+      "loss": 7.25,
+      "step": 4816
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.830004628134406e-05,
+      "loss": 6.4062,
+      "step": 4817
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.829918763198145e-05,
+      "loss": 5.7031,
+      "step": 4818
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.8298328775450995e-05,
+      "loss": 3.0469,
+      "step": 4819
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.8297469711762435e-05,
+      "loss": 3.9375,
+      "step": 4820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.82966104409255e-05,
+      "loss": 2.8906,
+      "step": 4821
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.829575096294989e-05,
+      "loss": 5.875,
+      "step": 4822
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.829489127784537e-05,
+      "loss": 3.6094,
+      "step": 4823
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.829403138562165e-05,
+      "loss": 5.4219,
+      "step": 4824
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.829317128628847e-05,
+      "loss": 6.125,
+      "step": 4825
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.829231097985557e-05,
+      "loss": 5.0469,
+      "step": 4826
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.829145046633269e-05,
+      "loss": 6.9844,
+      "step": 4827
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.829058974572956e-05,
+      "loss": 4.1719,
+      "step": 4828
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.828972881805594e-05,
+      "loss": 3.0,
+      "step": 4829
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.828886768332157e-05,
+      "loss": 7.4531,
+      "step": 4830
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.828800634153621e-05,
+      "loss": 3.6875,
+      "step": 4831
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.828714479270959e-05,
+      "loss": 4.0312,
+      "step": 4832
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.8286283036851476e-05,
+      "loss": 5.4062,
+      "step": 4833
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.828542107397163e-05,
+      "loss": 5.625,
+      "step": 4834
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.82845589040798e-05,
+      "loss": 5.4219,
+      "step": 4835
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.828369652718575e-05,
+      "loss": 4.4062,
+      "step": 4836
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.828283394329925e-05,
+      "loss": 3.625,
+      "step": 4837
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.828197115243005e-05,
+      "loss": 3.1562,
+      "step": 4838
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.828110815458793e-05,
+      "loss": 4.9062,
+      "step": 4839
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.828024494978266e-05,
+      "loss": 1.8438,
+      "step": 4840
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.8279381538024004e-05,
+      "loss": 3.0,
+      "step": 4841
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.8278517919321744e-05,
+      "loss": 2.9844,
+      "step": 4842
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.8277654093685654e-05,
+      "loss": 5.0469,
+      "step": 4843
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.827679006112552e-05,
+      "loss": 2.2188,
+      "step": 4844
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.8275925821651116e-05,
+      "loss": 3.0781,
+      "step": 4845
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.827506137527223e-05,
+      "loss": 3.3125,
+      "step": 4846
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.827419672199864e-05,
+      "loss": 3.1875,
+      "step": 4847
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.827333186184016e-05,
+      "loss": 1.5938,
+      "step": 4848
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.8272466794806546e-05,
+      "loss": 2.4688,
+      "step": 4849
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.8271601520907614e-05,
+      "loss": 4.2031,
+      "step": 4850
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.827073604015316e-05,
+      "loss": 3.6406,
+      "step": 4851
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.826987035255297e-05,
+      "loss": 7.0625,
+      "step": 4852
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.826900445811686e-05,
+      "loss": 5.7656,
+      "step": 4853
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.826813835685462e-05,
+      "loss": 3.6562,
+      "step": 4854
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.8267272048776056e-05,
+      "loss": 6.1562,
+      "step": 4855
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.826640553389098e-05,
+      "loss": 4.3594,
+      "step": 4856
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.826553881220921e-05,
+      "loss": 4.8594,
+      "step": 4857
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.8264671883740543e-05,
+      "loss": 1.875,
+      "step": 4858
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.82638047484948e-05,
+      "loss": 3.2812,
+      "step": 4859
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.8262937406481805e-05,
+      "loss": 1.7656,
+      "step": 4860
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.826206985771136e-05,
+      "loss": 4.7188,
+      "step": 4861
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.826120210219331e-05,
+      "loss": 6.3438,
+      "step": 4862
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.8260334139937456e-05,
+      "loss": 3.5781,
+      "step": 4863
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.8259465970953636e-05,
+      "loss": 4.1562,
+      "step": 4864
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.8258597595251684e-05,
+      "loss": 4.1094,
+      "step": 4865
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.825772901284142e-05,
+      "loss": 5.3438,
+      "step": 4866
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.825686022373268e-05,
+      "loss": 4.8438,
+      "step": 4867
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.825599122793531e-05,
+      "loss": 3.75,
+      "step": 4868
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.8255122025459126e-05,
+      "loss": 6.5469,
+      "step": 4869
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.8254252616313986e-05,
+      "loss": 5.0312,
+      "step": 4870
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.825338300050973e-05,
+      "loss": 6.2188,
+      "step": 4871
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.825251317805621e-05,
+      "loss": 5.3438,
+      "step": 4872
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.8251643148963255e-05,
+      "loss": 4.0312,
+      "step": 4873
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.825077291324073e-05,
+      "loss": 6.8906,
+      "step": 4874
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.8249902470898473e-05,
+      "loss": 4.8438,
+      "step": 4875
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.8249031821946356e-05,
+      "loss": 5.0156,
+      "step": 4876
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.824816096639422e-05,
+      "loss": 5.2969,
+      "step": 4877
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.824728990425194e-05,
+      "loss": 5.5781,
+      "step": 4878
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.824641863552936e-05,
+      "loss": 5.4062,
+      "step": 4879
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.8245547160236354e-05,
+      "loss": 4.7812,
+      "step": 4880
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.824467547838279e-05,
+      "loss": 4.9688,
+      "step": 4881
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.824380358997853e-05,
+      "loss": 3.0469,
+      "step": 4882
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.824293149503345e-05,
+      "loss": 4.9062,
+      "step": 4883
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.8242059193557415e-05,
+      "loss": 5.5625,
+      "step": 4884
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.8241186685560316e-05,
+      "loss": 4.0781,
+      "step": 4885
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.824031397105201e-05,
+      "loss": 5.1875,
+      "step": 4886
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.823944105004239e-05,
+      "loss": 4.1719,
+      "step": 4887
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.823856792254134e-05,
+      "loss": 2.9062,
+      "step": 4888
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.823769458855874e-05,
+      "loss": 4.4531,
+      "step": 4889
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.823682104810448e-05,
+      "loss": 4.9219,
+      "step": 4890
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.823594730118845e-05,
+      "loss": 5.2812,
+      "step": 4891
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.823507334782053e-05,
+      "loss": 6.7031,
+      "step": 4892
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8234199188010636e-05,
+      "loss": 1.75,
+      "step": 4893
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.823332482176865e-05,
+      "loss": 4.4375,
+      "step": 4894
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.823245024910446e-05,
+      "loss": 2.25,
+      "step": 4895
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.823157547002799e-05,
+      "loss": 3.9531,
+      "step": 4896
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8230700484549134e-05,
+      "loss": 5.6406,
+      "step": 4897
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.82298252926778e-05,
+      "loss": 5.1875,
+      "step": 4898
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8228949894423895e-05,
+      "loss": 3.4688,
+      "step": 4899
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.822807428979732e-05,
+      "loss": 5.0938,
+      "step": 4900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8227198478808006e-05,
+      "loss": 3.2969,
+      "step": 4901
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.822632246146585e-05,
+      "loss": 4.7812,
+      "step": 4902
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8225446237780784e-05,
+      "loss": 3.8906,
+      "step": 4903
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.822456980776272e-05,
+      "loss": 6.7188,
+      "step": 4904
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.822369317142158e-05,
+      "loss": 6.2656,
+      "step": 4905
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8222816328767293e-05,
+      "loss": 4.9219,
+      "step": 4906
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.822193927980978e-05,
+      "loss": 3.3438,
+      "step": 4907
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.822106202455898e-05,
+      "loss": 3.5,
+      "step": 4908
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8220184563024806e-05,
+      "loss": 3.3594,
+      "step": 4909
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8219306895217214e-05,
+      "loss": 5.8438,
+      "step": 4910
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.821842902114613e-05,
+      "loss": 3.75,
+      "step": 4911
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.821755094082149e-05,
+      "loss": 5.1719,
+      "step": 4912
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.8216672654253235e-05,
+      "loss": 4.9531,
+      "step": 4913
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.821579416145132e-05,
+      "loss": 3.3281,
+      "step": 4914
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.821491546242568e-05,
+      "loss": 2.7812,
+      "step": 4915
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.821403655718626e-05,
+      "loss": 4.3594,
+      "step": 4916
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.821315744574301e-05,
+      "loss": 2.1562,
+      "step": 4917
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.821227812810589e-05,
+      "loss": 2.3125,
+      "step": 4918
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.821139860428485e-05,
+      "loss": 5.2812,
+      "step": 4919
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8210518874289856e-05,
+      "loss": 3.6875,
+      "step": 4920
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.820963893813086e-05,
+      "loss": 3.7656,
+      "step": 4921
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.820875879581782e-05,
+      "loss": 4.9062,
+      "step": 4922
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8207878447360706e-05,
+      "loss": 3.5938,
+      "step": 4923
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.820699789276948e-05,
+      "loss": 4.3906,
+      "step": 4924
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8206117132054124e-05,
+      "loss": 4.6094,
+      "step": 4925
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8205236165224595e-05,
+      "loss": 2.8125,
+      "step": 4926
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.820435499229087e-05,
+      "loss": 6.5312,
+      "step": 4927
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.820347361326293e-05,
+      "loss": 6.4219,
+      "step": 4928
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.820259202815074e-05,
+      "loss": 6.5312,
+      "step": 4929
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8201710236964295e-05,
+      "loss": 6.8594,
+      "step": 4930
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8200828239713575e-05,
+      "loss": 4.625,
+      "step": 4931
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.819994603640856e-05,
+      "loss": 3.5312,
+      "step": 4932
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.819906362705924e-05,
+      "loss": 5.0625,
+      "step": 4933
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.819818101167561e-05,
+      "loss": 5.0781,
+      "step": 4934
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8197298190267644e-05,
+      "loss": 5.7344,
+      "step": 4935
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8196415162845364e-05,
+      "loss": 6.8281,
+      "step": 4936
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8195531929418755e-05,
+      "loss": 4.625,
+      "step": 4937
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.81946484899978e-05,
+      "loss": 5.0312,
+      "step": 4938
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8193764844592525e-05,
+      "loss": 1.7031,
+      "step": 4939
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.819288099321292e-05,
+      "loss": 8.7031,
+      "step": 4940
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8191996935868996e-05,
+      "loss": 3.9062,
+      "step": 4941
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8191112672570764e-05,
+      "loss": 3.3125,
+      "step": 4942
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.819022820332823e-05,
+      "loss": 7.2188,
+      "step": 4943
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.8189343528151404e-05,
+      "loss": 3.3125,
+      "step": 4944
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.81884586470503e-05,
+      "loss": 6.4062,
+      "step": 4945
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.818757356003495e-05,
+      "loss": 2.6406,
+      "step": 4946
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.818668826711537e-05,
+      "loss": 3.7188,
+      "step": 4947
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.8185802768301575e-05,
+      "loss": 4.8125,
+      "step": 4948
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.818491706360359e-05,
+      "loss": 6.3594,
+      "step": 4949
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.8184031153031454e-05,
+      "loss": 3.3438,
+      "step": 4950
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.818314503659518e-05,
+      "loss": 6.0625,
+      "step": 4951
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.818225871430481e-05,
+      "loss": 4.6875,
+      "step": 4952
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.818137218617038e-05,
+      "loss": 5.6719,
+      "step": 4953
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.818048545220192e-05,
+      "loss": 5.5312,
+      "step": 4954
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.817959851240947e-05,
+      "loss": 8.8594,
+      "step": 4955
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.817871136680307e-05,
+      "loss": 6.1875,
+      "step": 4956
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.817782401539277e-05,
+      "loss": 3.3906,
+      "step": 4957
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.817693645818861e-05,
+      "loss": 6.7188,
+      "step": 4958
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.8176048695200645e-05,
+      "loss": 6.5,
+      "step": 4959
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.817516072643892e-05,
+      "loss": 3.1562,
+      "step": 4960
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.817427255191349e-05,
+      "loss": 2.3438,
+      "step": 4961
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.817338417163441e-05,
+      "loss": 3.2031,
+      "step": 4962
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.817249558561172e-05,
+      "loss": 5.9375,
+      "step": 4963
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.817160679385551e-05,
+      "loss": 4.9375,
+      "step": 4964
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.817071779637583e-05,
+      "loss": 5.625,
+      "step": 4965
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.8169828593182735e-05,
+      "loss": 3.1094,
+      "step": 4966
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.816893918428631e-05,
+      "loss": 2.9688,
+      "step": 4967
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.81680495696966e-05,
+      "loss": 4.7188,
+      "step": 4968
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.81671597494237e-05,
+      "loss": 4.6875,
+      "step": 4969
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.816626972347767e-05,
+      "loss": 5.4688,
+      "step": 4970
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.816537949186859e-05,
+      "loss": 5.1875,
+      "step": 4971
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.816448905460654e-05,
+      "loss": 5.7969,
+      "step": 4972
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.81635984117016e-05,
+      "loss": 3.7969,
+      "step": 4973
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.816270756316385e-05,
+      "loss": 4.7344,
+      "step": 4974
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.816181650900338e-05,
+      "loss": 2.0312,
+      "step": 4975
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.816092524923027e-05,
+      "loss": 2.9688,
+      "step": 4976
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.816003378385462e-05,
+      "loss": 5.6094,
+      "step": 4977
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.8159142112886514e-05,
+      "loss": 6.0312,
+      "step": 4978
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.815825023633606e-05,
+      "loss": 4.7188,
+      "step": 4979
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.815735815421333e-05,
+      "loss": 3.7812,
+      "step": 4980
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.815646586652845e-05,
+      "loss": 5.0781,
+      "step": 4981
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.815557337329152e-05,
+      "loss": 5.4531,
+      "step": 4982
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.815468067451262e-05,
+      "loss": 7.0156,
+      "step": 4983
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.815378777020187e-05,
+      "loss": 2.4062,
+      "step": 4984
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.815289466036939e-05,
+      "loss": 6.25,
+      "step": 4985
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.815200134502528e-05,
+      "loss": 3.2188,
+      "step": 4986
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.815110782417965e-05,
+      "loss": 4.375,
+      "step": 4987
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.815021409784262e-05,
+      "loss": 3.6719,
+      "step": 4988
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.814932016602431e-05,
+      "loss": 4.3125,
+      "step": 4989
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.8148426028734834e-05,
+      "loss": 4.0469,
+      "step": 4990
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.814753168598433e-05,
+      "loss": 2.9375,
+      "step": 4991
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.8146637137782905e-05,
+      "loss": 4.0469,
+      "step": 4992
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.8145742384140695e-05,
+      "loss": 3.375,
+      "step": 4993
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.814484742506782e-05,
+      "loss": 4.375,
+      "step": 4994
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.814395226057443e-05,
+      "loss": 9.5312,
+      "step": 4995
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.8143056890670645e-05,
+      "loss": 6.8125,
+      "step": 4996
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.814216131536661e-05,
+      "loss": 6.4375,
+      "step": 4997
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.8141265534672456e-05,
+      "loss": 3.8438,
+      "step": 4998
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.8140369548598324e-05,
+      "loss": 3.5312,
+      "step": 4999
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.8139473357154365e-05,
+      "loss": 3.5938,
+      "step": 5000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.813857696035073e-05,
+      "loss": 4.2188,
+      "step": 5001
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.813768035819755e-05,
+      "loss": 4.1406,
+      "step": 5002
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.813678355070499e-05,
+      "loss": 2.3438,
+      "step": 5003
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.813588653788319e-05,
+      "loss": 2.9844,
+      "step": 5004
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.813498931974232e-05,
+      "loss": 3.4062,
+      "step": 5005
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.813409189629252e-05,
+      "loss": 4.4219,
+      "step": 5006
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.813319426754396e-05,
+      "loss": 5.25,
+      "step": 5007
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.813229643350682e-05,
+      "loss": 5.4375,
+      "step": 5008
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.813139839419124e-05,
+      "loss": 4.9062,
+      "step": 5009
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.813050014960738e-05,
+      "loss": 7.2812,
+      "step": 5010
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.8129601699765425e-05,
+      "loss": 5.9375,
+      "step": 5011
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.8128703044675555e-05,
+      "loss": 5.2344,
+      "step": 5012
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.812780418434793e-05,
+      "loss": 4.3438,
+      "step": 5013
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.812690511879273e-05,
+      "loss": 5.5938,
+      "step": 5014
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.812600584802013e-05,
+      "loss": 4.5625,
+      "step": 5015
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.812510637204031e-05,
+      "loss": 5.8125,
+      "step": 5016
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.812420669086346e-05,
+      "loss": 4.0938,
+      "step": 5017
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.812330680449976e-05,
+      "loss": 5.2188,
+      "step": 5018
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.81224067129594e-05,
+      "loss": 2.5938,
+      "step": 5019
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.812150641625257e-05,
+      "loss": 3.5469,
+      "step": 5020
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.8120605914389464e-05,
+      "loss": 4.9531,
+      "step": 5021
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.8119705207380277e-05,
+      "loss": 4.9062,
+      "step": 5022
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.81188042952352e-05,
+      "loss": 5.1875,
+      "step": 5023
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.811790317796443e-05,
+      "loss": 5.3438,
+      "step": 5024
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.8117001855578185e-05,
+      "loss": 2.7656,
+      "step": 5025
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.811610032808665e-05,
+      "loss": 2.125,
+      "step": 5026
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.811519859550005e-05,
+      "loss": 5.8438,
+      "step": 5027
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.811429665782858e-05,
+      "loss": 5.4375,
+      "step": 5028
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.811339451508245e-05,
+      "loss": 5.2656,
+      "step": 5029
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.8112492167271886e-05,
+      "loss": 5.0078,
+      "step": 5030
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.811158961440709e-05,
+      "loss": 5.5625,
+      "step": 5031
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.8110686856498286e-05,
+      "loss": 3.8594,
+      "step": 5032
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.8109783893555696e-05,
+      "loss": 3.3438,
+      "step": 5033
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.810888072558954e-05,
+      "loss": 3.8906,
+      "step": 5034
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.810797735261004e-05,
+      "loss": 5.7656,
+      "step": 5035
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.810707377462743e-05,
+      "loss": 9.1719,
+      "step": 5036
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.810616999165194e-05,
+      "loss": 3.8438,
+      "step": 5037
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.810526600369379e-05,
+      "loss": 6.7188,
+      "step": 5038
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.810436181076322e-05,
+      "loss": 5.4219,
+      "step": 5039
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.810345741287048e-05,
+      "loss": 4.3828,
+      "step": 5040
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.810255281002579e-05,
+      "loss": 3.4844,
+      "step": 5041
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8101648002239405e-05,
+      "loss": 4.625,
+      "step": 5042
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8100742989521556e-05,
+      "loss": 3.3281,
+      "step": 5043
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8099837771882495e-05,
+      "loss": 3.4531,
+      "step": 5044
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.809893234933247e-05,
+      "loss": 6.0469,
+      "step": 5045
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.809802672188173e-05,
+      "loss": 3.4062,
+      "step": 5046
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8097120889540536e-05,
+      "loss": 5.1875,
+      "step": 5047
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.809621485231914e-05,
+      "loss": 3.7812,
+      "step": 5048
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8095308610227786e-05,
+      "loss": 3.25,
+      "step": 5049
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8094402163276745e-05,
+      "loss": 5.1562,
+      "step": 5050
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8093495511476275e-05,
+      "loss": 2.5469,
+      "step": 5051
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.809258865483665e-05,
+      "loss": 3.7031,
+      "step": 5052
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.809168159336812e-05,
+      "loss": 4.4375,
+      "step": 5053
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8090774327080966e-05,
+      "loss": 2.625,
+      "step": 5054
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8089866855985465e-05,
+      "loss": 2.9688,
+      "step": 5055
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.808895918009187e-05,
+      "loss": 6.2344,
+      "step": 5056
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.808805129941048e-05,
+      "loss": 5.6094,
+      "step": 5057
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.808714321395155e-05,
+      "loss": 3.0625,
+      "step": 5058
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8086234923725385e-05,
+      "loss": 3.375,
+      "step": 5059
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.808532642874225e-05,
+      "loss": 5.0781,
+      "step": 5060
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.808441772901244e-05,
+      "loss": 3.4844,
+      "step": 5061
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.808350882454623e-05,
+      "loss": 2.8906,
+      "step": 5062
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.808259971535393e-05,
+      "loss": 6.0,
+      "step": 5063
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8081690401445815e-05,
+      "loss": 6.7031,
+      "step": 5064
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8080780882832187e-05,
+      "loss": 8.875,
+      "step": 5065
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.807987115952333e-05,
+      "loss": 6.4844,
+      "step": 5066
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.807896123152957e-05,
+      "loss": 4.625,
+      "step": 5067
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.807805109886118e-05,
+      "loss": 3.375,
+      "step": 5068
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8077140761528486e-05,
+      "loss": 5.7344,
+      "step": 5069
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.807623021954178e-05,
+      "loss": 5.0,
+      "step": 5070
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.807531947291137e-05,
+      "loss": 3.8594,
+      "step": 5071
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8074408521647576e-05,
+      "loss": 3.4531,
+      "step": 5072
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.807349736576071e-05,
+      "loss": 3.75,
+      "step": 5073
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.807258600526107e-05,
+      "loss": 5.4375,
+      "step": 5074
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8071674440159e-05,
+      "loss": 4.0781,
+      "step": 5075
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.807076267046481e-05,
+      "loss": 2.7656,
+      "step": 5076
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.806985069618881e-05,
+      "loss": 5.8438,
+      "step": 5077
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.806893851734134e-05,
+      "loss": 3.0312,
+      "step": 5078
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.806802613393272e-05,
+      "loss": 3.3594,
+      "step": 5079
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8067113545973276e-05,
+      "loss": 2.5312,
+      "step": 5080
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8066200753473345e-05,
+      "loss": 4.4688,
+      "step": 5081
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.806528775644326e-05,
+      "loss": 3.9375,
+      "step": 5082
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.806437455489336e-05,
+      "loss": 3.0938,
+      "step": 5083
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.806346114883398e-05,
+      "loss": 5.0312,
+      "step": 5084
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8062547538275464e-05,
+      "loss": 2.7812,
+      "step": 5085
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.806163372322815e-05,
+      "loss": 3.4062,
+      "step": 5086
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.806071970370238e-05,
+      "loss": 3.4688,
+      "step": 5087
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.805980547970851e-05,
+      "loss": 1.8125,
+      "step": 5088
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8058891051256896e-05,
+      "loss": 3.3125,
+      "step": 5089
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8057976418357875e-05,
+      "loss": 3.6562,
+      "step": 5090
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.80570615810218e-05,
+      "loss": 2.4688,
+      "step": 5091
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.805614653925905e-05,
+      "loss": 4.3906,
+      "step": 5092
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8055231293079965e-05,
+      "loss": 4.7031,
+      "step": 5093
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.805431584249492e-05,
+      "loss": 3.1875,
+      "step": 5094
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.805340018751426e-05,
+      "loss": 6.2031,
+      "step": 5095
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8052484328148375e-05,
+      "loss": 5.2656,
+      "step": 5096
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.805156826440761e-05,
+      "loss": 4.625,
+      "step": 5097
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8050651996302356e-05,
+      "loss": 2.2188,
+      "step": 5098
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.804973552384297e-05,
+      "loss": 8.2031,
+      "step": 5099
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.804881884703984e-05,
+      "loss": 4.6875,
+      "step": 5100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.804790196590333e-05,
+      "loss": 4.6406,
+      "step": 5101
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.804698488044384e-05,
+      "loss": 3.4062,
+      "step": 5102
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.804606759067174e-05,
+      "loss": 2.0,
+      "step": 5103
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8045150096597416e-05,
+      "loss": 4.1094,
+      "step": 5104
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.804423239823125e-05,
+      "loss": 4.2188,
+      "step": 5105
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.804331449558364e-05,
+      "loss": 2.1875,
+      "step": 5106
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.804239638866498e-05,
+      "loss": 3.6094,
+      "step": 5107
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.804147807748565e-05,
+      "loss": 2.5781,
+      "step": 5108
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8040559562056055e-05,
+      "loss": 1.8125,
+      "step": 5109
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.80396408423866e-05,
+      "loss": 3.5,
+      "step": 5110
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8038721918487674e-05,
+      "loss": 4.2344,
+      "step": 5111
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8037802790369685e-05,
+      "loss": 4.3125,
+      "step": 5112
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.803688345804305e-05,
+      "loss": 4.875,
+      "step": 5113
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.803596392151815e-05,
+      "loss": 4.2031,
+      "step": 5114
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.803504418080542e-05,
+      "loss": 5.6562,
+      "step": 5115
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8034124235915265e-05,
+      "loss": 4.0938,
+      "step": 5116
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8033204086858104e-05,
+      "loss": 5.1875,
+      "step": 5117
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.803228373364434e-05,
+      "loss": 3.7812,
+      "step": 5118
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.80313631762844e-05,
+      "loss": 1.8125,
+      "step": 5119
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.803044241478871e-05,
+      "loss": 5.3438,
+      "step": 5120
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.802952144916769e-05,
+      "loss": 3.7344,
+      "step": 5121
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.802860027943177e-05,
+      "loss": 3.7656,
+      "step": 5122
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.802767890559138e-05,
+      "loss": 5.2109,
+      "step": 5123
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.802675732765694e-05,
+      "loss": 3.0938,
+      "step": 5124
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.802583554563889e-05,
+      "loss": 3.4375,
+      "step": 5125
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.802491355954767e-05,
+      "loss": 6.0312,
+      "step": 5126
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8023991369393714e-05,
+      "loss": 6.5469,
+      "step": 5127
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8023068975187465e-05,
+      "loss": 5.125,
+      "step": 5128
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8022146376939354e-05,
+      "loss": 4.1562,
+      "step": 5129
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8021223574659844e-05,
+      "loss": 4.2812,
+      "step": 5130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.802030056835937e-05,
+      "loss": 4.7812,
+      "step": 5131
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.801937735804838e-05,
+      "loss": 7.0312,
+      "step": 5132
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8018453943737344e-05,
+      "loss": 3.9844,
+      "step": 5133
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8017530325436694e-05,
+      "loss": 7.6562,
+      "step": 5134
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.801660650315689e-05,
+      "loss": 4.8906,
+      "step": 5135
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.80156824769084e-05,
+      "loss": 6.0938,
+      "step": 5136
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.801475824670168e-05,
+      "loss": 7.75,
+      "step": 5137
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8013833812547196e-05,
+      "loss": 6.7188,
+      "step": 5138
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.801290917445541e-05,
+      "loss": 3.4688,
+      "step": 5139
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.801198433243679e-05,
+      "loss": 2.8281,
+      "step": 5140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.801105928650181e-05,
+      "loss": 5.5547,
+      "step": 5141
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.801013403666094e-05,
+      "loss": 7.2656,
+      "step": 5142
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8009208582924656e-05,
+      "loss": 3.9375,
+      "step": 5143
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.800828292530344e-05,
+      "loss": 3.5312,
+      "step": 5144
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.800735706380776e-05,
+      "loss": 6.4688,
+      "step": 5145
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.80064309984481e-05,
+      "loss": 3.8438,
+      "step": 5146
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.800550472923495e-05,
+      "loss": 4.8281,
+      "step": 5147
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.80045782561788e-05,
+      "loss": 3.8438,
+      "step": 5148
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8003651579290125e-05,
+      "loss": 3.9375,
+      "step": 5149
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8002724698579426e-05,
+      "loss": 3.0312,
+      "step": 5150
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8001797614057204e-05,
+      "loss": 6.0,
+      "step": 5151
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8000870325733935e-05,
+      "loss": 2.3281,
+      "step": 5152
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.799994283362013e-05,
+      "loss": 4.6875,
+      "step": 5153
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.799901513772629e-05,
+      "loss": 4.2344,
+      "step": 5154
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.7998087238062915e-05,
+      "loss": 6.4844,
+      "step": 5155
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.7997159134640506e-05,
+      "loss": 6.9062,
+      "step": 5156
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.7996230827469565e-05,
+      "loss": 4.5938,
+      "step": 5157
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.799530231656062e-05,
+      "loss": 4.7969,
+      "step": 5158
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.7994373601924167e-05,
+      "loss": 6.625,
+      "step": 5159
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.799344468357073e-05,
+      "loss": 3.6641,
+      "step": 5160
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.799251556151082e-05,
+      "loss": 4.25,
+      "step": 5161
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.799158623575496e-05,
+      "loss": 2.4062,
+      "step": 5162
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.7990656706313665e-05,
+      "loss": 5.2188,
+      "step": 5163
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.798972697319745e-05,
+      "loss": 4.0,
+      "step": 5164
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.798879703641687e-05,
+      "loss": 4.375,
+      "step": 5165
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.798786689598243e-05,
+      "loss": 3.5312,
+      "step": 5166
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.798693655190466e-05,
+      "loss": 5.5938,
+      "step": 5167
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.7986006004194095e-05,
+      "loss": 4.2188,
+      "step": 5168
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.798507525286128e-05,
+      "loss": 3.6875,
+      "step": 5169
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.7984144297916736e-05,
+      "loss": 3.4688,
+      "step": 5170
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.798321313937102e-05,
+      "loss": 4.25,
+      "step": 5171
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.798228177723466e-05,
+      "loss": 4.6719,
+      "step": 5172
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.7981350211518206e-05,
+      "loss": 4.8281,
+      "step": 5173
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.79804184422322e-05,
+      "loss": 2.6406,
+      "step": 5174
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.7979486469387204e-05,
+      "loss": 1.8438,
+      "step": 5175
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.7978554292993755e-05,
+      "loss": 3.8438,
+      "step": 5176
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.79776219130624e-05,
+      "loss": 3.25,
+      "step": 5177
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.797668932960371e-05,
+      "loss": 3.6562,
+      "step": 5178
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.797575654262825e-05,
+      "loss": 2.5625,
+      "step": 5179
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.7974823552146555e-05,
+      "loss": 5.5156,
+      "step": 5180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.79738903581692e-05,
+      "loss": 3.1406,
+      "step": 5181
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.7972956960706765e-05,
+      "loss": 3.625,
+      "step": 5182
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.797202335976979e-05,
+      "loss": 3.5,
+      "step": 5183
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.797108955536886e-05,
+      "loss": 4.2188,
+      "step": 5184
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.797015554751454e-05,
+      "loss": 6.0,
+      "step": 5185
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.796922133621741e-05,
+      "loss": 3.5625,
+      "step": 5186
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.796828692148805e-05,
+      "loss": 2.5,
+      "step": 5187
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.796735230333703e-05,
+      "loss": 6.125,
+      "step": 5188
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.796641748177493e-05,
+      "loss": 3.6562,
+      "step": 5189
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.7965482456812334e-05,
+      "loss": 3.5312,
+      "step": 5190
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.796454722845984e-05,
+      "loss": 2.3594,
+      "step": 5191
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.796361179672801e-05,
+      "loss": 6.8125,
+      "step": 5192
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.796267616162747e-05,
+      "loss": 4.6094,
+      "step": 5193
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.796174032316879e-05,
+      "loss": 2.0781,
+      "step": 5194
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7960804281362554e-05,
+      "loss": 3.25,
+      "step": 5195
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.795986803621939e-05,
+      "loss": 3.6094,
+      "step": 5196
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7958931587749865e-05,
+      "loss": 5.9688,
+      "step": 5197
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7957994935964605e-05,
+      "loss": 5.9688,
+      "step": 5198
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7957058080874206e-05,
+      "loss": 3.5156,
+      "step": 5199
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.795612102248927e-05,
+      "loss": 3.8906,
+      "step": 5200
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.795518376082041e-05,
+      "loss": 3.8906,
+      "step": 5201
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.795424629587823e-05,
+      "loss": 2.6875,
+      "step": 5202
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.795330862767336e-05,
+      "loss": 3.125,
+      "step": 5203
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.795237075621639e-05,
+      "loss": 4.5469,
+      "step": 5204
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7951432681517964e-05,
+      "loss": 6.5,
+      "step": 5205
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.795049440358869e-05,
+      "loss": 5.9375,
+      "step": 5206
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7949555922439185e-05,
+      "loss": 6.25,
+      "step": 5207
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.794861723808008e-05,
+      "loss": 4.0938,
+      "step": 5208
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7947678350522e-05,
+      "loss": 3.5938,
+      "step": 5209
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.794673925977558e-05,
+      "loss": 3.0469,
+      "step": 5210
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7945799965851445e-05,
+      "loss": 3.7812,
+      "step": 5211
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.794486046876023e-05,
+      "loss": 2.2812,
+      "step": 5212
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7943920768512575e-05,
+      "loss": 2.9844,
+      "step": 5213
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.794298086511911e-05,
+      "loss": 3.9375,
+      "step": 5214
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.794204075859049e-05,
+      "loss": 3.8438,
+      "step": 5215
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7941100448937344e-05,
+      "loss": 3.0312,
+      "step": 5216
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.794015993617032e-05,
+      "loss": 3.5,
+      "step": 5217
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.793921922030007e-05,
+      "loss": 3.2344,
+      "step": 5218
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.793827830133725e-05,
+      "loss": 5.3281,
+      "step": 5219
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.7937337179292495e-05,
+      "loss": 4.0,
+      "step": 5220
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7936395854176475e-05,
+      "loss": 2.0938,
+      "step": 5221
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7935454325999844e-05,
+      "loss": 4.9062,
+      "step": 5222
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7934512594773254e-05,
+      "loss": 4.9531,
+      "step": 5223
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.793357066050737e-05,
+      "loss": 5.9688,
+      "step": 5224
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7932628523212855e-05,
+      "loss": 6.7188,
+      "step": 5225
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.793168618290039e-05,
+      "loss": 8.6875,
+      "step": 5226
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.793074363958061e-05,
+      "loss": 5.5312,
+      "step": 5227
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.792980089326422e-05,
+      "loss": 5.0469,
+      "step": 5228
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.792885794396187e-05,
+      "loss": 3.3125,
+      "step": 5229
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7927914791684255e-05,
+      "loss": 3.2344,
+      "step": 5230
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.792697143644204e-05,
+      "loss": 2.125,
+      "step": 5231
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.79260278782459e-05,
+      "loss": 3.8906,
+      "step": 5232
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.792508411710652e-05,
+      "loss": 4.5625,
+      "step": 5233
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.79241401530346e-05,
+      "loss": 6.25,
+      "step": 5234
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7923195986040814e-05,
+      "loss": 6.5,
+      "step": 5235
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7922251616135846e-05,
+      "loss": 5.5938,
+      "step": 5236
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7921307043330395e-05,
+      "loss": 6.2812,
+      "step": 5237
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.792036226763515e-05,
+      "loss": 4.0,
+      "step": 5238
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.791941728906081e-05,
+      "loss": 6.4375,
+      "step": 5239
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.791847210761808e-05,
+      "loss": 5.3438,
+      "step": 5240
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.791752672331765e-05,
+      "loss": 4.8438,
+      "step": 5241
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7916581136170226e-05,
+      "loss": 3.8281,
+      "step": 5242
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7915635346186515e-05,
+      "loss": 3.75,
+      "step": 5243
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.791468935337722e-05,
+      "loss": 3.5469,
+      "step": 5244
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.791374315775306e-05,
+      "loss": 2.5,
+      "step": 5245
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7912796759324735e-05,
+      "loss": 2.1875,
+      "step": 5246
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.791185015810297e-05,
+      "loss": 2.4062,
+      "step": 5247
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7910903354098476e-05,
+      "loss": 4.0469,
+      "step": 5248
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.7909956347321965e-05,
+      "loss": 3.9062,
+      "step": 5249
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.790900913778418e-05,
+      "loss": 3.75,
+      "step": 5250
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.790806172549582e-05,
+      "loss": 3.3438,
+      "step": 5251
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7907114110467625e-05,
+      "loss": 6.0781,
+      "step": 5252
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7906166292710314e-05,
+      "loss": 4.1094,
+      "step": 5253
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.790521827223463e-05,
+      "loss": 8.6719,
+      "step": 5254
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.790427004905129e-05,
+      "loss": 5.2188,
+      "step": 5255
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.790332162317104e-05,
+      "loss": 4.9062,
+      "step": 5256
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.790237299460462e-05,
+      "loss": 4.8906,
+      "step": 5257
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.790142416336276e-05,
+      "loss": 4.0469,
+      "step": 5258
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.790047512945621e-05,
+      "loss": 4.0156,
+      "step": 5259
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.789952589289571e-05,
+      "loss": 5.2266,
+      "step": 5260
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7898576453692e-05,
+      "loss": 5.2969,
+      "step": 5261
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.789762681185584e-05,
+      "loss": 4.5625,
+      "step": 5262
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7896676967397976e-05,
+      "loss": 5.0625,
+      "step": 5263
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.789572692032916e-05,
+      "loss": 3.5,
+      "step": 5264
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.789477667066015e-05,
+      "loss": 5.9531,
+      "step": 5265
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.78938262184017e-05,
+      "loss": 3.375,
+      "step": 5266
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7892875563564585e-05,
+      "loss": 6.3125,
+      "step": 5267
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.789192470615955e-05,
+      "loss": 1.9062,
+      "step": 5268
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.789097364619736e-05,
+      "loss": 4.4531,
+      "step": 5269
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.789002238368879e-05,
+      "loss": 4.0312,
+      "step": 5270
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.788907091864461e-05,
+      "loss": 4.4688,
+      "step": 5271
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7888119251075595e-05,
+      "loss": 7.0469,
+      "step": 5272
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.788716738099251e-05,
+      "loss": 4.4688,
+      "step": 5273
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7886215308406123e-05,
+      "loss": 4.125,
+      "step": 5274
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7885263033327236e-05,
+      "loss": 4.6875,
+      "step": 5275
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7884310555766614e-05,
+      "loss": 4.0,
+      "step": 5276
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7883357875735044e-05,
+      "loss": 2.1094,
+      "step": 5277
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.788240499324331e-05,
+      "loss": 3.0938,
+      "step": 5278
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.78814519083022e-05,
+      "loss": 3.4375,
+      "step": 5279
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.788049862092251e-05,
+      "loss": 3.6562,
+      "step": 5280
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.787954513111502e-05,
+      "loss": 4.7344,
+      "step": 5281
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.787859143889054e-05,
+      "loss": 2.4062,
+      "step": 5282
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7877637544259854e-05,
+      "loss": 4.4062,
+      "step": 5283
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7876683447233765e-05,
+      "loss": 3.9062,
+      "step": 5284
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.787572914782307e-05,
+      "loss": 4.1875,
+      "step": 5285
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.787477464603859e-05,
+      "loss": 6.5156,
+      "step": 5286
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.787381994189111e-05,
+      "loss": 2.9062,
+      "step": 5287
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7872865035391454e-05,
+      "loss": 4.1875,
+      "step": 5288
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7871909926550425e-05,
+      "loss": 3.9688,
+      "step": 5289
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.787095461537884e-05,
+      "loss": 5.1094,
+      "step": 5290
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7869999101887505e-05,
+      "loss": 2.9688,
+      "step": 5291
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7869043386087243e-05,
+      "loss": 2.6094,
+      "step": 5292
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.786808746798888e-05,
+      "loss": 5.9062,
+      "step": 5293
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.786713134760323e-05,
+      "loss": 5.1562,
+      "step": 5294
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.786617502494112e-05,
+      "loss": 4.8281,
+      "step": 5295
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7865218500013375e-05,
+      "loss": 5.6406,
+      "step": 5296
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.786426177283083e-05,
+      "loss": 6.4688,
+      "step": 5297
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.786330484340431e-05,
+      "loss": 5.6875,
+      "step": 5298
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7862347711744655e-05,
+      "loss": 4.0312,
+      "step": 5299
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.786139037786269e-05,
+      "loss": 5.1562,
+      "step": 5300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.786043284176926e-05,
+      "loss": 2.8281,
+      "step": 5301
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.785947510347521e-05,
+      "loss": 2.6875,
+      "step": 5302
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7858517162991375e-05,
+      "loss": 2.9375,
+      "step": 5303
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7857559020328604e-05,
+      "loss": 5.2656,
+      "step": 5304
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7856600675497736e-05,
+      "loss": 3.7969,
+      "step": 5305
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7855642128509635e-05,
+      "loss": 6.7188,
+      "step": 5306
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.785468337937514e-05,
+      "loss": 3.4062,
+      "step": 5307
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7853724428105114e-05,
+      "loss": 5.0625,
+      "step": 5308
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.785276527471041e-05,
+      "loss": 2.7812,
+      "step": 5309
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7851805919201885e-05,
+      "loss": 2.8125,
+      "step": 5310
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7850846361590396e-05,
+      "loss": 2.5312,
+      "step": 5311
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.784988660188682e-05,
+      "loss": 2.7656,
+      "step": 5312
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.784892664010202e-05,
+      "loss": 4.2188,
+      "step": 5313
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.784796647624685e-05,
+      "loss": 4.5,
+      "step": 5314
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.784700611033219e-05,
+      "loss": 4.9375,
+      "step": 5315
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.784604554236891e-05,
+      "loss": 1.8594,
+      "step": 5316
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7845084772367885e-05,
+      "loss": 4.5,
+      "step": 5317
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.784412380033999e-05,
+      "loss": 4.7969,
+      "step": 5318
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7843162626296114e-05,
+      "loss": 5.1875,
+      "step": 5319
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.784220125024713e-05,
+      "loss": 3.9688,
+      "step": 5320
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7841239672203923e-05,
+      "loss": 3.8281,
+      "step": 5321
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.784027789217738e-05,
+      "loss": 5.2812,
+      "step": 5322
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.783931591017838e-05,
+      "loss": 3.4062,
+      "step": 5323
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.783835372621783e-05,
+      "loss": 6.0156,
+      "step": 5324
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.783739134030662e-05,
+      "loss": 2.5938,
+      "step": 5325
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.783642875245564e-05,
+      "loss": 3.5781,
+      "step": 5326
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.783546596267578e-05,
+      "loss": 2.4844,
+      "step": 5327
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.783450297097796e-05,
+      "loss": 5.0312,
+      "step": 5328
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.783353977737306e-05,
+      "loss": 3.0156,
+      "step": 5329
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.783257638187201e-05,
+      "loss": 3.3438,
+      "step": 5330
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7831612784485695e-05,
+      "loss": 4.2188,
+      "step": 5331
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7830648985225025e-05,
+      "loss": 3.7812,
+      "step": 5332
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.782968498410093e-05,
+      "loss": 4.0938,
+      "step": 5333
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.78287207811243e-05,
+      "loss": 3.0312,
+      "step": 5334
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.782775637630606e-05,
+      "loss": 4.6719,
+      "step": 5335
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7826791769657135e-05,
+      "loss": 5.25,
+      "step": 5336
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.782582696118844e-05,
+      "loss": 5.3125,
+      "step": 5337
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.78248619509109e-05,
+      "loss": 5.7656,
+      "step": 5338
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.782389673883543e-05,
+      "loss": 3.0625,
+      "step": 5339
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.782293132497298e-05,
+      "loss": 4.1875,
+      "step": 5340
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7821965709334456e-05,
+      "loss": 5.7188,
+      "step": 5341
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.78209998919308e-05,
+      "loss": 6.9375,
+      "step": 5342
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.782003387277295e-05,
+      "loss": 5.625,
+      "step": 5343
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7819067651871835e-05,
+      "loss": 4.8438,
+      "step": 5344
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7818101229238394e-05,
+      "loss": 4.7812,
+      "step": 5345
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.781713460488357e-05,
+      "loss": 3.6719,
+      "step": 5346
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.781616777881831e-05,
+      "loss": 3.375,
+      "step": 5347
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7815200751053554e-05,
+      "loss": 3.3125,
+      "step": 5348
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.781423352160025e-05,
+      "loss": 7.7344,
+      "step": 5349
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.781326609046936e-05,
+      "loss": 3.875,
+      "step": 5350
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7812298457671815e-05,
+      "loss": 3.25,
+      "step": 5351
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.781133062321859e-05,
+      "loss": 7.4219,
+      "step": 5352
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7810362587120626e-05,
+      "loss": 4.3594,
+      "step": 5353
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7809394349388895e-05,
+      "loss": 5.7656,
+      "step": 5354
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7808425910034354e-05,
+      "loss": 4.8594,
+      "step": 5355
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.780745726906796e-05,
+      "loss": 4.3281,
+      "step": 5356
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.780648842650069e-05,
+      "loss": 2.3438,
+      "step": 5357
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.780551938234351e-05,
+      "loss": 4.375,
+      "step": 5358
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7804550136607384e-05,
+      "loss": 3.5625,
+      "step": 5359
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.780358068930329e-05,
+      "loss": 4.4688,
+      "step": 5360
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.78026110404422e-05,
+      "loss": 5.5156,
+      "step": 5361
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.78016411900351e-05,
+      "loss": 5.5,
+      "step": 5362
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.780067113809297e-05,
+      "loss": 2.6094,
+      "step": 5363
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.779970088462677e-05,
+      "loss": 4.0625,
+      "step": 5364
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.779873042964751e-05,
+      "loss": 7.2656,
+      "step": 5365
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7797759773166166e-05,
+      "loss": 5.9688,
+      "step": 5366
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.779678891519373e-05,
+      "loss": 2.2031,
+      "step": 5367
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7795817855741186e-05,
+      "loss": 4.7812,
+      "step": 5368
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7794846594819534e-05,
+      "loss": 6.2188,
+      "step": 5369
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.779387513243977e-05,
+      "loss": 4.4375,
+      "step": 5370
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.779290346861289e-05,
+      "loss": 4.5312,
+      "step": 5371
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.77919316033499e-05,
+      "loss": 2.25,
+      "step": 5372
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.779095953666179e-05,
+      "loss": 4.9219,
+      "step": 5373
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7789987268559575e-05,
+      "loss": 3.1562,
+      "step": 5374
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.778901479905426e-05,
+      "loss": 3.4688,
+      "step": 5375
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.778804212815686e-05,
+      "loss": 5.1875,
+      "step": 5376
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.778706925587837e-05,
+      "loss": 7.6875,
+      "step": 5377
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.778609618222983e-05,
+      "loss": 5.5625,
+      "step": 5378
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7785122907222226e-05,
+      "loss": 3.5,
+      "step": 5379
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.77841494308666e-05,
+      "loss": 4.9844,
+      "step": 5380
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7783175753173964e-05,
+      "loss": 2.5,
+      "step": 5381
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7782201874155344e-05,
+      "loss": 1.8281,
+      "step": 5382
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.778122779382176e-05,
+      "loss": 2.9062,
+      "step": 5383
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.778025351218424e-05,
+      "loss": 5.8281,
+      "step": 5384
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.777927902925383e-05,
+      "loss": 4.2812,
+      "step": 5385
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7778304345041536e-05,
+      "loss": 5.3906,
+      "step": 5386
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.777732945955841e-05,
+      "loss": 3.3906,
+      "step": 5387
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.777635437281548e-05,
+      "loss": 5.2188,
+      "step": 5388
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.77753790848238e-05,
+      "loss": 3.3125,
+      "step": 5389
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.777440359559439e-05,
+      "loss": 5.8438,
+      "step": 5390
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.777342790513832e-05,
+      "loss": 3.9688,
+      "step": 5391
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7772452013466604e-05,
+      "loss": 3.8906,
+      "step": 5392
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.777147592059032e-05,
+      "loss": 3.7812,
+      "step": 5393
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.77704996265205e-05,
+      "loss": 3.625,
+      "step": 5394
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.77695231312682e-05,
+      "loss": 4.2812,
+      "step": 5395
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.776854643484448e-05,
+      "loss": 3.6562,
+      "step": 5396
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.77675695372604e-05,
+      "loss": 5.4688,
+      "step": 5397
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7766592438527005e-05,
+      "loss": 4.75,
+      "step": 5398
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.776561513865537e-05,
+      "loss": 3.5625,
+      "step": 5399
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.776463763765655e-05,
+      "loss": 5.3906,
+      "step": 5400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.776365993554162e-05,
+      "loss": 2.2812,
+      "step": 5401
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.776268203232165e-05,
+      "loss": 5.7812,
+      "step": 5402
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.776170392800771e-05,
+      "loss": 3.2188,
+      "step": 5403
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7760725622610864e-05,
+      "loss": 3.9375,
+      "step": 5404
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.775974711614219e-05,
+      "loss": 5.0469,
+      "step": 5405
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7758768408612774e-05,
+      "loss": 5.75,
+      "step": 5406
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7757789500033686e-05,
+      "loss": 4.1406,
+      "step": 5407
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7756810390416017e-05,
+      "loss": 4.0156,
+      "step": 5408
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7755831079770846e-05,
+      "loss": 4.1875,
+      "step": 5409
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.775485156810926e-05,
+      "loss": 5.8906,
+      "step": 5410
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7753871855442357e-05,
+      "loss": 3.25,
+      "step": 5411
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.775289194178122e-05,
+      "loss": 4.875,
+      "step": 5412
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.775191182713694e-05,
+      "loss": 3.0938,
+      "step": 5413
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.775093151152062e-05,
+      "loss": 5.0625,
+      "step": 5414
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.774995099494336e-05,
+      "loss": 6.2812,
+      "step": 5415
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.774897027741625e-05,
+      "loss": 4.625,
+      "step": 5416
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.77479893589504e-05,
+      "loss": 3.625,
+      "step": 5417
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7747008239556914e-05,
+      "loss": 4.0938,
+      "step": 5418
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.774602691924691e-05,
+      "loss": 4.9688,
+      "step": 5419
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7745045398031477e-05,
+      "loss": 5.3438,
+      "step": 5420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.774406367592174e-05,
+      "loss": 3.3125,
+      "step": 5421
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.774308175292881e-05,
+      "loss": 3.8125,
+      "step": 5422
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.774209962906381e-05,
+      "loss": 4.75,
+      "step": 5423
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7741117304337844e-05,
+      "loss": 3.6562,
+      "step": 5424
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7740134778762045e-05,
+      "loss": 3.7188,
+      "step": 5425
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.773915205234753e-05,
+      "loss": 4.5,
+      "step": 5426
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.773816912510544e-05,
+      "loss": 3.75,
+      "step": 5427
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7737185997046874e-05,
+      "loss": 4.7188,
+      "step": 5428
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.773620266818299e-05,
+      "loss": 5.5312,
+      "step": 5429
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.77352191385249e-05,
+      "loss": 2.8125,
+      "step": 5430
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.773423540808376e-05,
+      "loss": 4.4062,
+      "step": 5431
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.773325147687069e-05,
+      "loss": 3.8125,
+      "step": 5432
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7732267344896835e-05,
+      "loss": 2.875,
+      "step": 5433
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7731283012173337e-05,
+      "loss": 4.9531,
+      "step": 5434
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.773029847871133e-05,
+      "loss": 2.0312,
+      "step": 5435
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.772931374452197e-05,
+      "loss": 3.1562,
+      "step": 5436
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.772832880961642e-05,
+      "loss": 3.875,
+      "step": 5437
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.772734367400579e-05,
+      "loss": 3.2656,
+      "step": 5438
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.772635833770127e-05,
+      "loss": 3.9062,
+      "step": 5439
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.772537280071401e-05,
+      "loss": 2.7969,
+      "step": 5440
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.772438706305515e-05,
+      "loss": 5.0156,
+      "step": 5441
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.772340112473586e-05,
+      "loss": 2.7812,
+      "step": 5442
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.77224149857673e-05,
+      "loss": 4.4375,
+      "step": 5443
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.772142864616064e-05,
+      "loss": 4.4844,
+      "step": 5444
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.772044210592704e-05,
+      "loss": 3.4375,
+      "step": 5445
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.771945536507767e-05,
+      "loss": 4.5,
+      "step": 5446
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.77184684236237e-05,
+      "loss": 4.7031,
+      "step": 5447
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.771748128157632e-05,
+      "loss": 4.25,
+      "step": 5448
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.771649393894667e-05,
+      "loss": 3.6562,
+      "step": 5449
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.771550639574596e-05,
+      "loss": 4.4062,
+      "step": 5450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.771451865198536e-05,
+      "loss": 4.0,
+      "step": 5451
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.771353070767605e-05,
+      "loss": 6.0938,
+      "step": 5452
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.771254256282921e-05,
+      "loss": 3.3438,
+      "step": 5453
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.771155421745604e-05,
+      "loss": 2.9688,
+      "step": 5454
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7710565671567716e-05,
+      "loss": 4.6719,
+      "step": 5455
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.770957692517544e-05,
+      "loss": 4.3438,
+      "step": 5456
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7708587978290404e-05,
+      "loss": 5.1562,
+      "step": 5457
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7707598830923796e-05,
+      "loss": 3.3438,
+      "step": 5458
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.770660948308682e-05,
+      "loss": 6.2188,
+      "step": 5459
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7705619934790676e-05,
+      "loss": 4.75,
+      "step": 5460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.770463018604657e-05,
+      "loss": 4.7812,
+      "step": 5461
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.77036402368657e-05,
+      "loss": 4.25,
+      "step": 5462
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.770265008725928e-05,
+      "loss": 5.5,
+      "step": 5463
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.770165973723852e-05,
+      "loss": 4.875,
+      "step": 5464
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.770066918681462e-05,
+      "loss": 3.7812,
+      "step": 5465
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.769967843599881e-05,
+      "loss": 4.1406,
+      "step": 5466
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.76986874848023e-05,
+      "loss": 4.5469,
+      "step": 5467
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7697696333236306e-05,
+      "loss": 4.2812,
+      "step": 5468
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.769670498131205e-05,
+      "loss": 3.5156,
+      "step": 5469
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7695713429040756e-05,
+      "loss": 4.875,
+      "step": 5470
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7694721676433655e-05,
+      "loss": 3.125,
+      "step": 5471
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.769372972350196e-05,
+      "loss": 2.8438,
+      "step": 5472
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.769273757025692e-05,
+      "loss": 3.4062,
+      "step": 5473
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7691745216709754e-05,
+      "loss": 4.0781,
+      "step": 5474
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7690752662871705e-05,
+      "loss": 2.7812,
+      "step": 5475
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7689759908754e-05,
+      "loss": 3.0625,
+      "step": 5476
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.768876695436789e-05,
+      "loss": 3.7969,
+      "step": 5477
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.768777379972461e-05,
+      "loss": 5.9062,
+      "step": 5478
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.76867804448354e-05,
+      "loss": 5.5312,
+      "step": 5479
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.76857868897115e-05,
+      "loss": 4.4688,
+      "step": 5480
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.768479313436419e-05,
+      "loss": 5.4219,
+      "step": 5481
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.768379917880468e-05,
+      "loss": 6.0156,
+      "step": 5482
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.768280502304425e-05,
+      "loss": 5.6406,
+      "step": 5483
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7681810667094136e-05,
+      "loss": 2.5625,
+      "step": 5484
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7680816110965616e-05,
+      "loss": 5.0625,
+      "step": 5485
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7679821354669936e-05,
+      "loss": 4.6875,
+      "step": 5486
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.767882639821836e-05,
+      "loss": 3.5938,
+      "step": 5487
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7677831241622144e-05,
+      "loss": 3.5,
+      "step": 5488
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7676835884892576e-05,
+      "loss": 3.6875,
+      "step": 5489
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7675840328040904e-05,
+      "loss": 4.8125,
+      "step": 5490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.767484457107841e-05,
+      "loss": 5.375,
+      "step": 5491
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.767384861401636e-05,
+      "loss": 6.0938,
+      "step": 5492
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.767285245686604e-05,
+      "loss": 3.25,
+      "step": 5493
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7671856099638716e-05,
+      "loss": 3.8594,
+      "step": 5494
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.767085954234567e-05,
+      "loss": 4.4062,
+      "step": 5495
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.7669862784998185e-05,
+      "loss": 3.7188,
+      "step": 5496
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.766886582760755e-05,
+      "loss": 4.8281,
+      "step": 5497
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.766786867018505e-05,
+      "loss": 3.125,
+      "step": 5498
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.766687131274197e-05,
+      "loss": 5.125,
+      "step": 5499
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.76658737552896e-05,
+      "loss": 7.625,
+      "step": 5500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.766487599783925e-05,
+      "loss": 5.6719,
+      "step": 5501
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.7663878040402194e-05,
+      "loss": 5.25,
+      "step": 5502
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.7662879882989744e-05,
+      "loss": 2.0156,
+      "step": 5503
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.766188152561319e-05,
+      "loss": 3.875,
+      "step": 5504
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.7660882968283845e-05,
+      "loss": 3.1875,
+      "step": 5505
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.765988421101301e-05,
+      "loss": 3.5938,
+      "step": 5506
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.7658885253811985e-05,
+      "loss": 5.2812,
+      "step": 5507
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.765788609669208e-05,
+      "loss": 6.5938,
+      "step": 5508
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.765688673966463e-05,
+      "loss": 3.25,
+      "step": 5509
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.765588718274091e-05,
+      "loss": 2.8125,
+      "step": 5510
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.7654887425932275e-05,
+      "loss": 4.8594,
+      "step": 5511
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.765388746925001e-05,
+      "loss": 4.5156,
+      "step": 5512
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.765288731270546e-05,
+      "loss": 3.4688,
+      "step": 5513
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.765188695630994e-05,
+      "loss": 3.4375,
+      "step": 5514
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.765088640007477e-05,
+      "loss": 3.0469,
+      "step": 5515
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.7649885644011285e-05,
+      "loss": 5.25,
+      "step": 5516
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.764888468813081e-05,
+      "loss": 4.8906,
+      "step": 5517
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.764788353244467e-05,
+      "loss": 5.6719,
+      "step": 5518
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.7646882176964215e-05,
+      "loss": 4.7188,
+      "step": 5519
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.7645880621700774e-05,
+      "loss": 3.0938,
+      "step": 5520
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.764487886666569e-05,
+      "loss": 4.2188,
+      "step": 5521
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.764387691187029e-05,
+      "loss": 5.3281,
+      "step": 5522
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.764287475732593e-05,
+      "loss": 3.5312,
+      "step": 5523
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.764187240304396e-05,
+      "loss": 4.4531,
+      "step": 5524
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.764086984903571e-05,
+      "loss": 3.7188,
+      "step": 5525
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.763986709531255e-05,
+      "loss": 3.0312,
+      "step": 5526
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.763886414188581e-05,
+      "loss": 3.9688,
+      "step": 5527
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.763786098876686e-05,
+      "loss": 3.125,
+      "step": 5528
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.7636857635967056e-05,
+      "loss": 3.9688,
+      "step": 5529
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.763585408349777e-05,
+      "loss": 3.5625,
+      "step": 5530
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.763485033137033e-05,
+      "loss": 3.3125,
+      "step": 5531
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.7633846379596126e-05,
+      "loss": 5.4844,
+      "step": 5532
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.7632842228186513e-05,
+      "loss": 6.5938,
+      "step": 5533
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.763183787715286e-05,
+      "loss": 2.25,
+      "step": 5534
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.7630833326506545e-05,
+      "loss": 3.125,
+      "step": 5535
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.762982857625893e-05,
+      "loss": 4.7031,
+      "step": 5536
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.76288236264214e-05,
+      "loss": 2.8906,
+      "step": 5537
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.7627818477005325e-05,
+      "loss": 8.4531,
+      "step": 5538
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.762681312802209e-05,
+      "loss": 6.3125,
+      "step": 5539
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.762580757948307e-05,
+      "loss": 2.75,
+      "step": 5540
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.7624801831399655e-05,
+      "loss": 5.8906,
+      "step": 5541
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.7623795883783223e-05,
+      "loss": 8.0938,
+      "step": 5542
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.762278973664517e-05,
+      "loss": 4.5312,
+      "step": 5543
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.762178338999689e-05,
+      "loss": 4.5312,
+      "step": 5544
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.7620776843849775e-05,
+      "loss": 4.6875,
+      "step": 5545
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.7619770098215205e-05,
+      "loss": 5.25,
+      "step": 5546
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.76187631531046e-05,
+      "loss": 8.0938,
+      "step": 5547
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.761775600852934e-05,
+      "loss": 3.9688,
+      "step": 5548
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.761674866450084e-05,
+      "loss": 4.0938,
+      "step": 5549
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.76157411210305e-05,
+      "loss": 4.0938,
+      "step": 5550
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.761473337812972e-05,
+      "loss": 3.875,
+      "step": 5551
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.761372543580992e-05,
+      "loss": 5.0625,
+      "step": 5552
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.76127172940825e-05,
+      "loss": 4.7031,
+      "step": 5553
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.761170895295889e-05,
+      "loss": 3.7188,
+      "step": 5554
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.76107004124505e-05,
+      "loss": 7.8438,
+      "step": 5555
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7609691672568726e-05,
+      "loss": 4.9531,
+      "step": 5556
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.760868273332501e-05,
+      "loss": 4.5156,
+      "step": 5557
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7607673594730776e-05,
+      "loss": 3.3125,
+      "step": 5558
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.760666425679744e-05,
+      "loss": 6.3438,
+      "step": 5559
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.760565471953642e-05,
+      "loss": 4.75,
+      "step": 5560
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.760464498295917e-05,
+      "loss": 5.3125,
+      "step": 5561
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7603635047077105e-05,
+      "loss": 4.5938,
+      "step": 5562
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.760262491190166e-05,
+      "loss": 2.8125,
+      "step": 5563
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.760161457744427e-05,
+      "loss": 6.2812,
+      "step": 5564
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.760060404371638e-05,
+      "loss": 2.25,
+      "step": 5565
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7599593310729414e-05,
+      "loss": 6.1094,
+      "step": 5566
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7598582378494836e-05,
+      "loss": 4.9531,
+      "step": 5567
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7597571247024075e-05,
+      "loss": 4.9375,
+      "step": 5568
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.759655991632859e-05,
+      "loss": 4.1875,
+      "step": 5569
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7595548386419826e-05,
+      "loss": 4.5625,
+      "step": 5570
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.759453665730923e-05,
+      "loss": 5.0312,
+      "step": 5571
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7593524729008255e-05,
+      "loss": 1.625,
+      "step": 5572
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.759251260152837e-05,
+      "loss": 4.6562,
+      "step": 5573
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.759150027488102e-05,
+      "loss": 4.2812,
+      "step": 5574
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.759048774907767e-05,
+      "loss": 3.5,
+      "step": 5575
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.758947502412978e-05,
+      "loss": 2.0312,
+      "step": 5576
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7588462100048825e-05,
+      "loss": 4.7969,
+      "step": 5577
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7587448976846266e-05,
+      "loss": 4.7031,
+      "step": 5578
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7586435654533575e-05,
+      "loss": 3.625,
+      "step": 5579
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.758542213312222e-05,
+      "loss": 3.3125,
+      "step": 5580
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.758440841262367e-05,
+      "loss": 5.875,
+      "step": 5581
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.758339449304942e-05,
+      "loss": 4.0625,
+      "step": 5582
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.758238037441093e-05,
+      "loss": 4.7031,
+      "step": 5583
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.758136605671969e-05,
+      "loss": 4.7656,
+      "step": 5584
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7580351539987186e-05,
+      "loss": 3.375,
+      "step": 5585
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.75793368242249e-05,
+      "loss": 3.8438,
+      "step": 5586
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.757832190944431e-05,
+      "loss": 3.1875,
+      "step": 5587
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.7577306795656916e-05,
+      "loss": 5.2656,
+      "step": 5588
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.7576291482874215e-05,
+      "loss": 4.8281,
+      "step": 5589
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.757527597110769e-05,
+      "loss": 2.7344,
+      "step": 5590
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.757426026036885e-05,
+      "loss": 4.5938,
+      "step": 5591
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.757324435066918e-05,
+      "loss": 4.2812,
+      "step": 5592
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.757222824202019e-05,
+      "loss": 10.625,
+      "step": 5593
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.7571211934433385e-05,
+      "loss": 2.9844,
+      "step": 5594
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.7570195427920274e-05,
+      "loss": 1.8438,
+      "step": 5595
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.756917872249236e-05,
+      "loss": 2.1562,
+      "step": 5596
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.7568161818161135e-05,
+      "loss": 6.1562,
+      "step": 5597
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.756714471493814e-05,
+      "loss": 2.9375,
+      "step": 5598
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.756612741283488e-05,
+      "loss": 3.6562,
+      "step": 5599
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.7565109911862865e-05,
+      "loss": 2.2812,
+      "step": 5600
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.7564092212033624e-05,
+      "loss": 3.4375,
+      "step": 5601
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.7563074313358675e-05,
+      "loss": 4.7812,
+      "step": 5602
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.756205621584954e-05,
+      "loss": 3.2344,
+      "step": 5603
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.756103791951774e-05,
+      "loss": 3.75,
+      "step": 5604
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.7560019424374814e-05,
+      "loss": 3.1875,
+      "step": 5605
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.755900073043229e-05,
+      "loss": 3.25,
+      "step": 5606
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.7557981837701704e-05,
+      "loss": 4.4688,
+      "step": 5607
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.755696274619458e-05,
+      "loss": 2.75,
+      "step": 5608
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.755594345592246e-05,
+      "loss": 4.0,
+      "step": 5609
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.755492396689688e-05,
+      "loss": 3.1562,
+      "step": 5610
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.755390427912939e-05,
+      "loss": 3.875,
+      "step": 5611
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.755288439263154e-05,
+      "loss": 6.2969,
+      "step": 5612
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.755186430741486e-05,
+      "loss": 3.2969,
+      "step": 5613
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.755084402349091e-05,
+      "loss": 2.5625,
+      "step": 5614
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.754982354087123e-05,
+      "loss": 5.0312,
+      "step": 5615
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.7548802859567376e-05,
+      "loss": 3.3438,
+      "step": 5616
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.7547781979590915e-05,
+      "loss": 4.3594,
+      "step": 5617
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.7546760900953386e-05,
+      "loss": 4.7031,
+      "step": 5618
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.754573962366636e-05,
+      "loss": 6.375,
+      "step": 5619
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.754471814774141e-05,
+      "loss": 4.75,
+      "step": 5620
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.7543696473190084e-05,
+      "loss": 2.9688,
+      "step": 5621
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.754267460002394e-05,
+      "loss": 5.7812,
+      "step": 5622
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.754165252825457e-05,
+      "loss": 5.7344,
+      "step": 5623
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.754063025789353e-05,
+      "loss": 3.8281,
+      "step": 5624
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.75396077889524e-05,
+      "loss": 5.0156,
+      "step": 5625
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.753858512144276e-05,
+      "loss": 3.4531,
+      "step": 5626
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.753756225537617e-05,
+      "loss": 6.6406,
+      "step": 5627
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.753653919076423e-05,
+      "loss": 4.0,
+      "step": 5628
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.753551592761851e-05,
+      "loss": 1.8438,
+      "step": 5629
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.753449246595059e-05,
+      "loss": 3.7344,
+      "step": 5630
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.7533468805772077e-05,
+      "loss": 5.5625,
+      "step": 5631
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.753244494709454e-05,
+      "loss": 6.0,
+      "step": 5632
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.753142088992958e-05,
+      "loss": 4.875,
+      "step": 5633
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.753039663428879e-05,
+      "loss": 3.75,
+      "step": 5634
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.7529372180183766e-05,
+      "loss": 3.125,
+      "step": 5635
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.752834752762609e-05,
+      "loss": 3.0312,
+      "step": 5636
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.75273226766274e-05,
+      "loss": 5.375,
+      "step": 5637
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.752629762719925e-05,
+      "loss": 3.0625,
+      "step": 5638
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.752527237935328e-05,
+      "loss": 1.5625,
+      "step": 5639
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.752424693310109e-05,
+      "loss": 2.4062,
+      "step": 5640
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.752322128845428e-05,
+      "loss": 7.5312,
+      "step": 5641
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.752219544542448e-05,
+      "loss": 3.2031,
+      "step": 5642
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.752116940402328e-05,
+      "loss": 2.875,
+      "step": 5643
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.7520143164262305e-05,
+      "loss": 4.4062,
+      "step": 5644
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.7519116726153175e-05,
+      "loss": 2.2188,
+      "step": 5645
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.751809008970752e-05,
+      "loss": 5.6562,
+      "step": 5646
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7517063254936947e-05,
+      "loss": 3.4375,
+      "step": 5647
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7516036221853084e-05,
+      "loss": 3.4375,
+      "step": 5648
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7515008990467566e-05,
+      "loss": 2.8125,
+      "step": 5649
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.751398156079202e-05,
+      "loss": 10.6094,
+      "step": 5650
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.751295393283807e-05,
+      "loss": 4.1875,
+      "step": 5651
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.751192610661736e-05,
+      "loss": 4.1406,
+      "step": 5652
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7510898082141516e-05,
+      "loss": 5.0938,
+      "step": 5653
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.750986985942219e-05,
+      "loss": 4.1875,
+      "step": 5654
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.750884143847101e-05,
+      "loss": 6.75,
+      "step": 5655
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7507812819299625e-05,
+      "loss": 5.0312,
+      "step": 5656
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7506784001919676e-05,
+      "loss": 8.375,
+      "step": 5657
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.750575498634281e-05,
+      "loss": 2.75,
+      "step": 5658
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.750472577258069e-05,
+      "loss": 3.2969,
+      "step": 5659
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.750369636064495e-05,
+      "loss": 5.7188,
+      "step": 5660
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.750266675054726e-05,
+      "loss": 3.6875,
+      "step": 5661
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7501636942299257e-05,
+      "loss": 4.8281,
+      "step": 5662
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7500606935912624e-05,
+      "loss": 4.5,
+      "step": 5663
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7499576731399e-05,
+      "loss": 4.4062,
+      "step": 5664
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.749854632877005e-05,
+      "loss": 5.4375,
+      "step": 5665
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.749751572803746e-05,
+      "loss": 3.1562,
+      "step": 5666
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7496484929212875e-05,
+      "loss": 3.0625,
+      "step": 5667
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.749545393230798e-05,
+      "loss": 1.8125,
+      "step": 5668
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7494422737334433e-05,
+      "loss": 2.625,
+      "step": 5669
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.749339134430392e-05,
+      "loss": 6.0938,
+      "step": 5670
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7492359753228116e-05,
+      "loss": 2.0625,
+      "step": 5671
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7491327964118696e-05,
+      "loss": 3.5312,
+      "step": 5672
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7490295976987346e-05,
+      "loss": 4.5938,
+      "step": 5673
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.748926379184574e-05,
+      "loss": 3.8281,
+      "step": 5674
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.748823140870557e-05,
+      "loss": 2.0312,
+      "step": 5675
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.748719882757852e-05,
+      "loss": 1.8438,
+      "step": 5676
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7486166048476286e-05,
+      "loss": 4.4375,
+      "step": 5677
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.748513307141056e-05,
+      "loss": 4.9375,
+      "step": 5678
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7484099896393026e-05,
+      "loss": 3.8906,
+      "step": 5679
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.74830665234354e-05,
+      "loss": 3.6875,
+      "step": 5680
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.748203295254936e-05,
+      "loss": 2.6406,
+      "step": 5681
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.748099918374662e-05,
+      "loss": 4.0625,
+      "step": 5682
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7479965217038875e-05,
+      "loss": 4.5,
+      "step": 5683
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.747893105243784e-05,
+      "loss": 4.7812,
+      "step": 5684
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.747789668995522e-05,
+      "loss": 2.0938,
+      "step": 5685
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.747686212960272e-05,
+      "loss": 3.2656,
+      "step": 5686
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.747582737139206e-05,
+      "loss": 4.7031,
+      "step": 5687
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.747479241533495e-05,
+      "loss": 4.0781,
+      "step": 5688
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7473757261443105e-05,
+      "loss": 3.7812,
+      "step": 5689
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.747272190972825e-05,
+      "loss": 4.5469,
+      "step": 5690
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.74716863602021e-05,
+      "loss": 4.1875,
+      "step": 5691
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7470650612876385e-05,
+      "loss": 2.6562,
+      "step": 5692
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7469614667762825e-05,
+      "loss": 2.9844,
+      "step": 5693
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.746857852487315e-05,
+      "loss": 2.3125,
+      "step": 5694
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.74675421842191e-05,
+      "loss": 3.9219,
+      "step": 5695
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.746650564581239e-05,
+      "loss": 3.0625,
+      "step": 5696
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.746546890966477e-05,
+      "loss": 3.1719,
+      "step": 5697
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.746443197578796e-05,
+      "loss": 6.5469,
+      "step": 5698
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7463394844193715e-05,
+      "loss": 3.7188,
+      "step": 5699
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.746235751489377e-05,
+      "loss": 5.3594,
+      "step": 5700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.746131998789987e-05,
+      "loss": 5.6875,
+      "step": 5701
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7460282263223764e-05,
+      "loss": 2.7812,
+      "step": 5702
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.74592443408772e-05,
+      "loss": 4.625,
+      "step": 5703
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.745820622087192e-05,
+      "loss": 8.0312,
+      "step": 5704
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7457167903219686e-05,
+      "loss": 4.375,
+      "step": 5705
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.745612938793225e-05,
+      "loss": 2.0625,
+      "step": 5706
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.745509067502137e-05,
+      "loss": 3.5,
+      "step": 5707
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.74540517644988e-05,
+      "loss": 2.9062,
+      "step": 5708
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.74530126563763e-05,
+      "loss": 3.625,
+      "step": 5709
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.745197335066565e-05,
+      "loss": 4.875,
+      "step": 5710
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7450933847378606e-05,
+      "loss": 3.0781,
+      "step": 5711
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7449894146526934e-05,
+      "loss": 4.5156,
+      "step": 5712
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.74488542481224e-05,
+      "loss": 4.6875,
+      "step": 5713
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.744781415217679e-05,
+      "loss": 4.2344,
+      "step": 5714
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.744677385870187e-05,
+      "loss": 4.7969,
+      "step": 5715
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.744573336770942e-05,
+      "loss": 3.4062,
+      "step": 5716
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7444692679211227e-05,
+      "loss": 3.6875,
+      "step": 5717
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7443651793219055e-05,
+      "loss": 1.75,
+      "step": 5718
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.74426107097447e-05,
+      "loss": 6.0625,
+      "step": 5719
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.744156942879995e-05,
+      "loss": 2.4219,
+      "step": 5720
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.744052795039659e-05,
+      "loss": 3.875,
+      "step": 5721
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.743948627454641e-05,
+      "loss": 5.0,
+      "step": 5722
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.743844440126121e-05,
+      "loss": 4.1562,
+      "step": 5723
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.743740233055277e-05,
+      "loss": 6.0625,
+      "step": 5724
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7436360062432905e-05,
+      "loss": 5.6719,
+      "step": 5725
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.743531759691339e-05,
+      "loss": 6.9375,
+      "step": 5726
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.743427493400606e-05,
+      "loss": 3.3125,
+      "step": 5727
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.74332320737227e-05,
+      "loss": 4.1562,
+      "step": 5728
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.743218901607511e-05,
+      "loss": 3.125,
+      "step": 5729
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.743114576107511e-05,
+      "loss": 6.0625,
+      "step": 5730
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7430102308734515e-05,
+      "loss": 4.1875,
+      "step": 5731
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7429058659065124e-05,
+      "loss": 2.9375,
+      "step": 5732
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.742801481207876e-05,
+      "loss": 2.6562,
+      "step": 5733
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.742697076778724e-05,
+      "loss": 4.4688,
+      "step": 5734
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.742592652620238e-05,
+      "loss": 2.3438,
+      "step": 5735
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7424882087336014e-05,
+      "loss": 5.8125,
+      "step": 5736
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.742383745119996e-05,
+      "loss": 4.0781,
+      "step": 5737
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.742279261780603e-05,
+      "loss": 5.4531,
+      "step": 5738
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7421747587166075e-05,
+      "loss": 2.875,
+      "step": 5739
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.742070235929192e-05,
+      "loss": 5.375,
+      "step": 5740
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7419656934195385e-05,
+      "loss": 2.5625,
+      "step": 5741
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.741861131188832e-05,
+      "loss": 4.25,
+      "step": 5742
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7417565492382546e-05,
+      "loss": 3.6406,
+      "step": 5743
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.741651947568993e-05,
+      "loss": 2.8281,
+      "step": 5744
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.741547326182229e-05,
+      "loss": 6.2031,
+      "step": 5745
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.741442685079148e-05,
+      "loss": 4.9062,
+      "step": 5746
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.741338024260935e-05,
+      "loss": 5.9844,
+      "step": 5747
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.741233343728774e-05,
+      "loss": 3.0,
+      "step": 5748
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7411286434838506e-05,
+      "loss": 7.0625,
+      "step": 5749
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7410239235273494e-05,
+      "loss": 4.7188,
+      "step": 5750
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.740919183860457e-05,
+      "loss": 5.7344,
+      "step": 5751
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.740814424484359e-05,
+      "loss": 4.8438,
+      "step": 5752
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.740709645400241e-05,
+      "loss": 3.375,
+      "step": 5753
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.74060484660929e-05,
+      "loss": 7.4844,
+      "step": 5754
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.740500028112691e-05,
+      "loss": 4.9531,
+      "step": 5755
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7403951899116314e-05,
+      "loss": 4.375,
+      "step": 5756
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.740290332007299e-05,
+      "loss": 5.7812,
+      "step": 5757
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7401854544008793e-05,
+      "loss": 4.2188,
+      "step": 5758
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.740080557093561e-05,
+      "loss": 2.9062,
+      "step": 5759
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.739975640086531e-05,
+      "loss": 3.875,
+      "step": 5760
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7398707033809765e-05,
+      "loss": 4.8594,
+      "step": 5761
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.739765746978087e-05,
+      "loss": 6.7344,
+      "step": 5762
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7396607708790495e-05,
+      "loss": 3.5938,
+      "step": 5763
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.739555775085053e-05,
+      "loss": 3.75,
+      "step": 5764
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.739450759597286e-05,
+      "loss": 2.3438,
+      "step": 5765
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.739345724416937e-05,
+      "loss": 5.1562,
+      "step": 5766
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7392406695451964e-05,
+      "loss": 2.625,
+      "step": 5767
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.739135594983252e-05,
+      "loss": 3.2969,
+      "step": 5768
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.739030500732294e-05,
+      "loss": 2.875,
+      "step": 5769
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.738925386793512e-05,
+      "loss": 3.2812,
+      "step": 5770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.738820253168097e-05,
+      "loss": 4.9062,
+      "step": 5771
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.738715099857238e-05,
+      "loss": 9.2188,
+      "step": 5772
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.738609926862126e-05,
+      "loss": 3.7969,
+      "step": 5773
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.738504734183952e-05,
+      "loss": 3.125,
+      "step": 5774
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.738399521823906e-05,
+      "loss": 3.9375,
+      "step": 5775
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.7382942897831796e-05,
+      "loss": 5.0938,
+      "step": 5776
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.738189038062965e-05,
+      "loss": 3.4375,
+      "step": 5777
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.738083766664452e-05,
+      "loss": 4.4062,
+      "step": 5778
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.737978475588834e-05,
+      "loss": 3.8438,
+      "step": 5779
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.737873164837302e-05,
+      "loss": 5.4219,
+      "step": 5780
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.737767834411049e-05,
+      "loss": 2.5938,
+      "step": 5781
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.7376624843112665e-05,
+      "loss": 6.625,
+      "step": 5782
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.737557114539148e-05,
+      "loss": 3.5,
+      "step": 5783
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.737451725095886e-05,
+      "loss": 6.1562,
+      "step": 5784
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.737346315982674e-05,
+      "loss": 5.4062,
+      "step": 5785
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.737240887200705e-05,
+      "loss": 6.4688,
+      "step": 5786
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.737135438751173e-05,
+      "loss": 9.6875,
+      "step": 5787
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.737029970635271e-05,
+      "loss": 4.9062,
+      "step": 5788
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.7369244828541935e-05,
+      "loss": 4.0,
+      "step": 5789
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.736818975409136e-05,
+      "loss": 5.6875,
+      "step": 5790
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.7367134483012904e-05,
+      "loss": 7.2344,
+      "step": 5791
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.7366079015318534e-05,
+      "loss": 5.3438,
+      "step": 5792
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.736502335102018e-05,
+      "loss": 2.1094,
+      "step": 5793
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.736396749012982e-05,
+      "loss": 4.6875,
+      "step": 5794
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.736291143265939e-05,
+      "loss": 2.7812,
+      "step": 5795
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.736185517862085e-05,
+      "loss": 2.2812,
+      "step": 5796
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.736079872802615e-05,
+      "loss": 4.2812,
+      "step": 5797
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.7359742080887265e-05,
+      "loss": 7.1094,
+      "step": 5798
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.735868523721615e-05,
+      "loss": 3.875,
+      "step": 5799
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.735762819702476e-05,
+      "loss": 3.4688,
+      "step": 5800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.735657096032508e-05,
+      "loss": 5.9688,
+      "step": 5801
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.735551352712907e-05,
+      "loss": 5.0469,
+      "step": 5802
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.7354455897448694e-05,
+      "loss": 3.0625,
+      "step": 5803
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.7353398071295944e-05,
+      "loss": 3.5625,
+      "step": 5804
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.735234004868278e-05,
+      "loss": 2.5,
+      "step": 5805
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.735128182962118e-05,
+      "loss": 4.9688,
+      "step": 5806
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.735022341412314e-05,
+      "loss": 4.8125,
+      "step": 5807
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.734916480220063e-05,
+      "loss": 5.7656,
+      "step": 5808
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.734810599386563e-05,
+      "loss": 4.1875,
+      "step": 5809
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.7347046989130137e-05,
+      "loss": 5.875,
+      "step": 5810
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.734598778800614e-05,
+      "loss": 2.8438,
+      "step": 5811
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.734492839050562e-05,
+      "loss": 6.625,
+      "step": 5812
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.7343868796640585e-05,
+      "loss": 6.2188,
+      "step": 5813
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.7342809006423026e-05,
+      "loss": 5.5938,
+      "step": 5814
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.734174901986493e-05,
+      "loss": 4.8984,
+      "step": 5815
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.734068883697831e-05,
+      "loss": 6.3438,
+      "step": 5816
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.7339628457775156e-05,
+      "loss": 4.9531,
+      "step": 5817
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.73385678822675e-05,
+      "loss": 8.8125,
+      "step": 5818
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.733750711046731e-05,
+      "loss": 4.7656,
+      "step": 5819
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.733644614238663e-05,
+      "loss": 3.6875,
+      "step": 5820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.7335384978037446e-05,
+      "loss": 7.2344,
+      "step": 5821
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.733432361743179e-05,
+      "loss": 2.9062,
+      "step": 5822
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.7333262060581667e-05,
+      "loss": 5.9688,
+      "step": 5823
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.733220030749909e-05,
+      "loss": 6.2656,
+      "step": 5824
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.7331138358196094e-05,
+      "loss": 6.3438,
+      "step": 5825
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.73300762126847e-05,
+      "loss": 3.9062,
+      "step": 5826
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.732901387097692e-05,
+      "loss": 2.9531,
+      "step": 5827
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.7327951333084795e-05,
+      "loss": 4.5781,
+      "step": 5828
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.732688859902034e-05,
+      "loss": 3.375,
+      "step": 5829
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.73258256687956e-05,
+      "loss": 5.4219,
+      "step": 5830
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.73247625424226e-05,
+      "loss": 3.8438,
+      "step": 5831
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.732369921991338e-05,
+      "loss": 6.75,
+      "step": 5832
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.732263570127997e-05,
+      "loss": 2.3125,
+      "step": 5833
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.732157198653442e-05,
+      "loss": 4.3438,
+      "step": 5834
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.732050807568878e-05,
+      "loss": 2.8438,
+      "step": 5835
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.7319443968755076e-05,
+      "loss": 3.625,
+      "step": 5836
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.731837966574536e-05,
+      "loss": 6.0156,
+      "step": 5837
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.731731516667168e-05,
+      "loss": 5.2344,
+      "step": 5838
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.73162504715461e-05,
+      "loss": 6.7812,
+      "step": 5839
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.731518558038066e-05,
+      "loss": 4.7188,
+      "step": 5840
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.731412049318742e-05,
+      "loss": 3.8438,
+      "step": 5841
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.7313055209978435e-05,
+      "loss": 4.125,
+      "step": 5842
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.731198973076578e-05,
+      "loss": 3.25,
+      "step": 5843
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.731092405556149e-05,
+      "loss": 3.6562,
+      "step": 5844
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.7309858184377665e-05,
+      "loss": 3.2812,
+      "step": 5845
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.730879211722634e-05,
+      "loss": 3.2656,
+      "step": 5846
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.73077258541196e-05,
+      "loss": 5.2969,
+      "step": 5847
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.730665939506951e-05,
+      "loss": 2.9062,
+      "step": 5848
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.7305592740088155e-05,
+      "loss": 6.0312,
+      "step": 5849
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.73045258891876e-05,
+      "loss": 4.0312,
+      "step": 5850
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.730345884237992e-05,
+      "loss": 3.2344,
+      "step": 5851
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.73023915996772e-05,
+      "loss": 5.4375,
+      "step": 5852
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.730132416109153e-05,
+      "loss": 3.5781,
+      "step": 5853
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.730025652663499e-05,
+      "loss": 6.8125,
+      "step": 5854
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.7299188696319654e-05,
+      "loss": 4.7812,
+      "step": 5855
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.729812067015763e-05,
+      "loss": 2.5938,
+      "step": 5856
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.729705244816099e-05,
+      "loss": 4.4062,
+      "step": 5857
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.729598403034185e-05,
+      "loss": 3.5,
+      "step": 5858
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.7294915416712294e-05,
+      "loss": 4.625,
+      "step": 5859
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7293846607284416e-05,
+      "loss": 3.5,
+      "step": 5860
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.729277760207032e-05,
+      "loss": 3.7812,
+      "step": 5861
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7291708401082104e-05,
+      "loss": 2.625,
+      "step": 5862
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.729063900433189e-05,
+      "loss": 4.2656,
+      "step": 5863
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.728956941183176e-05,
+      "loss": 3.0156,
+      "step": 5864
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.728849962359384e-05,
+      "loss": 4.5938,
+      "step": 5865
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.728742963963024e-05,
+      "loss": 5.2344,
+      "step": 5866
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7286359459953063e-05,
+      "loss": 4.8906,
+      "step": 5867
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7285289084574434e-05,
+      "loss": 3.1719,
+      "step": 5868
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.728421851350647e-05,
+      "loss": 3.1562,
+      "step": 5869
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.728314774676128e-05,
+      "loss": 4.6875,
+      "step": 5870
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7282076784351006e-05,
+      "loss": 4.4062,
+      "step": 5871
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.728100562628776e-05,
+      "loss": 3.0781,
+      "step": 5872
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7279934272583674e-05,
+      "loss": 2.8906,
+      "step": 5873
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.727886272325087e-05,
+      "loss": 3.3281,
+      "step": 5874
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7277790978301475e-05,
+      "loss": 5.8594,
+      "step": 5875
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7276719037747644e-05,
+      "loss": 3.3906,
+      "step": 5876
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.727564690160149e-05,
+      "loss": 4.7188,
+      "step": 5877
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.727457456987516e-05,
+      "loss": 3.9375,
+      "step": 5878
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.72735020425808e-05,
+      "loss": 3.0469,
+      "step": 5879
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.727242931973055e-05,
+      "loss": 3.5469,
+      "step": 5880
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.727135640133653e-05,
+      "loss": 5.0,
+      "step": 5881
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.727028328741093e-05,
+      "loss": 2.9688,
+      "step": 5882
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.726920997796587e-05,
+      "loss": 2.0625,
+      "step": 5883
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7268136473013496e-05,
+      "loss": 2.3438,
+      "step": 5884
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.726706277256598e-05,
+      "loss": 6.3906,
+      "step": 5885
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7265988876635474e-05,
+      "loss": 2.875,
+      "step": 5886
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.726491478523412e-05,
+      "loss": 2.9688,
+      "step": 5887
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.7263840498374105e-05,
+      "loss": 6.4688,
+      "step": 5888
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.726276601606757e-05,
+      "loss": 3.4062,
+      "step": 5889
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.726169133832668e-05,
+      "loss": 3.2344,
+      "step": 5890
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.726061646516361e-05,
+      "loss": 2.5469,
+      "step": 5891
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.725954139659053e-05,
+      "loss": 6.5469,
+      "step": 5892
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7258466132619606e-05,
+      "loss": 5.7031,
+      "step": 5893
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7257390673263006e-05,
+      "loss": 4.1094,
+      "step": 5894
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.725631501853292e-05,
+      "loss": 6.1094,
+      "step": 5895
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.725523916844151e-05,
+      "loss": 4.7031,
+      "step": 5896
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.725416312300096e-05,
+      "loss": 5.9062,
+      "step": 5897
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.725308688222346e-05,
+      "loss": 4.125,
+      "step": 5898
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.725201044612119e-05,
+      "loss": 4.625,
+      "step": 5899
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.725093381470633e-05,
+      "loss": 3.375,
+      "step": 5900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.724985698799108e-05,
+      "loss": 2.0938,
+      "step": 5901
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.724877996598762e-05,
+      "loss": 3.3125,
+      "step": 5902
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7247702748708154e-05,
+      "loss": 5.2188,
+      "step": 5903
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7246625336164865e-05,
+      "loss": 3.625,
+      "step": 5904
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.724554772836997e-05,
+      "loss": 3.4688,
+      "step": 5905
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.724446992533564e-05,
+      "loss": 3.8125,
+      "step": 5906
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7243391927074103e-05,
+      "loss": 3.625,
+      "step": 5907
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.724231373359755e-05,
+      "loss": 3.8125,
+      "step": 5908
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7241235344918193e-05,
+      "loss": 3.6875,
+      "step": 5909
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.724015676104824e-05,
+      "loss": 2.875,
+      "step": 5910
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.723907798199989e-05,
+      "loss": 3.4531,
+      "step": 5911
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.723799900778538e-05,
+      "loss": 3.9375,
+      "step": 5912
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7236919838416895e-05,
+      "loss": 5.1406,
+      "step": 5913
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7235840473906683e-05,
+      "loss": 4.2188,
+      "step": 5914
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.723476091426694e-05,
+      "loss": 2.0938,
+      "step": 5915
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.72336811595099e-05,
+      "loss": 5.25,
+      "step": 5916
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7232601209647786e-05,
+      "loss": 3.9375,
+      "step": 5917
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.723152106469282e-05,
+      "loss": 3.2656,
+      "step": 5918
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7230440724657243e-05,
+      "loss": 4.7812,
+      "step": 5919
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.7229360189553265e-05,
+      "loss": 3.5,
+      "step": 5920
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.722827945939314e-05,
+      "loss": 1.8125,
+      "step": 5921
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7227198534189076e-05,
+      "loss": 3.75,
+      "step": 5922
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7226117413953335e-05,
+      "loss": 2.6562,
+      "step": 5923
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.722503609869815e-05,
+      "loss": 2.6562,
+      "step": 5924
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.722395458843576e-05,
+      "loss": 2.9531,
+      "step": 5925
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7222872883178414e-05,
+      "loss": 2.8125,
+      "step": 5926
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.722179098293835e-05,
+      "loss": 4.0469,
+      "step": 5927
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.722070888772782e-05,
+      "loss": 6.1094,
+      "step": 5928
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.721962659755908e-05,
+      "loss": 3.0781,
+      "step": 5929
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.721854411244437e-05,
+      "loss": 4.6562,
+      "step": 5930
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.721746143239596e-05,
+      "loss": 6.0312,
+      "step": 5931
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.72163785574261e-05,
+      "loss": 3.8281,
+      "step": 5932
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.721529548754705e-05,
+      "loss": 4.7656,
+      "step": 5933
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.721421222277107e-05,
+      "loss": 2.4375,
+      "step": 5934
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.721312876311042e-05,
+      "loss": 6.7656,
+      "step": 5935
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.721204510857738e-05,
+      "loss": 4.9062,
+      "step": 5936
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.72109612591842e-05,
+      "loss": 4.3438,
+      "step": 5937
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7209877214943163e-05,
+      "loss": 5.2188,
+      "step": 5938
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.720879297586654e-05,
+      "loss": 9.0312,
+      "step": 5939
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7207708541966606e-05,
+      "loss": 4.9375,
+      "step": 5940
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.720662391325563e-05,
+      "loss": 5.5625,
+      "step": 5941
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.72055390897459e-05,
+      "loss": 4.125,
+      "step": 5942
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7204454071449696e-05,
+      "loss": 6.9844,
+      "step": 5943
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.72033688583793e-05,
+      "loss": 4.1562,
+      "step": 5944
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7202283450547004e-05,
+      "loss": 5.0625,
+      "step": 5945
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.720119784796508e-05,
+      "loss": 4.2188,
+      "step": 5946
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.720011205064583e-05,
+      "loss": 5.375,
+      "step": 5947
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.719902605860155e-05,
+      "loss": 7.0312,
+      "step": 5948
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7197939871844526e-05,
+      "loss": 5.8125,
+      "step": 5949
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.719685349038707e-05,
+      "loss": 1.9688,
+      "step": 5950
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7195766914241455e-05,
+      "loss": 4.5938,
+      "step": 5951
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.719468014342001e-05,
+      "loss": 3.625,
+      "step": 5952
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.719359317793501e-05,
+      "loss": 3.4062,
+      "step": 5953
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.719250601779879e-05,
+      "loss": 3.7031,
+      "step": 5954
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.719141866302363e-05,
+      "loss": 3.375,
+      "step": 5955
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.719033111362187e-05,
+      "loss": 2.1406,
+      "step": 5956
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.71892433696058e-05,
+      "loss": 2.7812,
+      "step": 5957
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.718815543098774e-05,
+      "loss": 2.3594,
+      "step": 5958
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.718706729778001e-05,
+      "loss": 3.5625,
+      "step": 5959
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7185978969994924e-05,
+      "loss": 4.7969,
+      "step": 5960
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7184890447644806e-05,
+      "loss": 6.25,
+      "step": 5961
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.718380173074198e-05,
+      "loss": 7.9062,
+      "step": 5962
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.718271281929877e-05,
+      "loss": 1.875,
+      "step": 5963
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.718162371332751e-05,
+      "loss": 2.8438,
+      "step": 5964
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.718053441284052e-05,
+      "loss": 2.7188,
+      "step": 5965
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.717944491785013e-05,
+      "loss": 2.9688,
+      "step": 5966
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7178355228368695e-05,
+      "loss": 3.0938,
+      "step": 5967
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.717726534440853e-05,
+      "loss": 3.4062,
+      "step": 5968
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7176175265981976e-05,
+      "loss": 2.9375,
+      "step": 5969
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.717508499310138e-05,
+      "loss": 2.7812,
+      "step": 5970
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.717399452577909e-05,
+      "loss": 3.8125,
+      "step": 5971
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.717290386402744e-05,
+      "loss": 3.2188,
+      "step": 5972
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.717181300785879e-05,
+      "loss": 4.8125,
+      "step": 5973
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.717072195728548e-05,
+      "loss": 2.8438,
+      "step": 5974
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7169630712319864e-05,
+      "loss": 4.2188,
+      "step": 5975
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.71685392729743e-05,
+      "loss": 3.1406,
+      "step": 5976
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.716744763926113e-05,
+      "loss": 5.0625,
+      "step": 5977
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7166355811192734e-05,
+      "loss": 1.5625,
+      "step": 5978
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.716526378878146e-05,
+      "loss": 6.125,
+      "step": 5979
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.7164171572039676e-05,
+      "loss": 3.1094,
+      "step": 5980
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.716307916097974e-05,
+      "loss": 3.9844,
+      "step": 5981
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.716198655561403e-05,
+      "loss": 4.9375,
+      "step": 5982
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.71608937559549e-05,
+      "loss": 6.3438,
+      "step": 5983
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.715980076201474e-05,
+      "loss": 5.9062,
+      "step": 5984
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7158707573805915e-05,
+      "loss": 4.2031,
+      "step": 5985
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7157614191340795e-05,
+      "loss": 3.7188,
+      "step": 5986
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7156520614631776e-05,
+      "loss": 2.6562,
+      "step": 5987
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.715542684369122e-05,
+      "loss": 3.2188,
+      "step": 5988
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.715433287853151e-05,
+      "loss": 3.375,
+      "step": 5989
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.715323871916505e-05,
+      "loss": 3.7969,
+      "step": 5990
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.715214436560421e-05,
+      "loss": 3.5312,
+      "step": 5991
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.715104981786138e-05,
+      "loss": 4.6094,
+      "step": 5992
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7149955075948964e-05,
+      "loss": 5.1406,
+      "step": 5993
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.714886013987934e-05,
+      "loss": 5.8438,
+      "step": 5994
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7147765009664925e-05,
+      "loss": 3.5625,
+      "step": 5995
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.714666968531809e-05,
+      "loss": 2.7812,
+      "step": 5996
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7145574166851255e-05,
+      "loss": 4.2188,
+      "step": 5997
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.714447845427681e-05,
+      "loss": 5.375,
+      "step": 5998
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.714338254760717e-05,
+      "loss": 3.6562,
+      "step": 5999
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7142286446854745e-05,
+      "loss": 3.6875,
+      "step": 6000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.714119015203193e-05,
+      "loss": 2.5,
+      "step": 6001
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7140093663151156e-05,
+      "loss": 4.5625,
+      "step": 6002
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7138996980224806e-05,
+      "loss": 3.125,
+      "step": 6003
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7137900103265326e-05,
+      "loss": 2.9219,
+      "step": 6004
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.713680303228512e-05,
+      "loss": 4.5938,
+      "step": 6005
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7135705767296604e-05,
+      "loss": 3.1562,
+      "step": 6006
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.713460830831222e-05,
+      "loss": 4.25,
+      "step": 6007
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7133510655344363e-05,
+      "loss": 5.9062,
+      "step": 6008
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7132412808405483e-05,
+      "loss": 3.0,
+      "step": 6009
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.7131314767508e-05,
+      "loss": 3.4688,
+      "step": 6010
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.713021653266434e-05,
+      "loss": 5.0156,
+      "step": 6011
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.712911810388695e-05,
+      "loss": 2.6562,
+      "step": 6012
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.712801948118826e-05,
+      "loss": 3.0312,
+      "step": 6013
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.712692066458071e-05,
+      "loss": 3.5625,
+      "step": 6014
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.712582165407672e-05,
+      "loss": 7.2656,
+      "step": 6015
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.712472244968876e-05,
+      "loss": 3.4844,
+      "step": 6016
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.712362305142926e-05,
+      "loss": 3.2812,
+      "step": 6017
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.712252345931067e-05,
+      "loss": 3.2969,
+      "step": 6018
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.712142367334543e-05,
+      "loss": 2.7812,
+      "step": 6019
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7120323693546004e-05,
+      "loss": 4.7188,
+      "step": 6020
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.711922351992484e-05,
+      "loss": 3.7969,
+      "step": 6021
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7118123152494387e-05,
+      "loss": 1.6562,
+      "step": 6022
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.711702259126711e-05,
+      "loss": 3.9062,
+      "step": 6023
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.711592183625547e-05,
+      "loss": 3.9688,
+      "step": 6024
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.711482088747192e-05,
+      "loss": 4.4375,
+      "step": 6025
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7113719744928935e-05,
+      "loss": 3.75,
+      "step": 6026
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.711261840863897e-05,
+      "loss": 3.4375,
+      "step": 6027
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.71115168786145e-05,
+      "loss": 6.9688,
+      "step": 6028
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.711041515486799e-05,
+      "loss": 5.1406,
+      "step": 6029
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.710931323741193e-05,
+      "loss": 4.7188,
+      "step": 6030
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7108211126258773e-05,
+      "loss": 2.125,
+      "step": 6031
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7107108821421e-05,
+      "loss": 4.9062,
+      "step": 6032
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.71060063229111e-05,
+      "loss": 3.0625,
+      "step": 6033
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.710490363074155e-05,
+      "loss": 2.75,
+      "step": 6034
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.710380074492484e-05,
+      "loss": 3.5781,
+      "step": 6035
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7102697665473446e-05,
+      "loss": 2.6406,
+      "step": 6036
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.710159439239986e-05,
+      "loss": 4.2031,
+      "step": 6037
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.710049092571657e-05,
+      "loss": 4.0625,
+      "step": 6038
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.7099387265436074e-05,
+      "loss": 3.6875,
+      "step": 6039
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.709828341157086e-05,
+      "loss": 2.5625,
+      "step": 6040
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.709717936413343e-05,
+      "loss": 5.5312,
+      "step": 6041
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.709607512313628e-05,
+      "loss": 3.0938,
+      "step": 6042
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.709497068859191e-05,
+      "loss": 5.8281,
+      "step": 6043
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7093866060512834e-05,
+      "loss": 3.6562,
+      "step": 6044
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.709276123891154e-05,
+      "loss": 3.7188,
+      "step": 6045
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.709165622380056e-05,
+      "loss": 7.5,
+      "step": 6046
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.709055101519238e-05,
+      "loss": 3.2969,
+      "step": 6047
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7089445613099524e-05,
+      "loss": 3.6406,
+      "step": 6048
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.70883400175345e-05,
+      "loss": 3.9531,
+      "step": 6049
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.708723422850984e-05,
+      "loss": 5.5625,
+      "step": 6050
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.708612824603804e-05,
+      "loss": 4.2812,
+      "step": 6051
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.708502207013163e-05,
+      "loss": 3.4219,
+      "step": 6052
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7083915700803145e-05,
+      "loss": 4.4062,
+      "step": 6053
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.70828091380651e-05,
+      "loss": 4.5,
+      "step": 6054
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.708170238193002e-05,
+      "loss": 5.1562,
+      "step": 6055
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.708059543241045e-05,
+      "loss": 2.5,
+      "step": 6056
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.70794882895189e-05,
+      "loss": 5.0625,
+      "step": 6057
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.707838095326792e-05,
+      "loss": 3.1406,
+      "step": 6058
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7077273423670035e-05,
+      "loss": 3.625,
+      "step": 6059
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7076165700737794e-05,
+      "loss": 4.3125,
+      "step": 6060
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7075057784483734e-05,
+      "loss": 6.2344,
+      "step": 6061
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.70739496749204e-05,
+      "loss": 1.9688,
+      "step": 6062
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.707284137206032e-05,
+      "loss": 1.7812,
+      "step": 6063
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7071732875916066e-05,
+      "loss": 5.4844,
+      "step": 6064
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.707062418650018e-05,
+      "loss": 4.0625,
+      "step": 6065
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7069515303825205e-05,
+      "loss": 3.75,
+      "step": 6066
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.70684062279037e-05,
+      "loss": 2.7969,
+      "step": 6067
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.706729695874823e-05,
+      "loss": 3.5938,
+      "step": 6068
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7066187496371335e-05,
+      "loss": 7.9688,
+      "step": 6069
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.7065077840785594e-05,
+      "loss": 4.0469,
+      "step": 6070
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.706396799200355e-05,
+      "loss": 5.8594,
+      "step": 6071
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.706285795003778e-05,
+      "loss": 4.0625,
+      "step": 6072
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7061747714900856e-05,
+      "loss": 2.9531,
+      "step": 6073
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.706063728660534e-05,
+      "loss": 3.4062,
+      "step": 6074
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.705952666516379e-05,
+      "loss": 2.6875,
+      "step": 6075
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7058415850588804e-05,
+      "loss": 3.3906,
+      "step": 6076
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7057304842892954e-05,
+      "loss": 3.9844,
+      "step": 6077
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.70561936420888e-05,
+      "loss": 3.6562,
+      "step": 6078
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7055082248188935e-05,
+      "loss": 6.8125,
+      "step": 6079
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7053970661205946e-05,
+      "loss": 5.7031,
+      "step": 6080
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.70528588811524e-05,
+      "loss": 5.4688,
+      "step": 6081
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7051746908040904e-05,
+      "loss": 5.375,
+      "step": 6082
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7050634741884026e-05,
+      "loss": 6.2344,
+      "step": 6083
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7049522382694376e-05,
+      "loss": 3.4688,
+      "step": 6084
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.704840983048453e-05,
+      "loss": 5.0,
+      "step": 6085
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7047297085267106e-05,
+      "loss": 5.3281,
+      "step": 6086
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7046184147054685e-05,
+      "loss": 4.4375,
+      "step": 6087
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.704507101585987e-05,
+      "loss": 5.6094,
+      "step": 6088
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.704395769169526e-05,
+      "loss": 6.1562,
+      "step": 6089
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.704284417457346e-05,
+      "loss": 5.4844,
+      "step": 6090
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.704173046450708e-05,
+      "loss": 3.3281,
+      "step": 6091
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.704061656150873e-05,
+      "loss": 3.9062,
+      "step": 6092
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7039502465591015e-05,
+      "loss": 3.5312,
+      "step": 6093
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.703838817676654e-05,
+      "loss": 3.5781,
+      "step": 6094
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7037273695047947e-05,
+      "loss": 3.7656,
+      "step": 6095
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.703615902044783e-05,
+      "loss": 4.3438,
+      "step": 6096
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7035044152978806e-05,
+      "loss": 3.375,
+      "step": 6097
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.7033929092653516e-05,
+      "loss": 3.375,
+      "step": 6098
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.703281383948457e-05,
+      "loss": 1.75,
+      "step": 6099
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.70316983934846e-05,
+      "loss": 4.5312,
+      "step": 6100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.703058275466623e-05,
+      "loss": 3.625,
+      "step": 6101
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.702946692304209e-05,
+      "loss": 2.9062,
+      "step": 6102
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.702835089862482e-05,
+      "loss": 5.625,
+      "step": 6103
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.702723468142704e-05,
+      "loss": 2.375,
+      "step": 6104
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.70261182714614e-05,
+      "loss": 4.5938,
+      "step": 6105
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.702500166874053e-05,
+      "loss": 3.8125,
+      "step": 6106
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.702388487327708e-05,
+      "loss": 7.5312,
+      "step": 6107
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.702276788508369e-05,
+      "loss": 3.0625,
+      "step": 6108
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.7021650704173e-05,
+      "loss": 5.1562,
+      "step": 6109
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.702053333055766e-05,
+      "loss": 3.1562,
+      "step": 6110
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.701941576425034e-05,
+      "loss": 3.9375,
+      "step": 6111
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.7018298005263665e-05,
+      "loss": 4.0625,
+      "step": 6112
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.7017180053610295e-05,
+      "loss": 2.625,
+      "step": 6113
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.70160619093029e-05,
+      "loss": 2.9062,
+      "step": 6114
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.7014943572354116e-05,
+      "loss": 7.4062,
+      "step": 6115
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.7013825042776624e-05,
+      "loss": 3.4375,
+      "step": 6116
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.701270632058308e-05,
+      "loss": 3.4531,
+      "step": 6117
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.701158740578615e-05,
+      "loss": 7.2969,
+      "step": 6118
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.7010468298398506e-05,
+      "loss": 2.2812,
+      "step": 6119
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.70093489984328e-05,
+      "loss": 5.0312,
+      "step": 6120
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.7008229505901726e-05,
+      "loss": 3.625,
+      "step": 6121
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.700710982081794e-05,
+      "loss": 2.875,
+      "step": 6122
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.700598994319413e-05,
+      "loss": 3.9062,
+      "step": 6123
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.700486987304297e-05,
+      "loss": 5.2656,
+      "step": 6124
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.7003749610377146e-05,
+      "loss": 6.0312,
+      "step": 6125
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.700262915520933e-05,
+      "loss": 1.2812,
+      "step": 6126
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.7001508507552215e-05,
+      "loss": 4.2969,
+      "step": 6127
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.700038766741848e-05,
+      "loss": 3.0,
+      "step": 6128
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.6999266634820826e-05,
+      "loss": 5.4219,
+      "step": 6129
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.699814540977194e-05,
+      "loss": 2.0,
+      "step": 6130
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.699702399228451e-05,
+      "loss": 4.4062,
+      "step": 6131
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.699590238237123e-05,
+      "loss": 3.3281,
+      "step": 6132
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.6994780580044814e-05,
+      "loss": 4.5312,
+      "step": 6133
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6993658585317944e-05,
+      "loss": 2.8281,
+      "step": 6134
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6992536398203335e-05,
+      "loss": 4.0312,
+      "step": 6135
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.699141401871368e-05,
+      "loss": 3.0625,
+      "step": 6136
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.69902914468617e-05,
+      "loss": 3.2656,
+      "step": 6137
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6989168682660086e-05,
+      "loss": 3.4219,
+      "step": 6138
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6988045726121564e-05,
+      "loss": 3.8594,
+      "step": 6139
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.698692257725884e-05,
+      "loss": 3.4375,
+      "step": 6140
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.698579923608463e-05,
+      "loss": 2.2344,
+      "step": 6141
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6984675702611644e-05,
+      "loss": 3.6562,
+      "step": 6142
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.698355197685263e-05,
+      "loss": 6.125,
+      "step": 6143
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6982428058820275e-05,
+      "loss": 3.6562,
+      "step": 6144
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.698130394852732e-05,
+      "loss": 2.0,
+      "step": 6145
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.698017964598649e-05,
+      "loss": 2.0,
+      "step": 6146
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6979055151210516e-05,
+      "loss": 2.8906,
+      "step": 6147
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.697793046421213e-05,
+      "loss": 4.9062,
+      "step": 6148
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.697680558500405e-05,
+      "loss": 3.75,
+      "step": 6149
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.697568051359903e-05,
+      "loss": 3.0625,
+      "step": 6150
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6974555250009784e-05,
+      "loss": 2.7656,
+      "step": 6151
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.697342979424907e-05,
+      "loss": 4.1875,
+      "step": 6152
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.697230414632964e-05,
+      "loss": 4.1875,
+      "step": 6153
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6971178306264206e-05,
+      "loss": 3.5,
+      "step": 6154
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6970052274065536e-05,
+      "loss": 1.9688,
+      "step": 6155
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6968926049746376e-05,
+      "loss": 3.5156,
+      "step": 6156
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6967799633319466e-05,
+      "loss": 5.7188,
+      "step": 6157
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6966673024797566e-05,
+      "loss": 5.4375,
+      "step": 6158
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.696554622419343e-05,
+      "loss": 3.3594,
+      "step": 6159
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6964419231519814e-05,
+      "loss": 3.9062,
+      "step": 6160
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.696329204678948e-05,
+      "loss": 4.375,
+      "step": 6161
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.696216467001518e-05,
+      "loss": 4.125,
+      "step": 6162
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6961037101209685e-05,
+      "loss": 3.8438,
+      "step": 6163
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.6959909340385756e-05,
+      "loss": 3.7812,
+      "step": 6164
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.695878138755617e-05,
+      "loss": 4.3438,
+      "step": 6165
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.695765324273368e-05,
+      "loss": 4.2188,
+      "step": 6166
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.6956524905931075e-05,
+      "loss": 2.6875,
+      "step": 6167
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.6955396377161116e-05,
+      "loss": 4.75,
+      "step": 6168
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.695426765643659e-05,
+      "loss": 3.0938,
+      "step": 6169
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.695313874377027e-05,
+      "loss": 4.9375,
+      "step": 6170
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.695200963917494e-05,
+      "loss": 4.6719,
+      "step": 6171
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.695088034266337e-05,
+      "loss": 3.0625,
+      "step": 6172
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.694975085424836e-05,
+      "loss": 5.7344,
+      "step": 6173
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.6948621173942686e-05,
+      "loss": 5.4219,
+      "step": 6174
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.694749130175915e-05,
+      "loss": 4.0312,
+      "step": 6175
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.6946361237710536e-05,
+      "loss": 2.7812,
+      "step": 6176
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.6945230981809633e-05,
+      "loss": 3.4844,
+      "step": 6177
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.694410053406924e-05,
+      "loss": 8.1875,
+      "step": 6178
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.694296989450216e-05,
+      "loss": 4.5,
+      "step": 6179
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.69418390631212e-05,
+      "loss": 4.5938,
+      "step": 6180
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.6940708039939134e-05,
+      "loss": 3.1406,
+      "step": 6181
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.69395768249688e-05,
+      "loss": 2.7969,
+      "step": 6182
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.693844541822298e-05,
+      "loss": 5.75,
+      "step": 6183
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.693731381971449e-05,
+      "loss": 3.4844,
+      "step": 6184
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.693618202945615e-05,
+      "loss": 4.0156,
+      "step": 6185
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.693505004746077e-05,
+      "loss": 3.8594,
+      "step": 6186
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.693391787374115e-05,
+      "loss": 3.6875,
+      "step": 6187
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.693278550831012e-05,
+      "loss": 2.9688,
+      "step": 6188
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.693165295118051e-05,
+      "loss": 3.3438,
+      "step": 6189
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.6930520202365126e-05,
+      "loss": 3.8438,
+      "step": 6190
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.6929387261876794e-05,
+      "loss": 3.7188,
+      "step": 6191
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.6928254129728345e-05,
+      "loss": 1.1875,
+      "step": 6192
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.692712080593261e-05,
+      "loss": 4.4531,
+      "step": 6193
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.69259872905024e-05,
+      "loss": 3.3594,
+      "step": 6194
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.692485358345057e-05,
+      "loss": 3.6406,
+      "step": 6195
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.692371968478995e-05,
+      "loss": 3.75,
+      "step": 6196
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.692258559453338e-05,
+      "loss": 6.5156,
+      "step": 6197
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.692145131269368e-05,
+      "loss": 2.4531,
+      "step": 6198
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.692031683928371e-05,
+      "loss": 2.9375,
+      "step": 6199
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.691918217431631e-05,
+      "loss": 3.1875,
+      "step": 6200
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.6918047317804326e-05,
+      "loss": 3.75,
+      "step": 6201
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.69169122697606e-05,
+      "loss": 3.5312,
+      "step": 6202
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.691577703019799e-05,
+      "loss": 4.1719,
+      "step": 6203
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.691464159912935e-05,
+      "loss": 3.0,
+      "step": 6204
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.691350597656752e-05,
+      "loss": 4.3594,
+      "step": 6205
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.691237016252537e-05,
+      "loss": 2.375,
+      "step": 6206
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.6911234157015744e-05,
+      "loss": 5.375,
+      "step": 6207
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.691009796005152e-05,
+      "loss": 4.5938,
+      "step": 6208
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.690896157164556e-05,
+      "loss": 2.4688,
+      "step": 6209
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.690782499181071e-05,
+      "loss": 3.7812,
+      "step": 6210
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.690668822055986e-05,
+      "loss": 2.9375,
+      "step": 6211
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.690555125790586e-05,
+      "loss": 3.2812,
+      "step": 6212
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.6904414103861605e-05,
+      "loss": 2.3125,
+      "step": 6213
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.6903276758439946e-05,
+      "loss": 3.8594,
+      "step": 6214
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.690213922165377e-05,
+      "loss": 3.0625,
+      "step": 6215
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.690100149351596e-05,
+      "loss": 2.8125,
+      "step": 6216
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.689986357403939e-05,
+      "loss": 4.0,
+      "step": 6217
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.689872546323693e-05,
+      "loss": 4.4531,
+      "step": 6218
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.689758716112149e-05,
+      "loss": 2.5156,
+      "step": 6219
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.689644866770595e-05,
+      "loss": 3.2188,
+      "step": 6220
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.689530998300318e-05,
+      "loss": 4.9375,
+      "step": 6221
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.689417110702609e-05,
+      "loss": 5.125,
+      "step": 6222
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.689303203978757e-05,
+      "loss": 5.4688,
+      "step": 6223
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.689189278130051e-05,
+      "loss": 4.5,
+      "step": 6224
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.689075333157782e-05,
+      "loss": 3.0938,
+      "step": 6225
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.688961369063238e-05,
+      "loss": 3.2031,
+      "step": 6226
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.688847385847711e-05,
+      "loss": 4.3438,
+      "step": 6227
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.68873338351249e-05,
+      "loss": 4.7188,
+      "step": 6228
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.688619362058868e-05,
+      "loss": 4.6094,
+      "step": 6229
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.6885053214881324e-05,
+      "loss": 6.0938,
+      "step": 6230
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.688391261801577e-05,
+      "loss": 3.8281,
+      "step": 6231
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.688277183000493e-05,
+      "loss": 8.9219,
+      "step": 6232
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.68816308508617e-05,
+      "loss": 4.4766,
+      "step": 6233
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.6880489680599013e-05,
+      "loss": 6.4219,
+      "step": 6234
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.687934831922978e-05,
+      "loss": 3.1562,
+      "step": 6235
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.6878206766766925e-05,
+      "loss": 5.7969,
+      "step": 6236
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.687706502322338e-05,
+      "loss": 3.7812,
+      "step": 6237
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.687592308861206e-05,
+      "loss": 6.0469,
+      "step": 6238
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.6874780962945896e-05,
+      "loss": 3.5469,
+      "step": 6239
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.687363864623782e-05,
+      "loss": 6.0156,
+      "step": 6240
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.687249613850076e-05,
+      "loss": 6.2969,
+      "step": 6241
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.6871353439747657e-05,
+      "loss": 3.4688,
+      "step": 6242
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.687021054999144e-05,
+      "loss": 2.9062,
+      "step": 6243
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.6869067469245064e-05,
+      "loss": 6.8438,
+      "step": 6244
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.686792419752145e-05,
+      "loss": 2.6094,
+      "step": 6245
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.686678073483354e-05,
+      "loss": 7.0156,
+      "step": 6246
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.6865637081194304e-05,
+      "loss": 6.4844,
+      "step": 6247
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.6864493236616675e-05,
+      "loss": 4.5,
+      "step": 6248
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.686334920111359e-05,
+      "loss": 2.2344,
+      "step": 6249
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.6862204974698016e-05,
+      "loss": 4.4375,
+      "step": 6250
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.6861060557382905e-05,
+      "loss": 8.8281,
+      "step": 6251
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.6859915949181214e-05,
+      "loss": 6.0781,
+      "step": 6252
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.6858771150105894e-05,
+      "loss": 3.3125,
+      "step": 6253
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.685762616016992e-05,
+      "loss": 3.1406,
+      "step": 6254
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.685648097938623e-05,
+      "loss": 3.0156,
+      "step": 6255
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.685533560776782e-05,
+      "loss": 3.5625,
+      "step": 6256
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.685419004532763e-05,
+      "loss": 3.1875,
+      "step": 6257
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.6853044292078646e-05,
+      "loss": 5.9219,
+      "step": 6258
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.6851898348033824e-05,
+      "loss": 4.125,
+      "step": 6259
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.685075221320615e-05,
+      "loss": 3.2188,
+      "step": 6260
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.6849605887608604e-05,
+      "loss": 4.5,
+      "step": 6261
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.6848459371254154e-05,
+      "loss": 3.9219,
+      "step": 6262
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.684731266415577e-05,
+      "loss": 2.3906,
+      "step": 6263
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.684616576632645e-05,
+      "loss": 4.4219,
+      "step": 6264
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.684501867777918e-05,
+      "loss": 4.9062,
+      "step": 6265
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.684387139852694e-05,
+      "loss": 5.5938,
+      "step": 6266
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.684272392858271e-05,
+      "loss": 6.0156,
+      "step": 6267
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.68415762679595e-05,
+      "loss": 5.9375,
+      "step": 6268
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.684042841667029e-05,
+      "loss": 4.2031,
+      "step": 6269
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.683928037472807e-05,
+      "loss": 3.4531,
+      "step": 6270
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.683813214214585e-05,
+      "loss": 5.1406,
+      "step": 6271
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.683698371893662e-05,
+      "loss": 3.3281,
+      "step": 6272
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.683583510511339e-05,
+      "loss": 3.8125,
+      "step": 6273
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.683468630068916e-05,
+      "loss": 4.2656,
+      "step": 6274
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.6833537305676925e-05,
+      "loss": 4.875,
+      "step": 6275
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.6832388120089706e-05,
+      "loss": 4.5156,
+      "step": 6276
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.683123874394052e-05,
+      "loss": 3.4531,
+      "step": 6277
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.683008917724236e-05,
+      "loss": 3.5938,
+      "step": 6278
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.682893942000825e-05,
+      "loss": 3.9219,
+      "step": 6279
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.682778947225121e-05,
+      "loss": 3.4375,
+      "step": 6280
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.682663933398424e-05,
+      "loss": 3.5,
+      "step": 6281
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.682548900522039e-05,
+      "loss": 2.8438,
+      "step": 6282
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.682433848597267e-05,
+      "loss": 2.75,
+      "step": 6283
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.682318777625409e-05,
+      "loss": 3.625,
+      "step": 6284
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.6822036876077706e-05,
+      "loss": 3.3281,
+      "step": 6285
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.682088578545653e-05,
+      "loss": 3.1562,
+      "step": 6286
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.681973450440359e-05,
+      "loss": 4.0,
+      "step": 6287
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.681858303293193e-05,
+      "loss": 5.1172,
+      "step": 6288
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.681743137105458e-05,
+      "loss": 3.0469,
+      "step": 6289
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.681627951878458e-05,
+      "loss": 4.0312,
+      "step": 6290
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.6815127476134966e-05,
+      "loss": 4.75,
+      "step": 6291
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.681397524311879e-05,
+      "loss": 2.5938,
+      "step": 6292
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.6812822819749096e-05,
+      "loss": 2.7812,
+      "step": 6293
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.681167020603892e-05,
+      "loss": 3.125,
+      "step": 6294
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.681051740200132e-05,
+      "loss": 5.4688,
+      "step": 6295
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.680936440764934e-05,
+      "loss": 1.9062,
+      "step": 6296
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.680821122299604e-05,
+      "loss": 2.8125,
+      "step": 6297
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.680705784805448e-05,
+      "loss": 4.2969,
+      "step": 6298
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.68059042828377e-05,
+      "loss": 6.5,
+      "step": 6299
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.6804750527358776e-05,
+      "loss": 6.2656,
+      "step": 6300
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.680359658163076e-05,
+      "loss": 2.0,
+      "step": 6301
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.6802442445666724e-05,
+      "loss": 5.4688,
+      "step": 6302
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.680128811947973e-05,
+      "loss": 2.5312,
+      "step": 6303
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.680013360308285e-05,
+      "loss": 3.2812,
+      "step": 6304
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.679897889648914e-05,
+      "loss": 3.2031,
+      "step": 6305
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.679782399971169e-05,
+      "loss": 2.8438,
+      "step": 6306
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.679666891276357e-05,
+      "loss": 2.5312,
+      "step": 6307
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.6795513635657846e-05,
+      "loss": 5.2344,
+      "step": 6308
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.679435816840761e-05,
+      "loss": 6.9844,
+      "step": 6309
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.6793202511025945e-05,
+      "loss": 4.0781,
+      "step": 6310
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.6792046663525924e-05,
+      "loss": 3.8594,
+      "step": 6311
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.679089062592064e-05,
+      "loss": 3.875,
+      "step": 6312
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.678973439822317e-05,
+      "loss": 2.9062,
+      "step": 6313
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.6788577980446626e-05,
+      "loss": 4.9844,
+      "step": 6314
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.678742137260408e-05,
+      "loss": 4.25,
+      "step": 6315
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.678626457470863e-05,
+      "loss": 3.3906,
+      "step": 6316
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.6785107586773374e-05,
+      "loss": 2.75,
+      "step": 6317
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.678395040881141e-05,
+      "loss": 5.7969,
+      "step": 6318
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.6782793040835835e-05,
+      "loss": 3.3438,
+      "step": 6319
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.678163548285977e-05,
+      "loss": 3.5312,
+      "step": 6320
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.67804777348963e-05,
+      "loss": 3.6875,
+      "step": 6321
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.677931979695853e-05,
+      "loss": 3.8438,
+      "step": 6322
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.6778161669059575e-05,
+      "loss": 4.2031,
+      "step": 6323
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.6777003351212554e-05,
+      "loss": 7.75,
+      "step": 6324
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.677584484343058e-05,
+      "loss": 6.7344,
+      "step": 6325
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.677468614572676e-05,
+      "loss": 3.4062,
+      "step": 6326
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.677352725811421e-05,
+      "loss": 3.5312,
+      "step": 6327
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.677236818060606e-05,
+      "loss": 4.2344,
+      "step": 6328
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.677120891321543e-05,
+      "loss": 3.1875,
+      "step": 6329
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.677004945595544e-05,
+      "loss": 6.75,
+      "step": 6330
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.676888980883921e-05,
+      "loss": 4.0469,
+      "step": 6331
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.676772997187989e-05,
+      "loss": 4.4688,
+      "step": 6332
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.676656994509059e-05,
+      "loss": 3.3281,
+      "step": 6333
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.676540972848445e-05,
+      "loss": 3.9062,
+      "step": 6334
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.67642493220746e-05,
+      "loss": 5.5781,
+      "step": 6335
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.676308872587419e-05,
+      "loss": 3.9688,
+      "step": 6336
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.6761927939896346e-05,
+      "loss": 2.3125,
+      "step": 6337
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.6760766964154215e-05,
+      "loss": 3.5312,
+      "step": 6338
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.675960579866095e-05,
+      "loss": 6.6719,
+      "step": 6339
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.675844444342968e-05,
+      "loss": 4.8594,
+      "step": 6340
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.6757282898473555e-05,
+      "loss": 3.3125,
+      "step": 6341
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.675612116380573e-05,
+      "loss": 3.25,
+      "step": 6342
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.675495923943937e-05,
+      "loss": 3.625,
+      "step": 6343
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.675379712538761e-05,
+      "loss": 3.6875,
+      "step": 6344
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.675263482166361e-05,
+      "loss": 4.5,
+      "step": 6345
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.675147232828053e-05,
+      "loss": 3.4688,
+      "step": 6346
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.6750309645251536e-05,
+      "loss": 3.2656,
+      "step": 6347
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.674914677258979e-05,
+      "loss": 4.3438,
+      "step": 6348
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.674798371030845e-05,
+      "loss": 3.5,
+      "step": 6349
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.674682045842069e-05,
+      "loss": 4.0,
+      "step": 6350
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.6745657016939675e-05,
+      "loss": 3.7344,
+      "step": 6351
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.674449338587858e-05,
+      "loss": 2.3047,
+      "step": 6352
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.6743329565250575e-05,
+      "loss": 2.0469,
+      "step": 6353
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.6742165555068835e-05,
+      "loss": 3.0625,
+      "step": 6354
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.674100135534655e-05,
+      "loss": 6.0469,
+      "step": 6355
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.673983696609688e-05,
+      "loss": 5.1562,
+      "step": 6356
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.6738672387333026e-05,
+      "loss": 4.4375,
+      "step": 6357
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.673750761906816e-05,
+      "loss": 6.5625,
+      "step": 6358
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.673634266131548e-05,
+      "loss": 5.2188,
+      "step": 6359
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.673517751408816e-05,
+      "loss": 3.2969,
+      "step": 6360
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.6734012177399406e-05,
+      "loss": 3.4688,
+      "step": 6361
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.67328466512624e-05,
+      "loss": 4.125,
+      "step": 6362
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.673168093569033e-05,
+      "loss": 3.1875,
+      "step": 6363
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.673051503069642e-05,
+      "loss": 2.4531,
+      "step": 6364
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.672934893629384e-05,
+      "loss": 2.5312,
+      "step": 6365
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.672818265249581e-05,
+      "loss": 3.4688,
+      "step": 6366
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.6727016179315526e-05,
+      "loss": 5.2812,
+      "step": 6367
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.6725849516766196e-05,
+      "loss": 2.4062,
+      "step": 6368
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.672468266486103e-05,
+      "loss": 3.7812,
+      "step": 6369
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.672351562361324e-05,
+      "loss": 2.4375,
+      "step": 6370
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.672234839303603e-05,
+      "loss": 2.25,
+      "step": 6371
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.672118097314262e-05,
+      "loss": 2.8438,
+      "step": 6372
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.672001336394623e-05,
+      "loss": 4.0156,
+      "step": 6373
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.671884556546006e-05,
+      "loss": 2.4062,
+      "step": 6374
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.6717677577697354e-05,
+      "loss": 5.0312,
+      "step": 6375
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.671650940067132e-05,
+      "loss": 5.1094,
+      "step": 6376
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.6715341034395205e-05,
+      "loss": 5.4375,
+      "step": 6377
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.6714172478882205e-05,
+      "loss": 3.5938,
+      "step": 6378
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.671300373414557e-05,
+      "loss": 4.6406,
+      "step": 6379
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.6711834800198527e-05,
+      "loss": 3.6875,
+      "step": 6380
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.6710665677054304e-05,
+      "loss": 4.2344,
+      "step": 6381
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.670949636472615e-05,
+      "loss": 3.7656,
+      "step": 6382
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.670832686322729e-05,
+      "loss": 4.7031,
+      "step": 6383
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.6707157172570964e-05,
+      "loss": 4.0312,
+      "step": 6384
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.6705987292770414e-05,
+      "loss": 6.9062,
+      "step": 6385
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.6704817223838906e-05,
+      "loss": 3.375,
+      "step": 6386
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.6703646965789655e-05,
+      "loss": 2.4062,
+      "step": 6387
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.670247651863593e-05,
+      "loss": 5.8125,
+      "step": 6388
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.670130588239098e-05,
+      "loss": 4.5312,
+      "step": 6389
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.670013505706805e-05,
+      "loss": 4.9062,
+      "step": 6390
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.66989640426804e-05,
+      "loss": 5.9688,
+      "step": 6391
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.669779283924129e-05,
+      "loss": 2.9375,
+      "step": 6392
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.669662144676397e-05,
+      "loss": 2.6094,
+      "step": 6393
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.6695449865261715e-05,
+      "loss": 7.2656,
+      "step": 6394
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.669427809474777e-05,
+      "loss": 5.0625,
+      "step": 6395
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.6693106135235416e-05,
+      "loss": 3.3125,
+      "step": 6396
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.6691933986737924e-05,
+      "loss": 5.3281,
+      "step": 6397
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.669076164926855e-05,
+      "loss": 2.7656,
+      "step": 6398
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.668958912284057e-05,
+      "loss": 4.7188,
+      "step": 6399
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.668841640746727e-05,
+      "loss": 3.0625,
+      "step": 6400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.668724350316192e-05,
+      "loss": 5.1094,
+      "step": 6401
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.668607040993779e-05,
+      "loss": 2.4688,
+      "step": 6402
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.668489712780816e-05,
+      "loss": 6.0625,
+      "step": 6403
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.668372365678633e-05,
+      "loss": 3.8594,
+      "step": 6404
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.668254999688557e-05,
+      "loss": 5.0625,
+      "step": 6405
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.668137614811918e-05,
+      "loss": 3.7188,
+      "step": 6406
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.6680202110500436e-05,
+      "loss": 2.4688,
+      "step": 6407
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6679027884042635e-05,
+      "loss": 3.0938,
+      "step": 6408
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.667785346875907e-05,
+      "loss": 2.9844,
+      "step": 6409
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.667667886466304e-05,
+      "loss": 3.25,
+      "step": 6410
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.667550407176784e-05,
+      "loss": 3.7344,
+      "step": 6411
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6674329090086773e-05,
+      "loss": 4.5625,
+      "step": 6412
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.667315391963313e-05,
+      "loss": 4.3125,
+      "step": 6413
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.667197856042023e-05,
+      "loss": 4.3906,
+      "step": 6414
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.667080301246137e-05,
+      "loss": 2.5156,
+      "step": 6415
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.666962727576987e-05,
+      "loss": 4.0938,
+      "step": 6416
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.666845135035902e-05,
+      "loss": 6.3438,
+      "step": 6417
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.666727523624215e-05,
+      "loss": 4.9844,
+      "step": 6418
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.666609893343258e-05,
+      "loss": 2.8125,
+      "step": 6419
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.666492244194361e-05,
+      "loss": 2.6875,
+      "step": 6420
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6663745761788564e-05,
+      "loss": 3.4375,
+      "step": 6421
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.666256889298077e-05,
+      "loss": 3.8125,
+      "step": 6422
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.666139183553354e-05,
+      "loss": 3.4062,
+      "step": 6423
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.666021458946022e-05,
+      "loss": 4.1719,
+      "step": 6424
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.665903715477411e-05,
+      "loss": 3.7656,
+      "step": 6425
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6657859531488564e-05,
+      "loss": 3.5312,
+      "step": 6426
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.66566817196169e-05,
+      "loss": 5.0938,
+      "step": 6427
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.665550371917246e-05,
+      "loss": 3.6094,
+      "step": 6428
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6654325530168575e-05,
+      "loss": 2.6562,
+      "step": 6429
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.665314715261859e-05,
+      "loss": 3.375,
+      "step": 6430
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.665196858653584e-05,
+      "loss": 4.3594,
+      "step": 6431
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6650789831933665e-05,
+      "loss": 2.7188,
+      "step": 6432
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6649610888825414e-05,
+      "loss": 5.4062,
+      "step": 6433
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.664843175722444e-05,
+      "loss": 3.5,
+      "step": 6434
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.664725243714409e-05,
+      "loss": 3.4062,
+      "step": 6435
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.66460729285977e-05,
+      "loss": 4.6719,
+      "step": 6436
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.6644893231598635e-05,
+      "loss": 2.5781,
+      "step": 6437
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.664371334616026e-05,
+      "loss": 4.2812,
+      "step": 6438
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6642533272295916e-05,
+      "loss": 3.1406,
+      "step": 6439
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6641353010018966e-05,
+      "loss": 5.3438,
+      "step": 6440
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.664017255934278e-05,
+      "loss": 2.9219,
+      "step": 6441
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.663899192028072e-05,
+      "loss": 6.1719,
+      "step": 6442
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.663781109284615e-05,
+      "loss": 3.7031,
+      "step": 6443
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6636630077052435e-05,
+      "loss": 4.3438,
+      "step": 6444
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.663544887291295e-05,
+      "loss": 4.5,
+      "step": 6445
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6634267480441064e-05,
+      "loss": 4.5625,
+      "step": 6446
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.663308589965016e-05,
+      "loss": 2.4062,
+      "step": 6447
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.66319041305536e-05,
+      "loss": 6.6875,
+      "step": 6448
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6630722173164775e-05,
+      "loss": 3.2344,
+      "step": 6449
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.662954002749706e-05,
+      "loss": 5.6875,
+      "step": 6450
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.662835769356385e-05,
+      "loss": 3.8438,
+      "step": 6451
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6627175171378515e-05,
+      "loss": 3.1875,
+      "step": 6452
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.662599246095445e-05,
+      "loss": 3.2188,
+      "step": 6453
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.662480956230505e-05,
+      "loss": 2.5312,
+      "step": 6454
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.66236264754437e-05,
+      "loss": 4.6094,
+      "step": 6455
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.662244320038378e-05,
+      "loss": 3.625,
+      "step": 6456
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.662125973713871e-05,
+      "loss": 7.7344,
+      "step": 6457
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6620076085721875e-05,
+      "loss": 2.8906,
+      "step": 6458
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6618892246146684e-05,
+      "loss": 4.3125,
+      "step": 6459
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6617708218426536e-05,
+      "loss": 3.0,
+      "step": 6460
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.661652400257483e-05,
+      "loss": 8.1094,
+      "step": 6461
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6615339598604975e-05,
+      "loss": 3.0625,
+      "step": 6462
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.661415500653038e-05,
+      "loss": 3.875,
+      "step": 6463
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.661297022636446e-05,
+      "loss": 2.2188,
+      "step": 6464
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.661178525812063e-05,
+      "loss": 4.0625,
+      "step": 6465
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.6610600101812294e-05,
+      "loss": 4.7812,
+      "step": 6466
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.660941475745288e-05,
+      "loss": 6.7344,
+      "step": 6467
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.66082292250558e-05,
+      "loss": 3.4531,
+      "step": 6468
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.6607043504634476e-05,
+      "loss": 3.9062,
+      "step": 6469
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.660585759620234e-05,
+      "loss": 3.4688,
+      "step": 6470
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.660467149977281e-05,
+      "loss": 5.2188,
+      "step": 6471
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.660348521535932e-05,
+      "loss": 1.75,
+      "step": 6472
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.66022987429753e-05,
+      "loss": 5.9844,
+      "step": 6473
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.6601112082634166e-05,
+      "loss": 4.3438,
+      "step": 6474
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.6599925234349375e-05,
+      "loss": 3.9375,
+      "step": 6475
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.659873819813435e-05,
+      "loss": 2.5,
+      "step": 6476
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.659755097400253e-05,
+      "loss": 4.0312,
+      "step": 6477
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.659636356196736e-05,
+      "loss": 3.6875,
+      "step": 6478
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.659517596204229e-05,
+      "loss": 3.375,
+      "step": 6479
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.659398817424075e-05,
+      "loss": 2.9219,
+      "step": 6480
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.6592800198576195e-05,
+      "loss": 6.2031,
+      "step": 6481
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.659161203506208e-05,
+      "loss": 4.0,
+      "step": 6482
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.659042368371184e-05,
+      "loss": 3.6406,
+      "step": 6483
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.658923514453894e-05,
+      "loss": 4.0312,
+      "step": 6484
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.658804641755683e-05,
+      "loss": 2.9688,
+      "step": 6485
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.658685750277898e-05,
+      "loss": 6.0156,
+      "step": 6486
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.6585668400218836e-05,
+      "loss": 3.3906,
+      "step": 6487
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.658447910988987e-05,
+      "loss": 8.25,
+      "step": 6488
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.6583289631805544e-05,
+      "loss": 1.875,
+      "step": 6489
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.658209996597931e-05,
+      "loss": 4.4375,
+      "step": 6490
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.658091011242466e-05,
+      "loss": 4.9688,
+      "step": 6491
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.657972007115505e-05,
+      "loss": 4.0,
+      "step": 6492
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.657852984218395e-05,
+      "loss": 7.1406,
+      "step": 6493
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.657733942552484e-05,
+      "loss": 4.6094,
+      "step": 6494
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.65761488211912e-05,
+      "loss": 5.375,
+      "step": 6495
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.657495802919651e-05,
+      "loss": 5.3906,
+      "step": 6496
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.657376704955424e-05,
+      "loss": 5.0938,
+      "step": 6497
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.657257588227789e-05,
+      "loss": 4.5,
+      "step": 6498
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.6571384527380924e-05,
+      "loss": 2.9375,
+      "step": 6499
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.657019298487685e-05,
+      "loss": 4.1562,
+      "step": 6500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.656900125477914e-05,
+      "loss": 3.125,
+      "step": 6501
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.65678093371013e-05,
+      "loss": 6.2656,
+      "step": 6502
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.656661723185682e-05,
+      "loss": 5.6094,
+      "step": 6503
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.6565424939059194e-05,
+      "loss": 2.9375,
+      "step": 6504
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.656423245872191e-05,
+      "loss": 4.7812,
+      "step": 6505
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.65630397908585e-05,
+      "loss": 3.4844,
+      "step": 6506
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.656184693548243e-05,
+      "loss": 2.0469,
+      "step": 6507
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.656065389260722e-05,
+      "loss": 2.875,
+      "step": 6508
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.655946066224638e-05,
+      "loss": 5.6562,
+      "step": 6509
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.655826724441341e-05,
+      "loss": 7.1875,
+      "step": 6510
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.655707363912184e-05,
+      "loss": 3.8906,
+      "step": 6511
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.655587984638515e-05,
+      "loss": 5.4375,
+      "step": 6512
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.6554685866216885e-05,
+      "loss": 6.9375,
+      "step": 6513
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.655349169863055e-05,
+      "loss": 4.2031,
+      "step": 6514
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.6552297343639665e-05,
+      "loss": 2.5625,
+      "step": 6515
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.655110280125776e-05,
+      "loss": 4.0,
+      "step": 6516
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.654990807149834e-05,
+      "loss": 2.5469,
+      "step": 6517
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.654871315437494e-05,
+      "loss": 3.0156,
+      "step": 6518
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.6547518049901094e-05,
+      "loss": 3.8594,
+      "step": 6519
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.6546322758090334e-05,
+      "loss": 3.625,
+      "step": 6520
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.654512727895618e-05,
+      "loss": 2.5625,
+      "step": 6521
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.654393161251216e-05,
+      "loss": 2.3438,
+      "step": 6522
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.654273575877184e-05,
+      "loss": 6.6875,
+      "step": 6523
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.6541539717748725e-05,
+      "loss": 3.0,
+      "step": 6524
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.654034348945638e-05,
+      "loss": 4.5,
+      "step": 6525
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.653914707390833e-05,
+      "loss": 3.9688,
+      "step": 6526
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.653795047111814e-05,
+      "loss": 4.25,
+      "step": 6527
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.653675368109934e-05,
+      "loss": 7.5312,
+      "step": 6528
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.653555670386548e-05,
+      "loss": 3.2188,
+      "step": 6529
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.653435953943012e-05,
+      "loss": 3.75,
+      "step": 6530
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.653316218780681e-05,
+      "loss": 3.8438,
+      "step": 6531
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6531964649009096e-05,
+      "loss": 3.8906,
+      "step": 6532
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.653076692305055e-05,
+      "loss": 3.1719,
+      "step": 6533
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.652956900994473e-05,
+      "loss": 2.8438,
+      "step": 6534
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6528370909705185e-05,
+      "loss": 5.5312,
+      "step": 6535
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6527172622345486e-05,
+      "loss": 4.6562,
+      "step": 6536
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6525974147879204e-05,
+      "loss": 3.5,
+      "step": 6537
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.65247754863199e-05,
+      "loss": 4.5781,
+      "step": 6538
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.652357663768115e-05,
+      "loss": 3.7812,
+      "step": 6539
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.652237760197652e-05,
+      "loss": 3.3281,
+      "step": 6540
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.652117837921959e-05,
+      "loss": 3.4062,
+      "step": 6541
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.651997896942394e-05,
+      "loss": 2.625,
+      "step": 6542
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6518779372603135e-05,
+      "loss": 2.6875,
+      "step": 6543
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6517579588770765e-05,
+      "loss": 3.3438,
+      "step": 6544
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6516379617940416e-05,
+      "loss": 3.8125,
+      "step": 6545
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6515179460125666e-05,
+      "loss": 3.0781,
+      "step": 6546
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.65139791153401e-05,
+      "loss": 3.7031,
+      "step": 6547
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6512778583597325e-05,
+      "loss": 3.4844,
+      "step": 6548
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6511577864910914e-05,
+      "loss": 3.0469,
+      "step": 6549
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.651037695929446e-05,
+      "loss": 2.6875,
+      "step": 6550
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6509175866761575e-05,
+      "loss": 3.5469,
+      "step": 6551
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.650797458732585e-05,
+      "loss": 5.9688,
+      "step": 6552
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6506773121000866e-05,
+      "loss": 5.2812,
+      "step": 6553
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.650557146780025e-05,
+      "loss": 2.875,
+      "step": 6554
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.65043696277376e-05,
+      "loss": 3.9062,
+      "step": 6555
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.650316760082651e-05,
+      "loss": 5.1875,
+      "step": 6556
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.65019653870806e-05,
+      "loss": 6.1875,
+      "step": 6557
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.650076298651348e-05,
+      "loss": 2.6719,
+      "step": 6558
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6499560399138766e-05,
+      "loss": 2.9844,
+      "step": 6559
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.649835762497006e-05,
+      "loss": 3.0312,
+      "step": 6560
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6497154664020984e-05,
+      "loss": 4.5938,
+      "step": 6561
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.649595151630516e-05,
+      "loss": 5.7812,
+      "step": 6562
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.649474818183621e-05,
+      "loss": 3.8125,
+      "step": 6563
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6493544660627754e-05,
+      "loss": 4.2812,
+      "step": 6564
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6492340952693415e-05,
+      "loss": 4.5781,
+      "step": 6565
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6491137058046824e-05,
+      "loss": 3.1562,
+      "step": 6566
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6489932976701615e-05,
+      "loss": 2.9375,
+      "step": 6567
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.648872870867141e-05,
+      "loss": 3.875,
+      "step": 6568
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6487524253969844e-05,
+      "loss": 3.3438,
+      "step": 6569
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.648631961261056e-05,
+      "loss": 1.1875,
+      "step": 6570
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6485114784607194e-05,
+      "loss": 4.0625,
+      "step": 6571
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.648390976997338e-05,
+      "loss": 1.9844,
+      "step": 6572
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.648270456872276e-05,
+      "loss": 3.3125,
+      "step": 6573
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6481499180868995e-05,
+      "loss": 4.4531,
+      "step": 6574
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.648029360642571e-05,
+      "loss": 4.4062,
+      "step": 6575
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.647908784540656e-05,
+      "loss": 3.5781,
+      "step": 6576
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.64778818978252e-05,
+      "loss": 3.6875,
+      "step": 6577
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.647667576369529e-05,
+      "loss": 2.5,
+      "step": 6578
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.647546944303046e-05,
+      "loss": 3.5469,
+      "step": 6579
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.647426293584439e-05,
+      "loss": 3.9219,
+      "step": 6580
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.647305624215073e-05,
+      "loss": 2.625,
+      "step": 6581
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6471849361963144e-05,
+      "loss": 4.0938,
+      "step": 6582
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6470642295295294e-05,
+      "loss": 3.4375,
+      "step": 6583
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6469435042160844e-05,
+      "loss": 3.9219,
+      "step": 6584
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6468227602573457e-05,
+      "loss": 4.7812,
+      "step": 6585
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.646701997654681e-05,
+      "loss": 6.0781,
+      "step": 6586
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.646581216409458e-05,
+      "loss": 3.4688,
+      "step": 6587
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.646460416523043e-05,
+      "loss": 8.1094,
+      "step": 6588
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.646339597996804e-05,
+      "loss": 3.4375,
+      "step": 6589
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.646218760832109e-05,
+      "loss": 3.4219,
+      "step": 6590
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.646097905030325e-05,
+      "loss": 2.25,
+      "step": 6591
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.645977030592822e-05,
+      "loss": 3.2812,
+      "step": 6592
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6458561375209666e-05,
+      "loss": 4.125,
+      "step": 6593
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6457352258161286e-05,
+      "loss": 3.9531,
+      "step": 6594
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.645614295479677e-05,
+      "loss": 4.9062,
+      "step": 6595
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.64549334651298e-05,
+      "loss": 7.6406,
+      "step": 6596
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6453723789174074e-05,
+      "loss": 4.75,
+      "step": 6597
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.645251392694328e-05,
+      "loss": 2.5,
+      "step": 6598
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.645130387845113e-05,
+      "loss": 2.5625,
+      "step": 6599
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.645009364371131e-05,
+      "loss": 4.75,
+      "step": 6600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.644888322273753e-05,
+      "loss": 2.6875,
+      "step": 6601
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.644767261554348e-05,
+      "loss": 3.9688,
+      "step": 6602
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.644646182214288e-05,
+      "loss": 2.1406,
+      "step": 6603
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.644525084254943e-05,
+      "loss": 3.2812,
+      "step": 6604
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6444039676776843e-05,
+      "loss": 3.9062,
+      "step": 6605
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.644282832483883e-05,
+      "loss": 5.8125,
+      "step": 6606
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.64416167867491e-05,
+      "loss": 6.8438,
+      "step": 6607
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.644040506252138e-05,
+      "loss": 3.5938,
+      "step": 6608
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.643919315216938e-05,
+      "loss": 2.7344,
+      "step": 6609
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.643798105570681e-05,
+      "loss": 4.2188,
+      "step": 6610
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.643676877314741e-05,
+      "loss": 3.2344,
+      "step": 6611
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.64355563045049e-05,
+      "loss": 4.0312,
+      "step": 6612
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.643434364979301e-05,
+      "loss": 2.6562,
+      "step": 6613
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6433130809025465e-05,
+      "loss": 3.1562,
+      "step": 6614
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6431917782215986e-05,
+      "loss": 5.6094,
+      "step": 6615
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.643070456937832e-05,
+      "loss": 3.1562,
+      "step": 6616
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.642949117052619e-05,
+      "loss": 6.2344,
+      "step": 6617
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6428277585673344e-05,
+      "loss": 2.0938,
+      "step": 6618
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.6427063814833506e-05,
+      "loss": 2.5312,
+      "step": 6619
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.642584985802044e-05,
+      "loss": 5.2188,
+      "step": 6620
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6424635715247874e-05,
+      "loss": 4.6875,
+      "step": 6621
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.642342138652955e-05,
+      "loss": 3.25,
+      "step": 6622
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.642220687187923e-05,
+      "loss": 3.375,
+      "step": 6623
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.642099217131065e-05,
+      "loss": 4.5312,
+      "step": 6624
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6419777284837567e-05,
+      "loss": 3.5,
+      "step": 6625
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.641856221247374e-05,
+      "loss": 3.125,
+      "step": 6626
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.641734695423292e-05,
+      "loss": 5.2969,
+      "step": 6627
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.641613151012886e-05,
+      "loss": 3.6562,
+      "step": 6628
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.641491588017534e-05,
+      "loss": 4.4062,
+      "step": 6629
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.641370006438609e-05,
+      "loss": 2.5938,
+      "step": 6630
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.64124840627749e-05,
+      "loss": 4.4062,
+      "step": 6631
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.641126787535553e-05,
+      "loss": 3.7031,
+      "step": 6632
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6410051502141756e-05,
+      "loss": 5.3438,
+      "step": 6633
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.640883494314733e-05,
+      "loss": 3.3438,
+      "step": 6634
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6407618198386036e-05,
+      "loss": 2.7188,
+      "step": 6635
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6406401267871656e-05,
+      "loss": 3.3438,
+      "step": 6636
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6405184151617954e-05,
+      "loss": 2.8125,
+      "step": 6637
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6403966849638705e-05,
+      "loss": 3.4375,
+      "step": 6638
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.640274936194771e-05,
+      "loss": 1.9219,
+      "step": 6639
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.640153168855874e-05,
+      "loss": 4.375,
+      "step": 6640
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.640031382948558e-05,
+      "loss": 5.0547,
+      "step": 6641
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.639909578474203e-05,
+      "loss": 5.4375,
+      "step": 6642
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.639787755434186e-05,
+      "loss": 4.2188,
+      "step": 6643
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6396659138298875e-05,
+      "loss": 7.2031,
+      "step": 6644
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.639544053662687e-05,
+      "loss": 6.2031,
+      "step": 6645
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6394221749339625e-05,
+      "loss": 5.1719,
+      "step": 6646
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.639300277645096e-05,
+      "loss": 5.4688,
+      "step": 6647
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.639178361797466e-05,
+      "loss": 5.8594,
+      "step": 6648
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.639056427392453e-05,
+      "loss": 6.1094,
+      "step": 6649
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.638934474431438e-05,
+      "loss": 5.625,
+      "step": 6650
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.638812502915801e-05,
+      "loss": 4.1875,
+      "step": 6651
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.638690512846923e-05,
+      "loss": 5.1719,
+      "step": 6652
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.638568504226186e-05,
+      "loss": 3.0,
+      "step": 6653
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.63844647705497e-05,
+      "loss": 3.6875,
+      "step": 6654
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.638324431334657e-05,
+      "loss": 3.4219,
+      "step": 6655
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.638202367066629e-05,
+      "loss": 4.0,
+      "step": 6656
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.638080284252267e-05,
+      "loss": 5.4844,
+      "step": 6657
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.637958182892954e-05,
+      "loss": 6.2812,
+      "step": 6658
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6378360629900714e-05,
+      "loss": 3.8125,
+      "step": 6659
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.637713924545003e-05,
+      "loss": 7.6719,
+      "step": 6660
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.637591767559131e-05,
+      "loss": 7.4062,
+      "step": 6661
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.637469592033838e-05,
+      "loss": 5.4062,
+      "step": 6662
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.637347397970507e-05,
+      "loss": 3.1719,
+      "step": 6663
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6372251853705226e-05,
+      "loss": 3.0,
+      "step": 6664
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.637102954235267e-05,
+      "loss": 4.4844,
+      "step": 6665
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.636980704566125e-05,
+      "loss": 6.0781,
+      "step": 6666
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.63685843636448e-05,
+      "loss": 5.7188,
+      "step": 6667
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.636736149631716e-05,
+      "loss": 3.5,
+      "step": 6668
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.636613844369218e-05,
+      "loss": 6.875,
+      "step": 6669
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6364915205783716e-05,
+      "loss": 5.9219,
+      "step": 6670
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6363691782605596e-05,
+      "loss": 5.7969,
+      "step": 6671
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.636246817417168e-05,
+      "loss": 3.9375,
+      "step": 6672
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.636124438049582e-05,
+      "loss": 7.3281,
+      "step": 6673
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.636002040159187e-05,
+      "loss": 8.8438,
+      "step": 6674
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6358796237473694e-05,
+      "loss": 3.0156,
+      "step": 6675
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6357571888155136e-05,
+      "loss": 4.2969,
+      "step": 6676
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.635634735365008e-05,
+      "loss": 4.6094,
+      "step": 6677
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.635512263397236e-05,
+      "loss": 4.2812,
+      "step": 6678
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.635389772913587e-05,
+      "loss": 3.4531,
+      "step": 6679
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.635267263915445e-05,
+      "loss": 5.6875,
+      "step": 6680
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.635144736404199e-05,
+      "loss": 3.6719,
+      "step": 6681
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.635022190381236e-05,
+      "loss": 4.7188,
+      "step": 6682
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.634899625847942e-05,
+      "loss": 4.0781,
+      "step": 6683
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6347770428057054e-05,
+      "loss": 4.4688,
+      "step": 6684
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.634654441255914e-05,
+      "loss": 3.9531,
+      "step": 6685
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.634531821199956e-05,
+      "loss": 2.9375,
+      "step": 6686
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6344091826392196e-05,
+      "loss": 4.3594,
+      "step": 6687
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.634286525575093e-05,
+      "loss": 5.9375,
+      "step": 6688
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.634163850008964e-05,
+      "loss": 4.0469,
+      "step": 6689
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.634041155942224e-05,
+      "loss": 2.5625,
+      "step": 6690
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.633918443376258e-05,
+      "loss": 5.8594,
+      "step": 6691
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.633795712312459e-05,
+      "loss": 2.9062,
+      "step": 6692
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.633672962752214e-05,
+      "loss": 2.4375,
+      "step": 6693
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.633550194696915e-05,
+      "loss": 2.4062,
+      "step": 6694
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.63342740814795e-05,
+      "loss": 2.8438,
+      "step": 6695
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6333046031067087e-05,
+      "loss": 4.25,
+      "step": 6696
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.633181779574583e-05,
+      "loss": 3.1406,
+      "step": 6697
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.633058937552963e-05,
+      "loss": 4.5938,
+      "step": 6698
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6329360770432385e-05,
+      "loss": 2.8594,
+      "step": 6699
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.632813198046801e-05,
+      "loss": 2.7188,
+      "step": 6700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6326903005650415e-05,
+      "loss": 7.8438,
+      "step": 6701
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.632567384599352e-05,
+      "loss": 4.3125,
+      "step": 6702
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6324444501511235e-05,
+      "loss": 2.7812,
+      "step": 6703
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6323214972217474e-05,
+      "loss": 5.625,
+      "step": 6704
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.632198525812617e-05,
+      "loss": 3.9688,
+      "step": 6705
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6320755359251227e-05,
+      "loss": 4.1875,
+      "step": 6706
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.631952527560658e-05,
+      "loss": 1.8594,
+      "step": 6707
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.631829500720615e-05,
+      "loss": 3.7656,
+      "step": 6708
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6317064554063875e-05,
+      "loss": 3.9062,
+      "step": 6709
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6315833916193674e-05,
+      "loss": 3.7188,
+      "step": 6710
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.631460309360948e-05,
+      "loss": 1.4688,
+      "step": 6711
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.6313372086325234e-05,
+      "loss": 4.125,
+      "step": 6712
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6312140894354866e-05,
+      "loss": 5.7969,
+      "step": 6713
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6310909517712314e-05,
+      "loss": 3.625,
+      "step": 6714
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6309677956411524e-05,
+      "loss": 3.4375,
+      "step": 6715
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.630844621046644e-05,
+      "loss": 2.6719,
+      "step": 6716
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6307214279891e-05,
+      "loss": 2.2188,
+      "step": 6717
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.630598216469915e-05,
+      "loss": 5.5938,
+      "step": 6718
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6304749864904853e-05,
+      "loss": 4.4375,
+      "step": 6719
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6303517380522046e-05,
+      "loss": 2.8125,
+      "step": 6720
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.630228471156468e-05,
+      "loss": 3.5938,
+      "step": 6721
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.630105185804672e-05,
+      "loss": 5.3125,
+      "step": 6722
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.629981881998213e-05,
+      "loss": 4.3281,
+      "step": 6723
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.629858559738485e-05,
+      "loss": 5.7812,
+      "step": 6724
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.629735219026884e-05,
+      "loss": 3.875,
+      "step": 6725
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6296118598648086e-05,
+      "loss": 5.7188,
+      "step": 6726
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6294884822536544e-05,
+      "loss": 3.3594,
+      "step": 6727
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.629365086194818e-05,
+      "loss": 2.8438,
+      "step": 6728
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6292416716896955e-05,
+      "loss": 3.3594,
+      "step": 6729
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6291182387396854e-05,
+      "loss": 3.1719,
+      "step": 6730
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.628994787346185e-05,
+      "loss": 3.3906,
+      "step": 6731
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.628871317510591e-05,
+      "loss": 5.25,
+      "step": 6732
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6287478292343014e-05,
+      "loss": 3.6094,
+      "step": 6733
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6286243225187154e-05,
+      "loss": 3.8125,
+      "step": 6734
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6285007973652306e-05,
+      "loss": 3.8281,
+      "step": 6735
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.628377253775245e-05,
+      "loss": 3.2969,
+      "step": 6736
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.628253691750157e-05,
+      "loss": 4.2031,
+      "step": 6737
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.628130111291367e-05,
+      "loss": 4.9844,
+      "step": 6738
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6280065124002726e-05,
+      "loss": 5.3125,
+      "step": 6739
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.627882895078274e-05,
+      "loss": 3.5781,
+      "step": 6740
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.6277592593267695e-05,
+      "loss": 4.6562,
+      "step": 6741
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.62763560514716e-05,
+      "loss": 6.1562,
+      "step": 6742
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.627511932540845e-05,
+      "loss": 6.4375,
+      "step": 6743
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6273882415092246e-05,
+      "loss": 4.3594,
+      "step": 6744
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6272645320536985e-05,
+      "loss": 2.2812,
+      "step": 6745
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.627140804175668e-05,
+      "loss": 4.8125,
+      "step": 6746
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.627017057876534e-05,
+      "loss": 2.25,
+      "step": 6747
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.626893293157697e-05,
+      "loss": 6.0625,
+      "step": 6748
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.626769510020558e-05,
+      "loss": 4.375,
+      "step": 6749
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.62664570846652e-05,
+      "loss": 2.8438,
+      "step": 6750
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6265218884969814e-05,
+      "loss": 3.2031,
+      "step": 6751
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6263980501133466e-05,
+      "loss": 2.4688,
+      "step": 6752
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6262741933170174e-05,
+      "loss": 2.25,
+      "step": 6753
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.626150318109394e-05,
+      "loss": 2.7969,
+      "step": 6754
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6260264244918815e-05,
+      "loss": 2.2188,
+      "step": 6755
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.625902512465881e-05,
+      "loss": 7.3125,
+      "step": 6756
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.625778582032795e-05,
+      "loss": 2.7812,
+      "step": 6757
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.625654633194028e-05,
+      "loss": 3.1875,
+      "step": 6758
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6255306659509815e-05,
+      "loss": 2.625,
+      "step": 6759
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.62540668030506e-05,
+      "loss": 4.2656,
+      "step": 6760
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.625282676257667e-05,
+      "loss": 2.6562,
+      "step": 6761
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.625158653810207e-05,
+      "loss": 2.3672,
+      "step": 6762
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6250346129640825e-05,
+      "loss": 5.3438,
+      "step": 6763
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.624910553720699e-05,
+      "loss": 2.7812,
+      "step": 6764
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6247864760814606e-05,
+      "loss": 8.8906,
+      "step": 6765
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.624662380047772e-05,
+      "loss": 3.9219,
+      "step": 6766
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6245382656210384e-05,
+      "loss": 2.4688,
+      "step": 6767
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.624414132802665e-05,
+      "loss": 3.0312,
+      "step": 6768
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.624289981594056e-05,
+      "loss": 2.7344,
+      "step": 6769
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6241658119966185e-05,
+      "loss": 4.25,
+      "step": 6770
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.6240416240117574e-05,
+      "loss": 2.9688,
+      "step": 6771
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.623917417640878e-05,
+      "loss": 2.2812,
+      "step": 6772
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.623793192885388e-05,
+      "loss": 4.1875,
+      "step": 6773
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6236689497466924e-05,
+      "loss": 3.8281,
+      "step": 6774
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.623544688226199e-05,
+      "loss": 4.9688,
+      "step": 6775
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.623420408325314e-05,
+      "loss": 5.5938,
+      "step": 6776
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.623296110045444e-05,
+      "loss": 2.0625,
+      "step": 6777
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.623171793387997e-05,
+      "loss": 3.4062,
+      "step": 6778
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6230474583543786e-05,
+      "loss": 6.0938,
+      "step": 6779
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.622923104945999e-05,
+      "loss": 3.7969,
+      "step": 6780
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.622798733164264e-05,
+      "loss": 3.3906,
+      "step": 6781
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6226743430105835e-05,
+      "loss": 1.9688,
+      "step": 6782
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6225499344863635e-05,
+      "loss": 3.25,
+      "step": 6783
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.622425507593014e-05,
+      "loss": 5.6562,
+      "step": 6784
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.622301062331944e-05,
+      "loss": 2.2031,
+      "step": 6785
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6221765987045607e-05,
+      "loss": 6.3906,
+      "step": 6786
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.622052116712274e-05,
+      "loss": 2.7656,
+      "step": 6787
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.621927616356494e-05,
+      "loss": 3.9844,
+      "step": 6788
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.621803097638629e-05,
+      "loss": 3.3594,
+      "step": 6789
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.62167856056009e-05,
+      "loss": 3.75,
+      "step": 6790
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.621554005122285e-05,
+      "loss": 3.9062,
+      "step": 6791
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.621429431326625e-05,
+      "loss": 3.8281,
+      "step": 6792
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.621304839174521e-05,
+      "loss": 6.6562,
+      "step": 6793
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.621180228667383e-05,
+      "loss": 3.2969,
+      "step": 6794
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.621055599806622e-05,
+      "loss": 3.5938,
+      "step": 6795
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.620930952593648e-05,
+      "loss": 2.1875,
+      "step": 6796
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.620806287029873e-05,
+      "loss": 3.3594,
+      "step": 6797
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.620681603116708e-05,
+      "loss": 4.4844,
+      "step": 6798
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6205569008555656e-05,
+      "loss": 5.2812,
+      "step": 6799
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.620432180247856e-05,
+      "loss": 2.1719,
+      "step": 6800
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.620307441294992e-05,
+      "loss": 3.625,
+      "step": 6801
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6201826839983865e-05,
+      "loss": 2.6094,
+      "step": 6802
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.6200579083594504e-05,
+      "loss": 4.4062,
+      "step": 6803
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6199331143795975e-05,
+      "loss": 4.2812,
+      "step": 6804
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.619808302060239e-05,
+      "loss": 3.7188,
+      "step": 6805
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.61968347140279e-05,
+      "loss": 2.9688,
+      "step": 6806
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.619558622408662e-05,
+      "loss": 2.0,
+      "step": 6807
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.61943375507927e-05,
+      "loss": 4.0938,
+      "step": 6808
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.619308869416027e-05,
+      "loss": 3.875,
+      "step": 6809
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.619183965420346e-05,
+      "loss": 5.2812,
+      "step": 6810
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.619059043093642e-05,
+      "loss": 5.8281,
+      "step": 6811
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6189341024373286e-05,
+      "loss": 2.9062,
+      "step": 6812
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.618809143452821e-05,
+      "loss": 2.4688,
+      "step": 6813
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6186841661415336e-05,
+      "loss": 5.1875,
+      "step": 6814
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.618559170504881e-05,
+      "loss": 4.0312,
+      "step": 6815
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.618434156544279e-05,
+      "loss": 3.0312,
+      "step": 6816
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.618309124261142e-05,
+      "loss": 3.875,
+      "step": 6817
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.618184073656886e-05,
+      "loss": 2.2344,
+      "step": 6818
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.618059004732927e-05,
+      "loss": 4.875,
+      "step": 6819
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.61793391749068e-05,
+      "loss": 6.7188,
+      "step": 6820
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.617808811931562e-05,
+      "loss": 4.75,
+      "step": 6821
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.617683688056989e-05,
+      "loss": 4.5156,
+      "step": 6822
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.617558545868378e-05,
+      "loss": 2.2344,
+      "step": 6823
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.617433385367145e-05,
+      "loss": 3.9531,
+      "step": 6824
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6173082065547074e-05,
+      "loss": 4.8438,
+      "step": 6825
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6171830094324824e-05,
+      "loss": 4.125,
+      "step": 6826
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6170577940018875e-05,
+      "loss": 3.9688,
+      "step": 6827
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.61693256026434e-05,
+      "loss": 3.125,
+      "step": 6828
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.616807308221257e-05,
+      "loss": 3.625,
+      "step": 6829
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.616682037874057e-05,
+      "loss": 1.2344,
+      "step": 6830
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.61655674922416e-05,
+      "loss": 8.4375,
+      "step": 6831
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.616431442272981e-05,
+      "loss": 2.7656,
+      "step": 6832
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6163061170219415e-05,
+      "loss": 4.2344,
+      "step": 6833
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.6161807734724594e-05,
+      "loss": 3.1875,
+      "step": 6834
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.6160554116259536e-05,
+      "loss": 1.75,
+      "step": 6835
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.6159300314838424e-05,
+      "loss": 3.0312,
+      "step": 6836
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.615804633047547e-05,
+      "loss": 3.375,
+      "step": 6837
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.615679216318486e-05,
+      "loss": 4.0625,
+      "step": 6838
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.6155537812980804e-05,
+      "loss": 4.0,
+      "step": 6839
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.615428327987749e-05,
+      "loss": 3.8125,
+      "step": 6840
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.615302856388911e-05,
+      "loss": 4.1562,
+      "step": 6841
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.61517736650299e-05,
+      "loss": 3.1406,
+      "step": 6842
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.6150518583314045e-05,
+      "loss": 2.875,
+      "step": 6843
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.614926331875576e-05,
+      "loss": 2.1562,
+      "step": 6844
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.6148007871369253e-05,
+      "loss": 3.4062,
+      "step": 6845
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.6146752241168745e-05,
+      "loss": 2.5781,
+      "step": 6846
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.614549642816845e-05,
+      "loss": 4.5312,
+      "step": 6847
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.6144240432382565e-05,
+      "loss": 3.5625,
+      "step": 6848
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.6142984253825334e-05,
+      "loss": 4.1875,
+      "step": 6849
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.614172789251097e-05,
+      "loss": 3.8125,
+      "step": 6850
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.61404713484537e-05,
+      "loss": 4.5938,
+      "step": 6851
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.613921462166774e-05,
+      "loss": 2.5938,
+      "step": 6852
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.613795771216732e-05,
+      "loss": 4.5,
+      "step": 6853
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.613670061996668e-05,
+      "loss": 2.8906,
+      "step": 6854
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.613544334508004e-05,
+      "loss": 2.3125,
+      "step": 6855
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.613418588752165e-05,
+      "loss": 2.7969,
+      "step": 6856
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.6132928247305713e-05,
+      "loss": 4.7812,
+      "step": 6857
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.6131670424446505e-05,
+      "loss": 3.4688,
+      "step": 6858
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.6130412418958245e-05,
+      "loss": 3.875,
+      "step": 6859
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.612915423085518e-05,
+      "loss": 3.3594,
+      "step": 6860
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.612789586015155e-05,
+      "loss": 2.7188,
+      "step": 6861
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.61266373068616e-05,
+      "loss": 2.9062,
+      "step": 6862
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.612537857099959e-05,
+      "loss": 1.75,
+      "step": 6863
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.612411965257977e-05,
+      "loss": 2.5625,
+      "step": 6864
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.612286055161637e-05,
+      "loss": 3.9375,
+      "step": 6865
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.612160126812367e-05,
+      "loss": 4.4688,
+      "step": 6866
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.612034180211592e-05,
+      "loss": 2.8438,
+      "step": 6867
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.611908215360737e-05,
+      "loss": 4.2031,
+      "step": 6868
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.611782232261228e-05,
+      "loss": 6.5312,
+      "step": 6869
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.611656230914493e-05,
+      "loss": 4.9688,
+      "step": 6870
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.611530211321957e-05,
+      "loss": 4.5625,
+      "step": 6871
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.6114041734850466e-05,
+      "loss": 3.7188,
+      "step": 6872
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.611278117405189e-05,
+      "loss": 7.6719,
+      "step": 6873
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.611152043083812e-05,
+      "loss": 3.1406,
+      "step": 6874
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.611025950522342e-05,
+      "loss": 4.5312,
+      "step": 6875
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.610899839722207e-05,
+      "loss": 4.2656,
+      "step": 6876
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.610773710684834e-05,
+      "loss": 3.8594,
+      "step": 6877
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.610647563411653e-05,
+      "loss": 3.8438,
+      "step": 6878
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.610521397904089e-05,
+      "loss": 4.7656,
+      "step": 6879
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.610395214163573e-05,
+      "loss": 6.0469,
+      "step": 6880
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.610269012191531e-05,
+      "loss": 2.2812,
+      "step": 6881
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.610142791989395e-05,
+      "loss": 3.3281,
+      "step": 6882
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.6100165535585913e-05,
+      "loss": 3.9062,
+      "step": 6883
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.609890296900551e-05,
+      "loss": 3.5312,
+      "step": 6884
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.609764022016701e-05,
+      "loss": 3.375,
+      "step": 6885
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.609637728908473e-05,
+      "loss": 4.0469,
+      "step": 6886
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.609511417577296e-05,
+      "loss": 4.5312,
+      "step": 6887
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.6093850880246e-05,
+      "loss": 3.9375,
+      "step": 6888
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.609258740251815e-05,
+      "loss": 3.4844,
+      "step": 6889
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.6091323742603726e-05,
+      "loss": 4.2344,
+      "step": 6890
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.609005990051702e-05,
+      "loss": 5.25,
+      "step": 6891
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.608879587627235e-05,
+      "loss": 4.2656,
+      "step": 6892
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.608753166988401e-05,
+      "loss": 4.25,
+      "step": 6893
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.608626728136633e-05,
+      "loss": 2.4375,
+      "step": 6894
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.6085002710733626e-05,
+      "loss": 4.8438,
+      "step": 6895
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.6083737958000195e-05,
+      "loss": 3.9844,
+      "step": 6896
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.608247302318037e-05,
+      "loss": 4.7188,
+      "step": 6897
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.608120790628847e-05,
+      "loss": 2.5625,
+      "step": 6898
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.607994260733881e-05,
+      "loss": 3.4062,
+      "step": 6899
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.6078677126345726e-05,
+      "loss": 2.375,
+      "step": 6900
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.6077411463323544e-05,
+      "loss": 5.2188,
+      "step": 6901
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.607614561828658e-05,
+      "loss": 6.2188,
+      "step": 6902
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.6074879591249175e-05,
+      "loss": 3.75,
+      "step": 6903
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.607361338222566e-05,
+      "loss": 3.6719,
+      "step": 6904
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.607234699123037e-05,
+      "loss": 3.625,
+      "step": 6905
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.607108041827764e-05,
+      "loss": 2.5938,
+      "step": 6906
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.606981366338182e-05,
+      "loss": 5.0156,
+      "step": 6907
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.606854672655723e-05,
+      "loss": 2.9375,
+      "step": 6908
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.606727960781823e-05,
+      "loss": 2.75,
+      "step": 6909
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.606601230717916e-05,
+      "loss": 2.5312,
+      "step": 6910
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.6064744824654365e-05,
+      "loss": 2.2188,
+      "step": 6911
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.6063477160258195e-05,
+      "loss": 5.0312,
+      "step": 6912
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.6062209314005007e-05,
+      "loss": 4.1719,
+      "step": 6913
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.606094128590915e-05,
+      "loss": 2.3906,
+      "step": 6914
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.6059673075984984e-05,
+      "loss": 2.75,
+      "step": 6915
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.605840468424686e-05,
+      "loss": 1.875,
+      "step": 6916
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.605713611070914e-05,
+      "loss": 3.0938,
+      "step": 6917
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.605586735538618e-05,
+      "loss": 3.4375,
+      "step": 6918
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.605459841829236e-05,
+      "loss": 3.125,
+      "step": 6919
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.605332929944203e-05,
+      "loss": 4.875,
+      "step": 6920
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.605205999884957e-05,
+      "loss": 4.75,
+      "step": 6921
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.605079051652934e-05,
+      "loss": 2.9688,
+      "step": 6922
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.6049520852495716e-05,
+      "loss": 3.0,
+      "step": 6923
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.604825100676307e-05,
+      "loss": 8.0312,
+      "step": 6924
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.604698097934578e-05,
+      "loss": 4.0156,
+      "step": 6925
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.604571077025823e-05,
+      "loss": 4.6562,
+      "step": 6926
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.6044440379514785e-05,
+      "loss": 3.4062,
+      "step": 6927
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.6043169807129844e-05,
+      "loss": 4.8125,
+      "step": 6928
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.6041899053117784e-05,
+      "loss": 4.4688,
+      "step": 6929
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.604062811749299e-05,
+      "loss": 3.5625,
+      "step": 6930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.6039357000269846e-05,
+      "loss": 3.8438,
+      "step": 6931
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.6038085701462755e-05,
+      "loss": 2.8125,
+      "step": 6932
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.6036814221086094e-05,
+      "loss": 5.0156,
+      "step": 6933
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.603554255915428e-05,
+      "loss": 4.6875,
+      "step": 6934
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.6034270715681684e-05,
+      "loss": 3.625,
+      "step": 6935
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.603299869068273e-05,
+      "loss": 4.4062,
+      "step": 6936
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.6031726484171785e-05,
+      "loss": 6.2969,
+      "step": 6937
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.603045409616329e-05,
+      "loss": 4.0312,
+      "step": 6938
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.602918152667163e-05,
+      "loss": 4.1562,
+      "step": 6939
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.6027908775711204e-05,
+      "loss": 4.4062,
+      "step": 6940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.602663584329644e-05,
+      "loss": 1.7812,
+      "step": 6941
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.602536272944174e-05,
+      "loss": 4.2188,
+      "step": 6942
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.602408943416152e-05,
+      "loss": 4.9844,
+      "step": 6943
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.6022815957470185e-05,
+      "loss": 4.0625,
+      "step": 6944
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.602154229938217e-05,
+      "loss": 3.25,
+      "step": 6945
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.602026845991188e-05,
+      "loss": 3.5938,
+      "step": 6946
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.601899443907373e-05,
+      "loss": 3.1562,
+      "step": 6947
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.601772023688216e-05,
+      "loss": 4.375,
+      "step": 6948
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.6016445853351596e-05,
+      "loss": 5.5469,
+      "step": 6949
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.6015171288496455e-05,
+      "loss": 3.1875,
+      "step": 6950
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.601389654233117e-05,
+      "loss": 3.7656,
+      "step": 6951
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.6012621614870173e-05,
+      "loss": 3.1719,
+      "step": 6952
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.60113465061279e-05,
+      "loss": 4.8438,
+      "step": 6953
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.601007121611878e-05,
+      "loss": 5.3438,
+      "step": 6954
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.600879574485725e-05,
+      "loss": 7.0469,
+      "step": 6955
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.600752009235777e-05,
+      "loss": 3.0625,
+      "step": 6956
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.6006244258634756e-05,
+      "loss": 1.9219,
+      "step": 6957
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.6004968243702666e-05,
+      "loss": 4.5,
+      "step": 6958
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.600369204757594e-05,
+      "loss": 3.2656,
+      "step": 6959
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.600241567026902e-05,
+      "loss": 3.0938,
+      "step": 6960
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.600113911179638e-05,
+      "loss": 3.5,
+      "step": 6961
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.599986237217245e-05,
+      "loss": 9.4375,
+      "step": 6962
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.599858545141169e-05,
+      "loss": 5.625,
+      "step": 6963
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.5997308349528556e-05,
+      "loss": 2.9688,
+      "step": 6964
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.599603106653751e-05,
+      "loss": 2.6875,
+      "step": 6965
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.599475360245301e-05,
+      "loss": 5.25,
+      "step": 6966
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.599347595728952e-05,
+      "loss": 5.6562,
+      "step": 6967
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.59921981310615e-05,
+      "loss": 3.6406,
+      "step": 6968
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.599092012378342e-05,
+      "loss": 4.2656,
+      "step": 6969
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.5989641935469744e-05,
+      "loss": 3.2188,
+      "step": 6970
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.598836356613495e-05,
+      "loss": 4.4375,
+      "step": 6971
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.5987085015793504e-05,
+      "loss": 4.5156,
+      "step": 6972
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.598580628445988e-05,
+      "loss": 5.5469,
+      "step": 6973
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.598452737214857e-05,
+      "loss": 3.625,
+      "step": 6974
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.598324827887403e-05,
+      "loss": 3.7188,
+      "step": 6975
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.598196900465075e-05,
+      "loss": 3.9688,
+      "step": 6976
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.598068954949321e-05,
+      "loss": 3.0938,
+      "step": 6977
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.597940991341591e-05,
+      "loss": 2.5938,
+      "step": 6978
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.597813009643333e-05,
+      "loss": 4.125,
+      "step": 6979
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.597685009855994e-05,
+      "loss": 4.0938,
+      "step": 6980
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.597556991981026e-05,
+      "loss": 4.0,
+      "step": 6981
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.597428956019876e-05,
+      "loss": 3.0469,
+      "step": 6982
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.597300901973995e-05,
+      "loss": 3.6875,
+      "step": 6983
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.597172829844832e-05,
+      "loss": 4.125,
+      "step": 6984
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.597044739633836e-05,
+      "loss": 3.6562,
+      "step": 6985
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.59691663134246e-05,
+      "loss": 5.0625,
+      "step": 6986
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.596788504972152e-05,
+      "loss": 5.3594,
+      "step": 6987
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.596660360524363e-05,
+      "loss": 3.0469,
+      "step": 6988
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.5965321980005434e-05,
+      "loss": 5.3594,
+      "step": 6989
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.5964040174021455e-05,
+      "loss": 3.9062,
+      "step": 6990
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.596275818730619e-05,
+      "loss": 4.6562,
+      "step": 6991
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.5961476019874156e-05,
+      "loss": 4.1406,
+      "step": 6992
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.596019367173987e-05,
+      "loss": 2.3438,
+      "step": 6993
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.595891114291785e-05,
+      "loss": 3.9219,
+      "step": 6994
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.595762843342262e-05,
+      "loss": 5.8125,
+      "step": 6995
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.5956345543268694e-05,
+      "loss": 4.9375,
+      "step": 6996
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.59550624724706e-05,
+      "loss": 4.5469,
+      "step": 6997
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.595377922104287e-05,
+      "loss": 3.5938,
+      "step": 6998
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.5952495789000014e-05,
+      "loss": 5.5938,
+      "step": 6999
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.595121217635658e-05,
+      "loss": 3.0625,
+      "step": 7000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.5949928383127086e-05,
+      "loss": 2.4688,
+      "step": 7001
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.594864440932608e-05,
+      "loss": 3.8438,
+      "step": 7002
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.594736025496809e-05,
+      "loss": 2.75,
+      "step": 7003
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.5946075920067655e-05,
+      "loss": 4.375,
+      "step": 7004
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.594479140463931e-05,
+      "loss": 4.8438,
+      "step": 7005
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.594350670869761e-05,
+      "loss": 5.9688,
+      "step": 7006
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.594222183225709e-05,
+      "loss": 4.0312,
+      "step": 7007
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.594093677533229e-05,
+      "loss": 3.6562,
+      "step": 7008
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.5939651537937774e-05,
+      "loss": 4.5156,
+      "step": 7009
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.5938366120088084e-05,
+      "loss": 3.3125,
+      "step": 7010
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.593708052179777e-05,
+      "loss": 3.5312,
+      "step": 7011
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.593579474308139e-05,
+      "loss": 2.9531,
+      "step": 7012
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.5934508783953504e-05,
+      "loss": 4.25,
+      "step": 7013
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.593322264442867e-05,
+      "loss": 4.0938,
+      "step": 7014
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.5931936324521435e-05,
+      "loss": 4.1406,
+      "step": 7015
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.593064982424638e-05,
+      "loss": 2.6875,
+      "step": 7016
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.5929363143618055e-05,
+      "loss": 3.5469,
+      "step": 7017
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.592807628265103e-05,
+      "loss": 2.6406,
+      "step": 7018
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.592678924135989e-05,
+      "loss": 3.5938,
+      "step": 7019
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.5925502019759184e-05,
+      "loss": 3.7969,
+      "step": 7020
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.592421461786349e-05,
+      "loss": 2.625,
+      "step": 7021
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.592292703568739e-05,
+      "loss": 5.4688,
+      "step": 7022
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.592163927324546e-05,
+      "loss": 6.625,
+      "step": 7023
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.592035133055228e-05,
+      "loss": 3.6875,
+      "step": 7024
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.591906320762242e-05,
+      "loss": 4.625,
+      "step": 7025
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.591777490447047e-05,
+      "loss": 2.9688,
+      "step": 7026
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.591648642111102e-05,
+      "loss": 6.9375,
+      "step": 7027
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.5915197757558644e-05,
+      "loss": 6.3438,
+      "step": 7028
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.591390891382795e-05,
+      "loss": 4.1562,
+      "step": 7029
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.591261988993351e-05,
+      "loss": 3.3906,
+      "step": 7030
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.591133068588993e-05,
+      "loss": 2.0625,
+      "step": 7031
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.59100413017118e-05,
+      "loss": 2.5625,
+      "step": 7032
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.590875173741372e-05,
+      "loss": 4.25,
+      "step": 7033
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.590746199301029e-05,
+      "loss": 4.5938,
+      "step": 7034
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.590617206851611e-05,
+      "loss": 3.1875,
+      "step": 7035
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.590488196394578e-05,
+      "loss": 4.0156,
+      "step": 7036
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.59035916793139e-05,
+      "loss": 4.4844,
+      "step": 7037
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.59023012146351e-05,
+      "loss": 6.4219,
+      "step": 7038
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.5901010569923965e-05,
+      "loss": 5.375,
+      "step": 7039
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.589971974519512e-05,
+      "loss": 6.1875,
+      "step": 7040
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.589842874046318e-05,
+      "loss": 4.5,
+      "step": 7041
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.589713755574275e-05,
+      "loss": 6.5469,
+      "step": 7042
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.589584619104845e-05,
+      "loss": 2.7969,
+      "step": 7043
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.589455464639492e-05,
+      "loss": 4.3438,
+      "step": 7044
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.589326292179675e-05,
+      "loss": 3.7656,
+      "step": 7045
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.589197101726858e-05,
+      "loss": 3.4062,
+      "step": 7046
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.5890678932825043e-05,
+      "loss": 5.0781,
+      "step": 7047
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.588938666848076e-05,
+      "loss": 3.6875,
+      "step": 7048
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.588809422425035e-05,
+      "loss": 2.7188,
+      "step": 7049
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.588680160014845e-05,
+      "loss": 3.6875,
+      "step": 7050
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.5885508796189716e-05,
+      "loss": 2.1719,
+      "step": 7051
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.5884215812388756e-05,
+      "loss": 5.625,
+      "step": 7052
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.588292264876022e-05,
+      "loss": 5.8438,
+      "step": 7053
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.588162930531874e-05,
+      "loss": 3.7812,
+      "step": 7054
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.588033578207898e-05,
+      "loss": 3.8438,
+      "step": 7055
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.5879042079055565e-05,
+      "loss": 2.7344,
+      "step": 7056
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.587774819626314e-05,
+      "loss": 4.625,
+      "step": 7057
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.5876454133716364e-05,
+      "loss": 4.75,
+      "step": 7058
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.587515989142988e-05,
+      "loss": 1.5625,
+      "step": 7059
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.587386546941834e-05,
+      "loss": 3.7812,
+      "step": 7060
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.58725708676964e-05,
+      "loss": 3.0312,
+      "step": 7061
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.587127608627872e-05,
+      "loss": 9.0156,
+      "step": 7062
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.586998112517996e-05,
+      "loss": 4.4062,
+      "step": 7063
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.586868598441477e-05,
+      "loss": 3.3125,
+      "step": 7064
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.5867390663997815e-05,
+      "loss": 3.375,
+      "step": 7065
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.586609516394377e-05,
+      "loss": 3.2188,
+      "step": 7066
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.586479948426728e-05,
+      "loss": 7.7031,
+      "step": 7067
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.5863503624983045e-05,
+      "loss": 2.2188,
+      "step": 7068
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.586220758610571e-05,
+      "loss": 2.1094,
+      "step": 7069
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.586091136764996e-05,
+      "loss": 3.1094,
+      "step": 7070
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.5859614969630466e-05,
+      "loss": 2.6406,
+      "step": 7071
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.5858318392061896e-05,
+      "loss": 3.0781,
+      "step": 7072
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.5857021634958947e-05,
+      "loss": 4.1719,
+      "step": 7073
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.585572469833629e-05,
+      "loss": 7.0938,
+      "step": 7074
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.5854427582208605e-05,
+      "loss": 5.0,
+      "step": 7075
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.585313028659058e-05,
+      "loss": 5.4844,
+      "step": 7076
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.5851832811496895e-05,
+      "loss": 3.0,
+      "step": 7077
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.5850535156942256e-05,
+      "loss": 2.0625,
+      "step": 7078
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.584923732294134e-05,
+      "loss": 4.8438,
+      "step": 7079
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.584793930950884e-05,
+      "loss": 4.6562,
+      "step": 7080
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.5846641116659465e-05,
+      "loss": 4.9062,
+      "step": 7081
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.5845342744407894e-05,
+      "loss": 2.2812,
+      "step": 7082
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.584404419276883e-05,
+      "loss": 3.625,
+      "step": 7083
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.5842745461756973e-05,
+      "loss": 2.9375,
+      "step": 7084
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.5841446551387045e-05,
+      "loss": 3.2812,
+      "step": 7085
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.5840147461673734e-05,
+      "loss": 2.9062,
+      "step": 7086
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.583884819263174e-05,
+      "loss": 4.0938,
+      "step": 7087
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.58375487442758e-05,
+      "loss": 4.9688,
+      "step": 7088
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.5836249116620596e-05,
+      "loss": 6.4375,
+      "step": 7089
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.583494930968085e-05,
+      "loss": 5.2812,
+      "step": 7090
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.583364932347128e-05,
+      "loss": 3.0938,
+      "step": 7091
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.583234915800662e-05,
+      "loss": 3.3438,
+      "step": 7092
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.5831048813301555e-05,
+      "loss": 4.2969,
+      "step": 7093
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.582974828937083e-05,
+      "loss": 4.4375,
+      "step": 7094
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.582844758622916e-05,
+      "loss": 4.7344,
+      "step": 7095
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.582714670389128e-05,
+      "loss": 2.625,
+      "step": 7096
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.5825845642371906e-05,
+      "loss": 4.5938,
+      "step": 7097
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.582454440168577e-05,
+      "loss": 4.5625,
+      "step": 7098
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.582324298184761e-05,
+      "loss": 7.7188,
+      "step": 7099
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.582194138287215e-05,
+      "loss": 3.0938,
+      "step": 7100
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.582063960477414e-05,
+      "loss": 3.8906,
+      "step": 7101
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.58193376475683e-05,
+      "loss": 4.2656,
+      "step": 7102
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.581803551126938e-05,
+      "loss": 3.25,
+      "step": 7103
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.581673319589212e-05,
+      "loss": 3.1875,
+      "step": 7104
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.5815430701451265e-05,
+      "loss": 6.6484,
+      "step": 7105
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.5814128027961564e-05,
+      "loss": 3.8125,
+      "step": 7106
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.5812825175437754e-05,
+      "loss": 2.6406,
+      "step": 7107
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.5811522143894595e-05,
+      "loss": 3.125,
+      "step": 7108
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.5810218933346825e-05,
+      "loss": 4.1562,
+      "step": 7109
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.5808915543809216e-05,
+      "loss": 3.7969,
+      "step": 7110
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.580761197529651e-05,
+      "loss": 3.4062,
+      "step": 7111
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.5806308227823475e-05,
+      "loss": 2.8906,
+      "step": 7112
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.580500430140487e-05,
+      "loss": 5.0938,
+      "step": 7113
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.580370019605544e-05,
+      "loss": 4.125,
+      "step": 7114
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.580239591178997e-05,
+      "loss": 4.7656,
+      "step": 7115
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.580109144862323e-05,
+      "loss": 8.5625,
+      "step": 7116
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.579978680656996e-05,
+      "loss": 4.7188,
+      "step": 7117
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.579848198564495e-05,
+      "loss": 3.4688,
+      "step": 7118
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.5797176985862966e-05,
+      "loss": 3.5625,
+      "step": 7119
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.579587180723879e-05,
+      "loss": 8.125,
+      "step": 7120
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.5794566449787185e-05,
+      "loss": 3.2188,
+      "step": 7121
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.5793260913522945e-05,
+      "loss": 3.7656,
+      "step": 7122
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.579195519846084e-05,
+      "loss": 4.1094,
+      "step": 7123
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.579064930461565e-05,
+      "loss": 2.3125,
+      "step": 7124
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.5789343232002166e-05,
+      "loss": 6.1562,
+      "step": 7125
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.5788036980635165e-05,
+      "loss": 4.3125,
+      "step": 7126
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.5786730550529446e-05,
+      "loss": 3.9062,
+      "step": 7127
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.57854239416998e-05,
+      "loss": 6.75,
+      "step": 7128
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.578411715416101e-05,
+      "loss": 2.6562,
+      "step": 7129
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.578281018792788e-05,
+      "loss": 3.5938,
+      "step": 7130
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.578150304301519e-05,
+      "loss": 3.0625,
+      "step": 7131
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.5780195719437754e-05,
+      "loss": 4.8281,
+      "step": 7132
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.577888821721037e-05,
+      "loss": 3.3125,
+      "step": 7133
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.5777580536347835e-05,
+      "loss": 3.9062,
+      "step": 7134
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.577627267686496e-05,
+      "loss": 3.0781,
+      "step": 7135
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.5774964638776545e-05,
+      "loss": 3.9219,
+      "step": 7136
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.577365642209741e-05,
+      "loss": 3.6875,
+      "step": 7137
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.5772348026842346e-05,
+      "loss": 3.7969,
+      "step": 7138
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.577103945302619e-05,
+      "loss": 4.2031,
+      "step": 7139
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.576973070066373e-05,
+      "loss": 2.3906,
+      "step": 7140
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.57684217697698e-05,
+      "loss": 4.0938,
+      "step": 7141
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.5767112660359216e-05,
+      "loss": 3.2812,
+      "step": 7142
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.5765803372446794e-05,
+      "loss": 4.625,
+      "step": 7143
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.576449390604736e-05,
+      "loss": 3.0469,
+      "step": 7144
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.576318426117574e-05,
+      "loss": 2.3906,
+      "step": 7145
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.576187443784676e-05,
+      "loss": 4.6562,
+      "step": 7146
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.576056443607525e-05,
+      "loss": 2.8906,
+      "step": 7147
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.575925425587603e-05,
+      "loss": 4.4688,
+      "step": 7148
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.575794389726395e-05,
+      "loss": 3.625,
+      "step": 7149
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.575663336025383e-05,
+      "loss": 3.125,
+      "step": 7150
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.575532264486052e-05,
+      "loss": 4.9375,
+      "step": 7151
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.575401175109884e-05,
+      "loss": 4.0625,
+      "step": 7152
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.575270067898365e-05,
+      "loss": 3.6562,
+      "step": 7153
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.575138942852979e-05,
+      "loss": 3.0938,
+      "step": 7154
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.575007799975209e-05,
+      "loss": 5.4844,
+      "step": 7155
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.574876639266541e-05,
+      "loss": 2.625,
+      "step": 7156
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.57474546072846e-05,
+      "loss": 3.0938,
+      "step": 7157
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.5746142643624505e-05,
+      "loss": 2.5938,
+      "step": 7158
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.574483050169998e-05,
+      "loss": 5.8438,
+      "step": 7159
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.574351818152588e-05,
+      "loss": 2.375,
+      "step": 7160
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.5742205683117054e-05,
+      "loss": 5.2188,
+      "step": 7161
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.574089300648838e-05,
+      "loss": 2.5,
+      "step": 7162
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.57395801516547e-05,
+      "loss": 6.7812,
+      "step": 7163
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.573826711863089e-05,
+      "loss": 4.3438,
+      "step": 7164
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.57369539074318e-05,
+      "loss": 5.6875,
+      "step": 7165
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.573564051807232e-05,
+      "loss": 4.2656,
+      "step": 7166
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.5734326950567294e-05,
+      "loss": 5.5312,
+      "step": 7167
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.573301320493161e-05,
+      "loss": 4.125,
+      "step": 7168
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.573169928118014e-05,
+      "loss": 4.4688,
+      "step": 7169
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.573038517932775e-05,
+      "loss": 4.9688,
+      "step": 7170
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.572907089938932e-05,
+      "loss": 2.9062,
+      "step": 7171
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.572775644137974e-05,
+      "loss": 5.5234,
+      "step": 7172
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.572644180531388e-05,
+      "loss": 4.3438,
+      "step": 7173
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.572512699120662e-05,
+      "loss": 5.7188,
+      "step": 7174
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.5723811999072856e-05,
+      "loss": 5.7812,
+      "step": 7175
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.5722496828927474e-05,
+      "loss": 4.3125,
+      "step": 7176
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.5721181480785355e-05,
+      "loss": 3.2031,
+      "step": 7177
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.5719865954661386e-05,
+      "loss": 2.5,
+      "step": 7178
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.571855025057048e-05,
+      "loss": 5.375,
+      "step": 7179
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.571723436852752e-05,
+      "loss": 4.6562,
+      "step": 7180
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.571591830854741e-05,
+      "loss": 4.0938,
+      "step": 7181
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.571460207064503e-05,
+      "loss": 2.625,
+      "step": 7182
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.5713285654835305e-05,
+      "loss": 5.2344,
+      "step": 7183
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.571196906113313e-05,
+      "loss": 4.6875,
+      "step": 7184
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.57106522895534e-05,
+      "loss": 3.1562,
+      "step": 7185
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.570933534011104e-05,
+      "loss": 4.0,
+      "step": 7186
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.5708018212820945e-05,
+      "loss": 3.2969,
+      "step": 7187
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.570670090769803e-05,
+      "loss": 3.0,
+      "step": 7188
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.570538342475722e-05,
+      "loss": 1.1875,
+      "step": 7189
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.570406576401341e-05,
+      "loss": 2.9844,
+      "step": 7190
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.570274792548154e-05,
+      "loss": 2.1875,
+      "step": 7191
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.570142990917651e-05,
+      "loss": 4.4844,
+      "step": 7192
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.5700111715113255e-05,
+      "loss": 2.5625,
+      "step": 7193
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.569879334330669e-05,
+      "loss": 3.1562,
+      "step": 7194
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.5697474793771736e-05,
+      "loss": 3.7578,
+      "step": 7195
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.569615606652334e-05,
+      "loss": 3.0625,
+      "step": 7196
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.569483716157641e-05,
+      "loss": 2.9375,
+      "step": 7197
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.569351807894589e-05,
+      "loss": 3.0312,
+      "step": 7198
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.569219881864671e-05,
+      "loss": 4.4062,
+      "step": 7199
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.5690879380693805e-05,
+      "loss": 4.8594,
+      "step": 7200
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.568955976510211e-05,
+      "loss": 7.8281,
+      "step": 7201
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.568823997188658e-05,
+      "loss": 3.25,
+      "step": 7202
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.568692000106213e-05,
+      "loss": 4.0781,
+      "step": 7203
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.5685599852643734e-05,
+      "loss": 6.8125,
+      "step": 7204
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.568427952664631e-05,
+      "loss": 2.5312,
+      "step": 7205
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.568295902308482e-05,
+      "loss": 4.6875,
+      "step": 7206
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.568163834197421e-05,
+      "loss": 2.8125,
+      "step": 7207
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.568031748332943e-05,
+      "loss": 2.75,
+      "step": 7208
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.567899644716544e-05,
+      "loss": 6.1875,
+      "step": 7209
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.567767523349719e-05,
+      "loss": 2.8438,
+      "step": 7210
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.567635384233965e-05,
+      "loss": 6.3125,
+      "step": 7211
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.5675032273707754e-05,
+      "loss": 4.8125,
+      "step": 7212
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.567371052761649e-05,
+      "loss": 3.5625,
+      "step": 7213
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.56723886040808e-05,
+      "loss": 2.0,
+      "step": 7214
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.567106650311567e-05,
+      "loss": 2.7969,
+      "step": 7215
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.566974422473605e-05,
+      "loss": 4.0938,
+      "step": 7216
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.566842176895692e-05,
+      "loss": 3.7344,
+      "step": 7217
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.566709913579325e-05,
+      "loss": 5.9688,
+      "step": 7218
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.5665776325260016e-05,
+      "loss": 4.0938,
+      "step": 7219
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.566445333737219e-05,
+      "loss": 2.0625,
+      "step": 7220
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.5663130172144745e-05,
+      "loss": 3.3906,
+      "step": 7221
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.5661806829592675e-05,
+      "loss": 4.7188,
+      "step": 7222
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.566048330973095e-05,
+      "loss": 2.4062,
+      "step": 7223
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.565915961257455e-05,
+      "loss": 4.0312,
+      "step": 7224
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.565783573813848e-05,
+      "loss": 3.2812,
+      "step": 7225
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.56565116864377e-05,
+      "loss": 3.3125,
+      "step": 7226
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.5655187457487226e-05,
+      "loss": 4.4375,
+      "step": 7227
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.5653863051302034e-05,
+      "loss": 3.2812,
+      "step": 7228
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.5652538467897126e-05,
+      "loss": 2.0938,
+      "step": 7229
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.56512137072875e-05,
+      "loss": 2.7812,
+      "step": 7230
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.564988876948814e-05,
+      "loss": 3.625,
+      "step": 7231
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.564856365451405e-05,
+      "loss": 2.7656,
+      "step": 7232
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.564723836238025e-05,
+      "loss": 3.6875,
+      "step": 7233
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.564591289310172e-05,
+      "loss": 6.7656,
+      "step": 7234
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.564458724669347e-05,
+      "loss": 3.4062,
+      "step": 7235
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.564326142317052e-05,
+      "loss": 2.9375,
+      "step": 7236
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.564193542254788e-05,
+      "loss": 1.7812,
+      "step": 7237
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.5640609244840556e-05,
+      "loss": 3.3906,
+      "step": 7238
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.563928289006355e-05,
+      "loss": 3.4531,
+      "step": 7239
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.563795635823189e-05,
+      "loss": 3.9062,
+      "step": 7240
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.56366296493606e-05,
+      "loss": 4.4375,
+      "step": 7241
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.563530276346469e-05,
+      "loss": 5.4688,
+      "step": 7242
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.563397570055918e-05,
+      "loss": 4.9219,
+      "step": 7243
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.5632648460659105e-05,
+      "loss": 7.125,
+      "step": 7244
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.563132104377948e-05,
+      "loss": 3.4375,
+      "step": 7245
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.562999344993534e-05,
+      "loss": 3.5781,
+      "step": 7246
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.562866567914171e-05,
+      "loss": 6.3125,
+      "step": 7247
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.5627337731413626e-05,
+      "loss": 3.1406,
+      "step": 7248
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.562600960676612e-05,
+      "loss": 4.3594,
+      "step": 7249
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.562468130521422e-05,
+      "loss": 3.3281,
+      "step": 7250
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.562335282677298e-05,
+      "loss": 4.7812,
+      "step": 7251
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.562202417145743e-05,
+      "loss": 5.2188,
+      "step": 7252
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.5620695339282614e-05,
+      "loss": 5.3438,
+      "step": 7253
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.5619366330263574e-05,
+      "loss": 4.0938,
+      "step": 7254
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.561803714441536e-05,
+      "loss": 4.2969,
+      "step": 7255
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.561670778175302e-05,
+      "loss": 4.2188,
+      "step": 7256
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.56153782422916e-05,
+      "loss": 3.8438,
+      "step": 7257
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.5614048526046145e-05,
+      "loss": 3.7031,
+      "step": 7258
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.5612718633031714e-05,
+      "loss": 6.5625,
+      "step": 7259
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.5611388563263374e-05,
+      "loss": 4.0,
+      "step": 7260
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.5610058316756175e-05,
+      "loss": 3.0625,
+      "step": 7261
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.5608727893525174e-05,
+      "loss": 3.6562,
+      "step": 7262
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.560739729358544e-05,
+      "loss": 3.9062,
+      "step": 7263
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.560606651695203e-05,
+      "loss": 3.5312,
+      "step": 7264
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.5604735563640006e-05,
+      "loss": 3.6406,
+      "step": 7265
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.560340443366445e-05,
+      "loss": 3.9219,
+      "step": 7266
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.560207312704043e-05,
+      "loss": 4.0312,
+      "step": 7267
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.5600741643783e-05,
+      "loss": 4.6562,
+      "step": 7268
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.559940998390725e-05,
+      "loss": 5.5469,
+      "step": 7269
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.559807814742825e-05,
+      "loss": 3.4531,
+      "step": 7270
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.5596746134361086e-05,
+      "loss": 1.9844,
+      "step": 7271
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.5595413944720834e-05,
+      "loss": 3.125,
+      "step": 7272
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.559408157852256e-05,
+      "loss": 3.3438,
+      "step": 7273
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.559274903578137e-05,
+      "loss": 5.3438,
+      "step": 7274
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.559141631651235e-05,
+      "loss": 3.5469,
+      "step": 7275
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.5590083420730565e-05,
+      "loss": 4.7031,
+      "step": 7276
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.558875034845113e-05,
+      "loss": 3.7969,
+      "step": 7277
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.558741709968912e-05,
+      "loss": 4.5,
+      "step": 7278
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.558608367445964e-05,
+      "loss": 5.0,
+      "step": 7279
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.5584750072777784e-05,
+      "loss": 2.1875,
+      "step": 7280
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.558341629465864e-05,
+      "loss": 3.9062,
+      "step": 7281
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.5582082340117315e-05,
+      "loss": 3.7344,
+      "step": 7282
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.558074820916892e-05,
+      "loss": 3.2812,
+      "step": 7283
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.557941390182854e-05,
+      "loss": 2.7812,
+      "step": 7284
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.557807941811129e-05,
+      "loss": 2.375,
+      "step": 7285
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.557674475803228e-05,
+      "loss": 4.0,
+      "step": 7286
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.5575409921606625e-05,
+      "loss": 3.9531,
+      "step": 7287
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.557407490884942e-05,
+      "loss": 2.3125,
+      "step": 7288
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.55727397197758e-05,
+      "loss": 4.7188,
+      "step": 7289
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.557140435440086e-05,
+      "loss": 4.0938,
+      "step": 7290
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.5570068812739733e-05,
+      "loss": 2.6875,
+      "step": 7291
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.556873309480754e-05,
+      "loss": 5.625,
+      "step": 7292
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.556739720061939e-05,
+      "loss": 1.5312,
+      "step": 7293
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.556606113019041e-05,
+      "loss": 5.4062,
+      "step": 7294
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.556472488353574e-05,
+      "loss": 3.0312,
+      "step": 7295
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.556338846067049e-05,
+      "loss": 2.5625,
+      "step": 7296
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.55620518616098e-05,
+      "loss": 3.0,
+      "step": 7297
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.556071508636879e-05,
+      "loss": 3.5469,
+      "step": 7298
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.5559378134962615e-05,
+      "loss": 3.5,
+      "step": 7299
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.5558041007406395e-05,
+      "loss": 4.7656,
+      "step": 7300
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.555670370371527e-05,
+      "loss": 2.9375,
+      "step": 7301
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.555536622390438e-05,
+      "loss": 2.875,
+      "step": 7302
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.555402856798887e-05,
+      "loss": 5.875,
+      "step": 7303
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.5552690735983874e-05,
+      "loss": 2.375,
+      "step": 7304
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.5551352727904555e-05,
+      "loss": 4.0312,
+      "step": 7305
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.555001454376605e-05,
+      "loss": 4.125,
+      "step": 7306
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.554867618358351e-05,
+      "loss": 2.8438,
+      "step": 7307
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.554733764737209e-05,
+      "loss": 4.3125,
+      "step": 7308
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.5545998935146935e-05,
+      "loss": 3.75,
+      "step": 7309
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.55446600469232e-05,
+      "loss": 3.8594,
+      "step": 7310
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.554332098271606e-05,
+      "loss": 4.1094,
+      "step": 7311
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.554198174254067e-05,
+      "loss": 6.375,
+      "step": 7312
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.5540642326412174e-05,
+      "loss": 4.4688,
+      "step": 7313
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.553930273434575e-05,
+      "loss": 6.375,
+      "step": 7314
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.553796296635656e-05,
+      "loss": 6.4531,
+      "step": 7315
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.553662302245978e-05,
+      "loss": 5.125,
+      "step": 7316
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.553528290267056e-05,
+      "loss": 4.875,
+      "step": 7317
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.55339426070041e-05,
+      "loss": 2.9531,
+      "step": 7318
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.553260213547555e-05,
+      "loss": 2.5,
+      "step": 7319
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.553126148810009e-05,
+      "loss": 3.1562,
+      "step": 7320
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.5529920664892906e-05,
+      "loss": 3.7188,
+      "step": 7321
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.552857966586917e-05,
+      "loss": 2.6875,
+      "step": 7322
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.5527238491044066e-05,
+      "loss": 2.2344,
+      "step": 7323
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.552589714043278e-05,
+      "loss": 5.0312,
+      "step": 7324
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.55245556140505e-05,
+      "loss": 4.9375,
+      "step": 7325
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.5523213911912406e-05,
+      "loss": 5.375,
+      "step": 7326
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.552187203403369e-05,
+      "loss": 3.6094,
+      "step": 7327
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.552052998042955e-05,
+      "loss": 3.5938,
+      "step": 7328
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.5519187751115175e-05,
+      "loss": 6.0938,
+      "step": 7329
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.551784534610575e-05,
+      "loss": 3.6875,
+      "step": 7330
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.551650276541649e-05,
+      "loss": 2.6562,
+      "step": 7331
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.551516000906258e-05,
+      "loss": 3.3125,
+      "step": 7332
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.5513817077059234e-05,
+      "loss": 3.8438,
+      "step": 7333
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.5512473969421655e-05,
+      "loss": 2.9531,
+      "step": 7334
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.551113068616504e-05,
+      "loss": 3.375,
+      "step": 7335
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.550978722730459e-05,
+      "loss": 3.4062,
+      "step": 7336
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.550844359285554e-05,
+      "loss": 3.4688,
+      "step": 7337
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.5507099782833076e-05,
+      "loss": 2.8125,
+      "step": 7338
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.5505755797252426e-05,
+      "loss": 3.4375,
+      "step": 7339
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.55044116361288e-05,
+      "loss": 3.5,
+      "step": 7340
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.550306729947742e-05,
+      "loss": 4.375,
+      "step": 7341
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.55017227873135e-05,
+      "loss": 3.6875,
+      "step": 7342
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.550037809965226e-05,
+      "loss": 3.7188,
+      "step": 7343
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.549903323650893e-05,
+      "loss": 5.6875,
+      "step": 7344
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.549768819789873e-05,
+      "loss": 3.25,
+      "step": 7345
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.549634298383689e-05,
+      "loss": 3.4219,
+      "step": 7346
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.549499759433864e-05,
+      "loss": 5.1094,
+      "step": 7347
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.5493652029419213e-05,
+      "loss": 5.6875,
+      "step": 7348
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.549230628909383e-05,
+      "loss": 3.1875,
+      "step": 7349
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.549096037337775e-05,
+      "loss": 4.0781,
+      "step": 7350
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.548961428228618e-05,
+      "loss": 4.4688,
+      "step": 7351
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.548826801583439e-05,
+      "loss": 2.4688,
+      "step": 7352
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.54869215740376e-05,
+      "loss": 3.75,
+      "step": 7353
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.548557495691106e-05,
+      "loss": 6.25,
+      "step": 7354
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.548422816447002e-05,
+      "loss": 2.4062,
+      "step": 7355
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.548288119672972e-05,
+      "loss": 2.7812,
+      "step": 7356
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.548153405370541e-05,
+      "loss": 4.0469,
+      "step": 7357
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.548018673541234e-05,
+      "loss": 3.9375,
+      "step": 7358
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.5478839241865775e-05,
+      "loss": 2.1875,
+      "step": 7359
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.547749157308096e-05,
+      "loss": 8.0625,
+      "step": 7360
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.547614372907314e-05,
+      "loss": 3.0156,
+      "step": 7361
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.54747957098576e-05,
+      "loss": 4.4062,
+      "step": 7362
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.5473447515449594e-05,
+      "loss": 3.75,
+      "step": 7363
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.547209914586437e-05,
+      "loss": 2.3594,
+      "step": 7364
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.547075060111721e-05,
+      "loss": 3.125,
+      "step": 7365
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.546940188122337e-05,
+      "loss": 3.6875,
+      "step": 7366
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.5468052986198125e-05,
+      "loss": 3.625,
+      "step": 7367
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.546670391605674e-05,
+      "loss": 4.125,
+      "step": 7368
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.5465354670814494e-05,
+      "loss": 5.5,
+      "step": 7369
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.546400525048666e-05,
+      "loss": 3.0312,
+      "step": 7370
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.546265565508852e-05,
+      "loss": 3.375,
+      "step": 7371
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.546130588463534e-05,
+      "loss": 7.75,
+      "step": 7372
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.545995593914242e-05,
+      "loss": 2.9531,
+      "step": 7373
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.545860581862502e-05,
+      "loss": 4.125,
+      "step": 7374
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.5457255523098445e-05,
+      "loss": 5.9688,
+      "step": 7375
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.545590505257797e-05,
+      "loss": 4.5625,
+      "step": 7376
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.545455440707889e-05,
+      "loss": 5.8125,
+      "step": 7377
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.545320358661649e-05,
+      "loss": 6.3125,
+      "step": 7378
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.5451852591206074e-05,
+      "loss": 2.5,
+      "step": 7379
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.545050142086293e-05,
+      "loss": 3.375,
+      "step": 7380
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.544915007560235e-05,
+      "loss": 3.375,
+      "step": 7381
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.544779855543963e-05,
+      "loss": 4.4688,
+      "step": 7382
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.544644686039009e-05,
+      "loss": 2.8438,
+      "step": 7383
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.5445094990469006e-05,
+      "loss": 4.2969,
+      "step": 7384
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.544374294569171e-05,
+      "loss": 4.5312,
+      "step": 7385
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.544239072607349e-05,
+      "loss": 2.8438,
+      "step": 7386
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.544103833162966e-05,
+      "loss": 3.7812,
+      "step": 7387
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.543968576237553e-05,
+      "loss": 4.75,
+      "step": 7388
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.543833301832642e-05,
+      "loss": 4.0625,
+      "step": 7389
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.543698009949764e-05,
+      "loss": 3.4688,
+      "step": 7390
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.5435627005904494e-05,
+      "loss": 4.2188,
+      "step": 7391
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.543427373756232e-05,
+      "loss": 5.3281,
+      "step": 7392
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.543292029448643e-05,
+      "loss": 3.4062,
+      "step": 7393
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.543156667669214e-05,
+      "loss": 3.4531,
+      "step": 7394
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.543021288419479e-05,
+      "loss": 3.375,
+      "step": 7395
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.542885891700969e-05,
+      "loss": 4.4688,
+      "step": 7396
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.542750477515218e-05,
+      "loss": 5.7031,
+      "step": 7397
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.542615045863759e-05,
+      "loss": 2.4688,
+      "step": 7398
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.542479596748124e-05,
+      "loss": 5.3594,
+      "step": 7399
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.542344130169848e-05,
+      "loss": 3.1406,
+      "step": 7400
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.5422086461304635e-05,
+      "loss": 4.5156,
+      "step": 7401
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.542073144631505e-05,
+      "loss": 2.5938,
+      "step": 7402
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.541937625674507e-05,
+      "loss": 4.0625,
+      "step": 7403
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.541802089261003e-05,
+      "loss": 4.6562,
+      "step": 7404
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.541666535392526e-05,
+      "loss": 4.125,
+      "step": 7405
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.5415309640706134e-05,
+      "loss": 4.8281,
+      "step": 7406
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.541395375296799e-05,
+      "loss": 5.6875,
+      "step": 7407
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.541259769072617e-05,
+      "loss": 3.9062,
+      "step": 7408
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.5411241453996036e-05,
+      "loss": 7.5,
+      "step": 7409
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.540988504279293e-05,
+      "loss": 5.3438,
+      "step": 7410
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.5408528457132224e-05,
+      "loss": 1.7656,
+      "step": 7411
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.540717169702926e-05,
+      "loss": 1.9219,
+      "step": 7412
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5405814762499416e-05,
+      "loss": 6.125,
+      "step": 7413
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.540445765355804e-05,
+      "loss": 4.0156,
+      "step": 7414
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.54031003702205e-05,
+      "loss": 3.5625,
+      "step": 7415
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.540174291250217e-05,
+      "loss": 4.375,
+      "step": 7416
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.54003852804184e-05,
+      "loss": 5.3438,
+      "step": 7417
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.539902747398458e-05,
+      "loss": 4.0938,
+      "step": 7418
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.539766949321606e-05,
+      "loss": 3.7188,
+      "step": 7419
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5396311338128236e-05,
+      "loss": 3.25,
+      "step": 7420
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.539495300873647e-05,
+      "loss": 3.7188,
+      "step": 7421
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5393594505056144e-05,
+      "loss": 2.1875,
+      "step": 7422
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5392235827102633e-05,
+      "loss": 5.625,
+      "step": 7423
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.539087697489134e-05,
+      "loss": 4.4375,
+      "step": 7424
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5389517948437615e-05,
+      "loss": 2.6562,
+      "step": 7425
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.538815874775687e-05,
+      "loss": 4.6406,
+      "step": 7426
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.538679937286448e-05,
+      "loss": 2.4375,
+      "step": 7427
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.538543982377584e-05,
+      "loss": 3.375,
+      "step": 7428
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.538408010050634e-05,
+      "loss": 1.5,
+      "step": 7429
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5382720203071376e-05,
+      "loss": 2.0625,
+      "step": 7430
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.538136013148634e-05,
+      "loss": 2.3125,
+      "step": 7431
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5379999885766626e-05,
+      "loss": 3.1094,
+      "step": 7432
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5378639465927644e-05,
+      "loss": 4.4219,
+      "step": 7433
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5377278871984794e-05,
+      "loss": 6.3125,
+      "step": 7434
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5375918103953464e-05,
+      "loss": 2.6562,
+      "step": 7435
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.5374557161849084e-05,
+      "loss": 3.9062,
+      "step": 7436
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.537319604568704e-05,
+      "loss": 2.8594,
+      "step": 7437
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.537183475548275e-05,
+      "loss": 4.0469,
+      "step": 7438
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.537047329125163e-05,
+      "loss": 3.6719,
+      "step": 7439
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.536911165300908e-05,
+      "loss": 3.6875,
+      "step": 7440
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.536774984077053e-05,
+      "loss": 4.4062,
+      "step": 7441
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.536638785455139e-05,
+      "loss": 4.9844,
+      "step": 7442
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.536502569436708e-05,
+      "loss": 3.0469,
+      "step": 7443
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.536366336023303e-05,
+      "loss": 2.5625,
+      "step": 7444
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.5362300852164644e-05,
+      "loss": 3.375,
+      "step": 7445
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.5360938170177355e-05,
+      "loss": 4.2188,
+      "step": 7446
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.5359575314286606e-05,
+      "loss": 5.0,
+      "step": 7447
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.5358212284507805e-05,
+      "loss": 4.25,
+      "step": 7448
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.535684908085639e-05,
+      "loss": 4.1562,
+      "step": 7449
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.5355485703347804e-05,
+      "loss": 2.0938,
+      "step": 7450
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.535412215199746e-05,
+      "loss": 1.625,
+      "step": 7451
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.535275842682082e-05,
+      "loss": 4.125,
+      "step": 7452
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.53513945278333e-05,
+      "loss": 3.4062,
+      "step": 7453
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.535003045505036e-05,
+      "loss": 3.5938,
+      "step": 7454
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.534866620848743e-05,
+      "loss": 7.3594,
+      "step": 7455
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.5347301788159965e-05,
+      "loss": 3.875,
+      "step": 7456
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.53459371940834e-05,
+      "loss": 2.75,
+      "step": 7457
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.53445724262732e-05,
+      "loss": 3.3125,
+      "step": 7458
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.53432074847448e-05,
+      "loss": 4.8906,
+      "step": 7459
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.534184236951366e-05,
+      "loss": 4.6562,
+      "step": 7460
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.5340477080595234e-05,
+      "loss": 5.6562,
+      "step": 7461
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.533911161800498e-05,
+      "loss": 3.25,
+      "step": 7462
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.533774598175835e-05,
+      "loss": 4.0312,
+      "step": 7463
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.533638017187081e-05,
+      "loss": 2.4062,
+      "step": 7464
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.5335014188357825e-05,
+      "loss": 4.8438,
+      "step": 7465
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.5333648031234856e-05,
+      "loss": 3.2188,
+      "step": 7466
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.533228170051737e-05,
+      "loss": 5.8906,
+      "step": 7467
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.533091519622083e-05,
+      "loss": 2.6875,
+      "step": 7468
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.532954851836072e-05,
+      "loss": 2.75,
+      "step": 7469
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.5328181666952496e-05,
+      "loss": 4.4688,
+      "step": 7470
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.5326814642011645e-05,
+      "loss": 4.0,
+      "step": 7471
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.532544744355363e-05,
+      "loss": 1.75,
+      "step": 7472
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.5324080071593943e-05,
+      "loss": 4.0,
+      "step": 7473
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.532271252614807e-05,
+      "loss": 3.25,
+      "step": 7474
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.532134480723147e-05,
+      "loss": 3.1875,
+      "step": 7475
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.531997691485964e-05,
+      "loss": 2.6875,
+      "step": 7476
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5318608849048065e-05,
+      "loss": 4.5625,
+      "step": 7477
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.531724060981224e-05,
+      "loss": 3.4688,
+      "step": 7478
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.531587219716764e-05,
+      "loss": 3.0625,
+      "step": 7479
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5314503611129776e-05,
+      "loss": 2.5312,
+      "step": 7480
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5313134851714124e-05,
+      "loss": 8.6719,
+      "step": 7481
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.531176591893619e-05,
+      "loss": 2.2812,
+      "step": 7482
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.531039681281147e-05,
+      "loss": 4.4375,
+      "step": 7483
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.530902753335546e-05,
+      "loss": 6.4531,
+      "step": 7484
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.530765808058367e-05,
+      "loss": 3.9688,
+      "step": 7485
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.53062884545116e-05,
+      "loss": 5.8438,
+      "step": 7486
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5304918655154754e-05,
+      "loss": 3.5625,
+      "step": 7487
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5303548682528635e-05,
+      "loss": 3.1875,
+      "step": 7488
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.530217853664876e-05,
+      "loss": 4.2188,
+      "step": 7489
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.530080821753064e-05,
+      "loss": 5.8125,
+      "step": 7490
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.529943772518978e-05,
+      "loss": 3.1406,
+      "step": 7491
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5298067059641714e-05,
+      "loss": 3.0938,
+      "step": 7492
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.529669622090194e-05,
+      "loss": 1.9375,
+      "step": 7493
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5295325208985995e-05,
+      "loss": 2.5312,
+      "step": 7494
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.529395402390938e-05,
+      "loss": 3.6875,
+      "step": 7495
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.529258266568764e-05,
+      "loss": 3.75,
+      "step": 7496
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.529121113433628e-05,
+      "loss": 2.6406,
+      "step": 7497
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.528983942987084e-05,
+      "loss": 4.2188,
+      "step": 7498
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.528846755230685e-05,
+      "loss": 4.7969,
+      "step": 7499
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.528709550165984e-05,
+      "loss": 5.5781,
+      "step": 7500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.528572327794533e-05,
+      "loss": 5.6328,
+      "step": 7501
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.5284350881178885e-05,
+      "loss": 2.6719,
+      "step": 7502
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.528297831137601e-05,
+      "loss": 2.7812,
+      "step": 7503
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5281605568552256e-05,
+      "loss": 5.9688,
+      "step": 7504
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5280232652723165e-05,
+      "loss": 5.3906,
+      "step": 7505
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.527885956390429e-05,
+      "loss": 4.8125,
+      "step": 7506
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.527748630211116e-05,
+      "loss": 3.625,
+      "step": 7507
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5276112867359324e-05,
+      "loss": 3.4844,
+      "step": 7508
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5274739259664336e-05,
+      "loss": 2.625,
+      "step": 7509
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.527336547904175e-05,
+      "loss": 4.7969,
+      "step": 7510
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5271991525507114e-05,
+      "loss": 2.4062,
+      "step": 7511
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.527061739907598e-05,
+      "loss": 4.9219,
+      "step": 7512
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.526924309976391e-05,
+      "loss": 4.5312,
+      "step": 7513
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5267868627586464e-05,
+      "loss": 5.1953,
+      "step": 7514
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5266493982559194e-05,
+      "loss": 3.0312,
+      "step": 7515
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.526511916469767e-05,
+      "loss": 3.3516,
+      "step": 7516
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.526374417401745e-05,
+      "loss": 1.875,
+      "step": 7517
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5262369010534107e-05,
+      "loss": 5.5625,
+      "step": 7518
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.526099367426321e-05,
+      "loss": 4.625,
+      "step": 7519
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.525961816522033e-05,
+      "loss": 3.4688,
+      "step": 7520
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.525824248342103e-05,
+      "loss": 3.9688,
+      "step": 7521
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.525686662888089e-05,
+      "loss": 2.4062,
+      "step": 7522
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5255490601615485e-05,
+      "loss": 4.5625,
+      "step": 7523
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.52541144016404e-05,
+      "loss": 1.9375,
+      "step": 7524
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.525273802897121e-05,
+      "loss": 2.4062,
+      "step": 7525
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.525136148362349e-05,
+      "loss": 3.0156,
+      "step": 7526
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.524998476561284e-05,
+      "loss": 7.3906,
+      "step": 7527
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.524860787495484e-05,
+      "loss": 4.0,
+      "step": 7528
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5247230811665065e-05,
+      "loss": 6.6016,
+      "step": 7529
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.524585357575911e-05,
+      "loss": 5.1562,
+      "step": 7530
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5244476167252583e-05,
+      "loss": 2.1875,
+      "step": 7531
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.524309858616106e-05,
+      "loss": 2.9688,
+      "step": 7532
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.524172083250015e-05,
+      "loss": 2.8438,
+      "step": 7533
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.524034290628544e-05,
+      "loss": 3.8281,
+      "step": 7534
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.523896480753254e-05,
+      "loss": 2.0625,
+      "step": 7535
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.523758653625704e-05,
+      "loss": 4.5312,
+      "step": 7536
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.523620809247455e-05,
+      "loss": 4.5469,
+      "step": 7537
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.523482947620068e-05,
+      "loss": 3.6719,
+      "step": 7538
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.523345068745102e-05,
+      "loss": 3.4375,
+      "step": 7539
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.52320717262412e-05,
+      "loss": 3.6562,
+      "step": 7540
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.523069259258683e-05,
+      "loss": 5.25,
+      "step": 7541
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.522931328650351e-05,
+      "loss": 3.7188,
+      "step": 7542
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.522793380800686e-05,
+      "loss": 4.6094,
+      "step": 7543
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.52265541571125e-05,
+      "loss": 3.6875,
+      "step": 7544
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5225174333836045e-05,
+      "loss": 7.1875,
+      "step": 7545
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.522379433819313e-05,
+      "loss": 2.2812,
+      "step": 7546
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.522241417019936e-05,
+      "loss": 3.6562,
+      "step": 7547
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.522103382987036e-05,
+      "loss": 4.1875,
+      "step": 7548
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.521965331722178e-05,
+      "loss": 5.7812,
+      "step": 7549
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.521827263226923e-05,
+      "loss": 4.5781,
+      "step": 7550
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5216891775028344e-05,
+      "loss": 3.8438,
+      "step": 7551
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.521551074551476e-05,
+      "loss": 2.2031,
+      "step": 7552
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.52141295437441e-05,
+      "loss": 3.7188,
+      "step": 7553
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.521274816973201e-05,
+      "loss": 5.375,
+      "step": 7554
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5211366623494136e-05,
+      "loss": 4.6719,
+      "step": 7555
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5209984905046105e-05,
+      "loss": 2.9844,
+      "step": 7556
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.520860301440357e-05,
+      "loss": 3.9375,
+      "step": 7557
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.520722095158217e-05,
+      "loss": 5.6875,
+      "step": 7558
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.520583871659755e-05,
+      "loss": 2.8594,
+      "step": 7559
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5204456309465355e-05,
+      "loss": 5.1406,
+      "step": 7560
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5203073730201254e-05,
+      "loss": 3.4062,
+      "step": 7561
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.520169097882088e-05,
+      "loss": 2.2812,
+      "step": 7562
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.520030805533989e-05,
+      "loss": 2.3125,
+      "step": 7563
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.519892495977395e-05,
+      "loss": 4.3281,
+      "step": 7564
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.519754169213872e-05,
+      "loss": 3.2188,
+      "step": 7565
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.519615825244984e-05,
+      "loss": 4.0312,
+      "step": 7566
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5194774640722995e-05,
+      "loss": 4.375,
+      "step": 7567
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5193390856973835e-05,
+      "loss": 6.875,
+      "step": 7568
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5192006901218025e-05,
+      "loss": 4.875,
+      "step": 7569
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.519062277347125e-05,
+      "loss": 5.0156,
+      "step": 7570
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5189238473749155e-05,
+      "loss": 7.4219,
+      "step": 7571
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.518785400206743e-05,
+      "loss": 4.0156,
+      "step": 7572
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5186469358441746e-05,
+      "loss": 2.1562,
+      "step": 7573
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5185084542887776e-05,
+      "loss": 5.0469,
+      "step": 7574
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.51836995554212e-05,
+      "loss": 2.6719,
+      "step": 7575
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5182314396057693e-05,
+      "loss": 2.5,
+      "step": 7576
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.518092906481294e-05,
+      "loss": 2.5312,
+      "step": 7577
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.517954356170262e-05,
+      "loss": 4.3594,
+      "step": 7578
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5178157886742425e-05,
+      "loss": 2.5312,
+      "step": 7579
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.517677203994804e-05,
+      "loss": 2.5469,
+      "step": 7580
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.517538602133516e-05,
+      "loss": 5.5312,
+      "step": 7581
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.517399983091946e-05,
+      "loss": 3.4062,
+      "step": 7582
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.517261346871666e-05,
+      "loss": 3.375,
+      "step": 7583
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.517122693474243e-05,
+      "loss": 2.1719,
+      "step": 7584
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5169840229012473e-05,
+      "loss": 2.4375,
+      "step": 7585
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.516845335154249e-05,
+      "loss": 3.0,
+      "step": 7586
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5167066302348185e-05,
+      "loss": 4.0,
+      "step": 7587
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5165679081445264e-05,
+      "loss": 3.5,
+      "step": 7588
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.516429168884942e-05,
+      "loss": 4.6406,
+      "step": 7589
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.516290412457638e-05,
+      "loss": 3.2344,
+      "step": 7590
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.516151638864183e-05,
+      "loss": 2.5156,
+      "step": 7591
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.516012848106149e-05,
+      "loss": 5.7812,
+      "step": 7592
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.515874040185108e-05,
+      "loss": 3.9844,
+      "step": 7593
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5157352151026305e-05,
+      "loss": 2.25,
+      "step": 7594
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.515596372860288e-05,
+      "loss": 4.8125,
+      "step": 7595
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.515457513459653e-05,
+      "loss": 4.4062,
+      "step": 7596
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.515318636902298e-05,
+      "loss": 3.7344,
+      "step": 7597
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.515179743189794e-05,
+      "loss": 3.0781,
+      "step": 7598
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.515040832323715e-05,
+      "loss": 4.0469,
+      "step": 7599
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.5149019043056315e-05,
+      "loss": 4.3438,
+      "step": 7600
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.514762959137118e-05,
+      "loss": 2.625,
+      "step": 7601
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.514623996819747e-05,
+      "loss": 3.7812,
+      "step": 7602
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.514485017355092e-05,
+      "loss": 3.5,
+      "step": 7603
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.514346020744726e-05,
+      "loss": 6.0,
+      "step": 7604
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.514207006990222e-05,
+      "loss": 4.25,
+      "step": 7605
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.5140679760931557e-05,
+      "loss": 3.25,
+      "step": 7606
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.5139289280550994e-05,
+      "loss": 2.9688,
+      "step": 7607
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.5137898628776276e-05,
+      "loss": 2.7812,
+      "step": 7608
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.513650780562316e-05,
+      "loss": 5.3438,
+      "step": 7609
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.5135116811107365e-05,
+      "loss": 4.875,
+      "step": 7610
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.513372564524466e-05,
+      "loss": 4.4531,
+      "step": 7611
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.5132334308050794e-05,
+      "loss": 4.875,
+      "step": 7612
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.513094279954151e-05,
+      "loss": 3.0938,
+      "step": 7613
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.512955111973257e-05,
+      "loss": 3.8594,
+      "step": 7614
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.512815926863971e-05,
+      "loss": 3.875,
+      "step": 7615
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.512676724627872e-05,
+      "loss": 3.9531,
+      "step": 7616
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.512537505266533e-05,
+      "loss": 3.2812,
+      "step": 7617
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.512398268781531e-05,
+      "loss": 3.9219,
+      "step": 7618
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.512259015174443e-05,
+      "loss": 4.25,
+      "step": 7619
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.5121197444468454e-05,
+      "loss": 2.8281,
+      "step": 7620
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.5119804566003136e-05,
+      "loss": 4.5938,
+      "step": 7621
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.5118411516364264e-05,
+      "loss": 3.5938,
+      "step": 7622
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.51170182955676e-05,
+      "loss": 3.7656,
+      "step": 7623
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.511562490362891e-05,
+      "loss": 4.1875,
+      "step": 7624
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.511423134056397e-05,
+      "loss": 2.9062,
+      "step": 7625
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.511283760638858e-05,
+      "loss": 2.125,
+      "step": 7626
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.5111443701118486e-05,
+      "loss": 1.7812,
+      "step": 7627
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.5110049624769484e-05,
+      "loss": 5.1562,
+      "step": 7628
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.510865537735736e-05,
+      "loss": 3.4531,
+      "step": 7629
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.510726095889789e-05,
+      "loss": 2.3906,
+      "step": 7630
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.510586636940687e-05,
+      "loss": 1.9688,
+      "step": 7631
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.510447160890008e-05,
+      "loss": 4.0938,
+      "step": 7632
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.5103076677393316e-05,
+      "loss": 2.1562,
+      "step": 7633
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.510168157490237e-05,
+      "loss": 4.2656,
+      "step": 7634
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.510028630144302e-05,
+      "loss": 8.6875,
+      "step": 7635
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.509889085703108e-05,
+      "loss": 2.5938,
+      "step": 7636
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.509749524168235e-05,
+      "loss": 3.0938,
+      "step": 7637
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.5096099455412626e-05,
+      "loss": 4.5625,
+      "step": 7638
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.5094703498237705e-05,
+      "loss": 3.9062,
+      "step": 7639
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.509330737017339e-05,
+      "loss": 3.9844,
+      "step": 7640
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.509191107123549e-05,
+      "loss": 3.1875,
+      "step": 7641
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.509051460143982e-05,
+      "loss": 5.3125,
+      "step": 7642
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.508911796080217e-05,
+      "loss": 5.1094,
+      "step": 7643
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.508772114933838e-05,
+      "loss": 3.625,
+      "step": 7644
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.508632416706424e-05,
+      "loss": 3.5,
+      "step": 7645
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.508492701399557e-05,
+      "loss": 2.9062,
+      "step": 7646
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.508352969014819e-05,
+      "loss": 2.0,
+      "step": 7647
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.508213219553793e-05,
+      "loss": 2.6562,
+      "step": 7648
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.508073453018059e-05,
+      "loss": 2.7188,
+      "step": 7649
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.5079336694092014e-05,
+      "loss": 4.2188,
+      "step": 7650
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.5077938687288004e-05,
+      "loss": 3.4062,
+      "step": 7651
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.507654050978441e-05,
+      "loss": 3.0625,
+      "step": 7652
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.5075142161597055e-05,
+      "loss": 3.1562,
+      "step": 7653
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.507374364274176e-05,
+      "loss": 4.0,
+      "step": 7654
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.5072344953234365e-05,
+      "loss": 3.4688,
+      "step": 7655
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.5070946093090706e-05,
+      "loss": 4.2031,
+      "step": 7656
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.506954706232662e-05,
+      "loss": 3.25,
+      "step": 7657
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.506814786095794e-05,
+      "loss": 2.1562,
+      "step": 7658
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.50667484890005e-05,
+      "loss": 3.3281,
+      "step": 7659
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.5065348946470166e-05,
+      "loss": 3.0938,
+      "step": 7660
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.5063949233382767e-05,
+      "loss": 3.25,
+      "step": 7661
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.506254934975414e-05,
+      "loss": 2.875,
+      "step": 7662
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.506114929560015e-05,
+      "loss": 4.3281,
+      "step": 7663
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.505974907093665e-05,
+      "loss": 5.375,
+      "step": 7664
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.5058348675779474e-05,
+      "loss": 5.0938,
+      "step": 7665
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.5056948110144496e-05,
+      "loss": 7.4688,
+      "step": 7666
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.505554737404755e-05,
+      "loss": 3.25,
+      "step": 7667
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.505414646750452e-05,
+      "loss": 3.3438,
+      "step": 7668
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.5052745390531236e-05,
+      "loss": 2.2188,
+      "step": 7669
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.505134414314358e-05,
+      "loss": 5.6562,
+      "step": 7670
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.5049942725357416e-05,
+      "loss": 4.7031,
+      "step": 7671
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.50485411371886e-05,
+      "loss": 3.0469,
+      "step": 7672
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.504713937865301e-05,
+      "loss": 3.7969,
+      "step": 7673
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.504573744976651e-05,
+      "loss": 3.0938,
+      "step": 7674
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.5044335350544966e-05,
+      "loss": 4.1172,
+      "step": 7675
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.504293308100426e-05,
+      "loss": 5.9688,
+      "step": 7676
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.504153064116026e-05,
+      "loss": 3.2188,
+      "step": 7677
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.504012803102885e-05,
+      "loss": 2.8828,
+      "step": 7678
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.503872525062591e-05,
+      "loss": 4.5312,
+      "step": 7679
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.503732229996732e-05,
+      "loss": 3.1562,
+      "step": 7680
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.503591917906895e-05,
+      "loss": 4.5938,
+      "step": 7681
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.5034515887946704e-05,
+      "loss": 4.0312,
+      "step": 7682
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.503311242661646e-05,
+      "loss": 3.5,
+      "step": 7683
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.50317087950941e-05,
+      "loss": 4.4531,
+      "step": 7684
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.503030499339553e-05,
+      "loss": 4.3438,
+      "step": 7685
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.502890102153663e-05,
+      "loss": 3.5625,
+      "step": 7686
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.5027496879533295e-05,
+      "loss": 5.4844,
+      "step": 7687
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.502609256740144e-05,
+      "loss": 4.1328,
+      "step": 7688
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.5024688085156936e-05,
+      "loss": 5.2188,
+      "step": 7689
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.50232834328157e-05,
+      "loss": 7.0,
+      "step": 7690
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.502187861039364e-05,
+      "loss": 3.2656,
+      "step": 7691
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.502047361790664e-05,
+      "loss": 4.9688,
+      "step": 7692
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.501906845537061e-05,
+      "loss": 3.5312,
+      "step": 7693
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.5017663122801475e-05,
+      "loss": 4.5625,
+      "step": 7694
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.501625762021514e-05,
+      "loss": 4.5938,
+      "step": 7695
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.5014851947627503e-05,
+      "loss": 2.375,
+      "step": 7696
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.5013446105054486e-05,
+      "loss": 2.0156,
+      "step": 7697
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.5012040092512016e-05,
+      "loss": 2.8438,
+      "step": 7698
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.501063391001599e-05,
+      "loss": 2.6875,
+      "step": 7699
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.500922755758234e-05,
+      "loss": 3.6875,
+      "step": 7700
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.5007821035226994e-05,
+      "loss": 4.0938,
+      "step": 7701
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.500641434296586e-05,
+      "loss": 4.125,
+      "step": 7702
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.5005007480814865e-05,
+      "loss": 3.1562,
+      "step": 7703
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.500360044878995e-05,
+      "loss": 2.2969,
+      "step": 7704
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.500219324690703e-05,
+      "loss": 2.8438,
+      "step": 7705
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.5000785875182045e-05,
+      "loss": 4.0938,
+      "step": 7706
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.4999378333630925e-05,
+      "loss": 6.5156,
+      "step": 7707
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.49979706222696e-05,
+      "loss": 1.9844,
+      "step": 7708
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.499656274111402e-05,
+      "loss": 3.0469,
+      "step": 7709
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.499515469018011e-05,
+      "loss": 6.0938,
+      "step": 7710
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.4993746469483816e-05,
+      "loss": 3.5625,
+      "step": 7711
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.499233807904109e-05,
+      "loss": 2.375,
+      "step": 7712
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.499092951886786e-05,
+      "loss": 3.6875,
+      "step": 7713
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.4989520788980076e-05,
+      "loss": 3.1719,
+      "step": 7714
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.498811188939369e-05,
+      "loss": 5.2812,
+      "step": 7715
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.498670282012466e-05,
+      "loss": 8.125,
+      "step": 7716
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.498529358118893e-05,
+      "loss": 3.5625,
+      "step": 7717
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.498388417260246e-05,
+      "loss": 4.4688,
+      "step": 7718
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.4982474594381194e-05,
+      "loss": 7.9688,
+      "step": 7719
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.49810648465411e-05,
+      "loss": 2.9375,
+      "step": 7720
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.4979654929098135e-05,
+      "loss": 3.125,
+      "step": 7721
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.497824484206827e-05,
+      "loss": 5.6406,
+      "step": 7722
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.4976834585467445e-05,
+      "loss": 4.3594,
+      "step": 7723
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.4975424159311654e-05,
+      "loss": 5.5,
+      "step": 7724
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.4974013563616844e-05,
+      "loss": 3.4688,
+      "step": 7725
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.4972602798398995e-05,
+      "loss": 4.9062,
+      "step": 7726
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.497119186367408e-05,
+      "loss": 4.1719,
+      "step": 7727
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.496978075945806e-05,
+      "loss": 5.3906,
+      "step": 7728
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.496836948576692e-05,
+      "loss": 2.375,
+      "step": 7729
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.496695804261664e-05,
+      "loss": 6.0781,
+      "step": 7730
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.496554643002319e-05,
+      "loss": 4.3125,
+      "step": 7731
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.496413464800256e-05,
+      "loss": 4.0781,
+      "step": 7732
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.496272269657073e-05,
+      "loss": 3.4219,
+      "step": 7733
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.496131057574368e-05,
+      "loss": 2.7812,
+      "step": 7734
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.495989828553741e-05,
+      "loss": 3.0938,
+      "step": 7735
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.495848582596788e-05,
+      "loss": 4.4375,
+      "step": 7736
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.495707319705111e-05,
+      "loss": 4.0,
+      "step": 7737
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.495566039880309e-05,
+      "loss": 3.8125,
+      "step": 7738
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.49542474312398e-05,
+      "loss": 4.1875,
+      "step": 7739
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.495283429437724e-05,
+      "loss": 3.4375,
+      "step": 7740
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.495142098823142e-05,
+      "loss": 7.3594,
+      "step": 7741
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.4950007512818325e-05,
+      "loss": 3.8594,
+      "step": 7742
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.494859386815397e-05,
+      "loss": 3.3125,
+      "step": 7743
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.494718005425435e-05,
+      "loss": 3.2344,
+      "step": 7744
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.494576607113548e-05,
+      "loss": 2.1875,
+      "step": 7745
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.4944351918813355e-05,
+      "loss": 4.8438,
+      "step": 7746
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.4942937597303994e-05,
+      "loss": 3.8438,
+      "step": 7747
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.494152310662341e-05,
+      "loss": 4.7812,
+      "step": 7748
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.4940108446787616e-05,
+      "loss": 2.7656,
+      "step": 7749
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.4938693617812624e-05,
+      "loss": 3.7266,
+      "step": 7750
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.4937278619714446e-05,
+      "loss": 5.4375,
+      "step": 7751
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.4935863452509115e-05,
+      "loss": 3.25,
+      "step": 7752
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.493444811621264e-05,
+      "loss": 3.7188,
+      "step": 7753
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.493303261084105e-05,
+      "loss": 5.3125,
+      "step": 7754
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.493161693641037e-05,
+      "loss": 3.9375,
+      "step": 7755
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.4930201092936634e-05,
+      "loss": 2.8594,
+      "step": 7756
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.492878508043586e-05,
+      "loss": 3.1094,
+      "step": 7757
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.4927368898924076e-05,
+      "loss": 3.9375,
+      "step": 7758
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.492595254841733e-05,
+      "loss": 3.8125,
+      "step": 7759
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.492453602893164e-05,
+      "loss": 3.9219,
+      "step": 7760
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.492311934048306e-05,
+      "loss": 3.8125,
+      "step": 7761
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.492170248308761e-05,
+      "loss": 4.6406,
+      "step": 7762
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.492028545676134e-05,
+      "loss": 11.0312,
+      "step": 7763
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.4918868261520306e-05,
+      "loss": 4.6562,
+      "step": 7764
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.491745089738052e-05,
+      "loss": 4.6719,
+      "step": 7765
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.491603336435806e-05,
+      "loss": 5.0,
+      "step": 7766
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.4914615662468954e-05,
+      "loss": 5.375,
+      "step": 7767
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.491319779172926e-05,
+      "loss": 3.4688,
+      "step": 7768
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.491177975215503e-05,
+      "loss": 4.0547,
+      "step": 7769
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.491036154376231e-05,
+      "loss": 3.25,
+      "step": 7770
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.490894316656718e-05,
+      "loss": 3.9375,
+      "step": 7771
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.4907524620585665e-05,
+      "loss": 9.0625,
+      "step": 7772
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.490610590583385e-05,
+      "loss": 5.3281,
+      "step": 7773
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.490468702232777e-05,
+      "loss": 2.8125,
+      "step": 7774
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.4903267970083515e-05,
+      "loss": 2.7812,
+      "step": 7775
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.4901848749117134e-05,
+      "loss": 5.5,
+      "step": 7776
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.490042935944471e-05,
+      "loss": 8.1562,
+      "step": 7777
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.489900980108229e-05,
+      "loss": 3.75,
+      "step": 7778
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.489759007404597e-05,
+      "loss": 2.625,
+      "step": 7779
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.48961701783518e-05,
+      "loss": 4.9219,
+      "step": 7780
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.4894750114015866e-05,
+      "loss": 6.3125,
+      "step": 7781
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.489332988105425e-05,
+      "loss": 6.9375,
+      "step": 7782
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.489190947948301e-05,
+      "loss": 5.6562,
+      "step": 7783
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.4890488909318256e-05,
+      "loss": 3.5156,
+      "step": 7784
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.488906817057605e-05,
+      "loss": 2.6719,
+      "step": 7785
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.488764726327248e-05,
+      "loss": 3.9688,
+      "step": 7786
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.4886226187423633e-05,
+      "loss": 3.75,
+      "step": 7787
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.48848049430456e-05,
+      "loss": 4.3125,
+      "step": 7788
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.488338353015447e-05,
+      "loss": 3.1562,
+      "step": 7789
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.4881961948766337e-05,
+      "loss": 5.2656,
+      "step": 7790
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.488054019889729e-05,
+      "loss": 3.8125,
+      "step": 7791
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.4879118280563427e-05,
+      "loss": 2.2188,
+      "step": 7792
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.4877696193780846e-05,
+      "loss": 3.75,
+      "step": 7793
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.487627393856564e-05,
+      "loss": 5.9688,
+      "step": 7794
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.487485151493393e-05,
+      "loss": 2.7812,
+      "step": 7795
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.48734289229018e-05,
+      "loss": 3.2188,
+      "step": 7796
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.487200616248536e-05,
+      "loss": 3.2656,
+      "step": 7797
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.487058323370072e-05,
+      "loss": 4.0938,
+      "step": 7798
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.486916013656399e-05,
+      "loss": 1.5625,
+      "step": 7799
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.4867736871091275e-05,
+      "loss": 4.3438,
+      "step": 7800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.48663134372987e-05,
+      "loss": 2.4688,
+      "step": 7801
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.4864889835202366e-05,
+      "loss": 3.0156,
+      "step": 7802
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.4863466064818406e-05,
+      "loss": 3.6562,
+      "step": 7803
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.4862042126162916e-05,
+      "loss": 2.1562,
+      "step": 7804
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.486061801925204e-05,
+      "loss": 3.8125,
+      "step": 7805
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.4859193744101884e-05,
+      "loss": 2.875,
+      "step": 7806
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.4857769300728584e-05,
+      "loss": 4.3281,
+      "step": 7807
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.4856344689148256e-05,
+      "loss": 2.8438,
+      "step": 7808
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.4854919909377035e-05,
+      "loss": 5.625,
+      "step": 7809
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.4853494961431047e-05,
+      "loss": 5.3281,
+      "step": 7810
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.485206984532643e-05,
+      "loss": 4.625,
+      "step": 7811
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.485064456107931e-05,
+      "loss": 4.0625,
+      "step": 7812
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.4849219108705834e-05,
+      "loss": 2.5938,
+      "step": 7813
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.484779348822212e-05,
+      "loss": 3.7188,
+      "step": 7814
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.484636769964433e-05,
+      "loss": 4.6719,
+      "step": 7815
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.48449417429886e-05,
+      "loss": 5.0625,
+      "step": 7816
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.484351561827106e-05,
+      "loss": 4.0,
+      "step": 7817
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.484208932550787e-05,
+      "loss": 3.0,
+      "step": 7818
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.4840662864715176e-05,
+      "loss": 3.5,
+      "step": 7819
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.483923623590911e-05,
+      "loss": 2.5625,
+      "step": 7820
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.4837809439105856e-05,
+      "loss": 3.1094,
+      "step": 7821
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.483638247432153e-05,
+      "loss": 3.9375,
+      "step": 7822
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.4834955341572315e-05,
+      "loss": 6.875,
+      "step": 7823
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.4833528040874355e-05,
+      "loss": 3.1094,
+      "step": 7824
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.4832100572243805e-05,
+      "loss": 5.9688,
+      "step": 7825
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.483067293569684e-05,
+      "loss": 2.4062,
+      "step": 7826
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.482924513124961e-05,
+      "loss": 4.75,
+      "step": 7827
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.482781715891829e-05,
+      "loss": 4.2656,
+      "step": 7828
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.482638901871903e-05,
+      "loss": 4.25,
+      "step": 7829
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.482496071066801e-05,
+      "loss": 3.5938,
+      "step": 7830
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.482353223478139e-05,
+      "loss": 8.1562,
+      "step": 7831
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.482210359107536e-05,
+      "loss": 3.125,
+      "step": 7832
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.482067477956609e-05,
+      "loss": 3.6016,
+      "step": 7833
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.481924580026974e-05,
+      "loss": 3.7812,
+      "step": 7834
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.4817816653202496e-05,
+      "loss": 3.1562,
+      "step": 7835
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.481638733838054e-05,
+      "loss": 3.7031,
+      "step": 7836
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.4814957855820054e-05,
+      "loss": 1.8594,
+      "step": 7837
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.481352820553722e-05,
+      "loss": 3.0625,
+      "step": 7838
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.481209838754822e-05,
+      "loss": 3.4375,
+      "step": 7839
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.481066840186925e-05,
+      "loss": 5.2969,
+      "step": 7840
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.480923824851648e-05,
+      "loss": 3.1875,
+      "step": 7841
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.4807807927506134e-05,
+      "loss": 3.0469,
+      "step": 7842
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.4806377438854366e-05,
+      "loss": 4.5312,
+      "step": 7843
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.48049467825774e-05,
+      "loss": 1.9219,
+      "step": 7844
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.4803515958691426e-05,
+      "loss": 2.0469,
+      "step": 7845
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.480208496721263e-05,
+      "loss": 3.8594,
+      "step": 7846
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.480065380815723e-05,
+      "loss": 4.7344,
+      "step": 7847
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.479922248154141e-05,
+      "loss": 7.4375,
+      "step": 7848
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.479779098738139e-05,
+      "loss": 2.75,
+      "step": 7849
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.479635932569337e-05,
+      "loss": 6.5781,
+      "step": 7850
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.479492749649356e-05,
+      "loss": 2.5938,
+      "step": 7851
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.4793495499798165e-05,
+      "loss": 4.6562,
+      "step": 7852
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.47920633356234e-05,
+      "loss": 3.7812,
+      "step": 7853
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.479063100398548e-05,
+      "loss": 3.2344,
+      "step": 7854
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.478919850490062e-05,
+      "loss": 3.1875,
+      "step": 7855
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.478776583838504e-05,
+      "loss": 6.4844,
+      "step": 7856
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.478633300445496e-05,
+      "loss": 4.6719,
+      "step": 7857
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.478490000312659e-05,
+      "loss": 3.0625,
+      "step": 7858
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.4783466834416165e-05,
+      "loss": 4.4531,
+      "step": 7859
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.4782033498339905e-05,
+      "loss": 4.7812,
+      "step": 7860
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.4780599994914025e-05,
+      "loss": 5.5,
+      "step": 7861
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.477916632415479e-05,
+      "loss": 4.9688,
+      "step": 7862
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.477773248607839e-05,
+      "loss": 2.9375,
+      "step": 7863
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.4776298480701086e-05,
+      "loss": 3.6406,
+      "step": 7864
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.47748643080391e-05,
+      "loss": 1.9219,
+      "step": 7865
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.477342996810867e-05,
+      "loss": 3.9062,
+      "step": 7866
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.4771995460926034e-05,
+      "loss": 5.0625,
+      "step": 7867
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.4770560786507435e-05,
+      "loss": 3.5938,
+      "step": 7868
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.476912594486911e-05,
+      "loss": 4.8438,
+      "step": 7869
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.476769093602731e-05,
+      "loss": 4.375,
+      "step": 7870
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.476625575999828e-05,
+      "loss": 4.25,
+      "step": 7871
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.476482041679826e-05,
+      "loss": 3.5,
+      "step": 7872
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.47633849064435e-05,
+      "loss": 5.7344,
+      "step": 7873
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.476194922895027e-05,
+      "loss": 5.2812,
+      "step": 7874
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.4760513384334806e-05,
+      "loss": 3.125,
+      "step": 7875
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.475907737261336e-05,
+      "loss": 2.3125,
+      "step": 7876
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.47576411938022e-05,
+      "loss": 2.9062,
+      "step": 7877
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.475620484791759e-05,
+      "loss": 5.6094,
+      "step": 7878
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.475476833497578e-05,
+      "loss": 2.0625,
+      "step": 7879
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.475333165499304e-05,
+      "loss": 4.1875,
+      "step": 7880
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.475189480798562e-05,
+      "loss": 2.9844,
+      "step": 7881
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.4750457793969804e-05,
+      "loss": 3.5312,
+      "step": 7882
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.4749020612961854e-05,
+      "loss": 3.1562,
+      "step": 7883
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.4747583264978034e-05,
+      "loss": 1.8438,
+      "step": 7884
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.4746145750034636e-05,
+      "loss": 4.875,
+      "step": 7885
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.4744708068147915e-05,
+      "loss": 4.125,
+      "step": 7886
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.4743270219334155e-05,
+      "loss": 5.6562,
+      "step": 7887
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.474183220360963e-05,
+      "loss": 4.3906,
+      "step": 7888
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.474039402099063e-05,
+      "loss": 5.9375,
+      "step": 7889
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.473895567149342e-05,
+      "loss": 4.1406,
+      "step": 7890
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.47375171551343e-05,
+      "loss": 3.3438,
+      "step": 7891
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.473607847192955e-05,
+      "loss": 3.125,
+      "step": 7892
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.473463962189546e-05,
+      "loss": 3.0,
+      "step": 7893
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.473320060504831e-05,
+      "loss": 2.1562,
+      "step": 7894
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.47317614214044e-05,
+      "loss": 4.2812,
+      "step": 7895
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.473032207098002e-05,
+      "loss": 3.1562,
+      "step": 7896
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.4728882553791465e-05,
+      "loss": 4.875,
+      "step": 7897
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.4727442869855034e-05,
+      "loss": 7.0938,
+      "step": 7898
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.4726003019187025e-05,
+      "loss": 4.3438,
+      "step": 7899
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.472456300180374e-05,
+      "loss": 3.625,
+      "step": 7900
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.472312281772148e-05,
+      "loss": 2.5938,
+      "step": 7901
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.472168246695655e-05,
+      "loss": 2.9688,
+      "step": 7902
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.4720241949525254e-05,
+      "loss": 1.8906,
+      "step": 7903
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.471880126544391e-05,
+      "loss": 2.9531,
+      "step": 7904
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.4717360414728816e-05,
+      "loss": 4.0625,
+      "step": 7905
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.4715919397396284e-05,
+      "loss": 2.1406,
+      "step": 7906
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.471447821346264e-05,
+      "loss": 4.5,
+      "step": 7907
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.4713036862944186e-05,
+      "loss": 2.6562,
+      "step": 7908
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.471159534585725e-05,
+      "loss": 4.5781,
+      "step": 7909
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.471015366221815e-05,
+      "loss": 3.2188,
+      "step": 7910
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.47087118120432e-05,
+      "loss": 3.6562,
+      "step": 7911
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.470726979534873e-05,
+      "loss": 3.5312,
+      "step": 7912
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.4705827612151066e-05,
+      "loss": 5.75,
+      "step": 7913
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.470438526246653e-05,
+      "loss": 3.5,
+      "step": 7914
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.4702942746311464e-05,
+      "loss": 1.9062,
+      "step": 7915
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.470150006370219e-05,
+      "loss": 3.0,
+      "step": 7916
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.470005721465503e-05,
+      "loss": 3.5312,
+      "step": 7917
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.469861419918633e-05,
+      "loss": 4.3906,
+      "step": 7918
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.469717101731242e-05,
+      "loss": 2.0781,
+      "step": 7919
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.469572766904966e-05,
+      "loss": 3.3125,
+      "step": 7920
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.469428415441437e-05,
+      "loss": 2.8125,
+      "step": 7921
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.469284047342289e-05,
+      "loss": 2.3047,
+      "step": 7922
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.469139662609157e-05,
+      "loss": 5.4062,
+      "step": 7923
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.468995261243676e-05,
+      "loss": 2.9688,
+      "step": 7924
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.468850843247481e-05,
+      "loss": 3.9688,
+      "step": 7925
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.468706408622206e-05,
+      "loss": 3.5781,
+      "step": 7926
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.468561957369487e-05,
+      "loss": 4.2969,
+      "step": 7927
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.4684174894909585e-05,
+      "loss": 9.4688,
+      "step": 7928
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.4682730049882564e-05,
+      "loss": 3.2344,
+      "step": 7929
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.4681285038630166e-05,
+      "loss": 4.2812,
+      "step": 7930
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.467983986116875e-05,
+      "loss": 4.1797,
+      "step": 7931
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.467839451751468e-05,
+      "loss": 3.9375,
+      "step": 7932
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.467694900768432e-05,
+      "loss": 2.9062,
+      "step": 7933
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.467550333169402e-05,
+      "loss": 3.4219,
+      "step": 7934
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.467405748956016e-05,
+      "loss": 5.125,
+      "step": 7935
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.4672611481299104e-05,
+      "loss": 5.5625,
+      "step": 7936
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.4671165306927225e-05,
+      "loss": 4.7969,
+      "step": 7937
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.46697189664609e-05,
+      "loss": 4.7656,
+      "step": 7938
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.466827245991649e-05,
+      "loss": 4.9375,
+      "step": 7939
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.466682578731038e-05,
+      "loss": 4.7188,
+      "step": 7940
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.466537894865895e-05,
+      "loss": 3.125,
+      "step": 7941
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.466393194397858e-05,
+      "loss": 2.25,
+      "step": 7942
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.466248477328564e-05,
+      "loss": 3.7812,
+      "step": 7943
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.4661037436596526e-05,
+      "loss": 2.7812,
+      "step": 7944
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.4659589933927624e-05,
+      "loss": 4.6719,
+      "step": 7945
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.4658142265295314e-05,
+      "loss": 2.875,
+      "step": 7946
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.465669443071598e-05,
+      "loss": 4.0312,
+      "step": 7947
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.4655246430206034e-05,
+      "loss": 5.5312,
+      "step": 7948
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.465379826378185e-05,
+      "loss": 3.875,
+      "step": 7949
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.465234993145984e-05,
+      "loss": 3.5781,
+      "step": 7950
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.465090143325638e-05,
+      "loss": 3.7812,
+      "step": 7951
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.464945276918788e-05,
+      "loss": 4.625,
+      "step": 7952
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.4648003939270735e-05,
+      "loss": 3.5312,
+      "step": 7953
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.464655494352136e-05,
+      "loss": 3.3594,
+      "step": 7954
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.464510578195615e-05,
+      "loss": 3.5156,
+      "step": 7955
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.4643656454591505e-05,
+      "loss": 3.7188,
+      "step": 7956
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.464220696144385e-05,
+      "loss": 4.1875,
+      "step": 7957
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.464075730252958e-05,
+      "loss": 1.9375,
+      "step": 7958
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.4639307477865114e-05,
+      "loss": 2.4375,
+      "step": 7959
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.4637857487466856e-05,
+      "loss": 2.4688,
+      "step": 7960
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4636407331351244e-05,
+      "loss": 4.9844,
+      "step": 7961
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.463495700953467e-05,
+      "loss": 1.7188,
+      "step": 7962
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4633506522033564e-05,
+      "loss": 4.5469,
+      "step": 7963
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4632055868864354e-05,
+      "loss": 4.8594,
+      "step": 7964
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.463060505004346e-05,
+      "loss": 4.4375,
+      "step": 7965
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.462915406558729e-05,
+      "loss": 3.2188,
+      "step": 7966
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4627702915512294e-05,
+      "loss": 5.9531,
+      "step": 7967
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.462625159983488e-05,
+      "loss": 3.0156,
+      "step": 7968
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.46248001185715e-05,
+      "loss": 3.4688,
+      "step": 7969
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.462334847173857e-05,
+      "loss": 6.2031,
+      "step": 7970
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4621896659352534e-05,
+      "loss": 3.1562,
+      "step": 7971
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.462044468142982e-05,
+      "loss": 2.7188,
+      "step": 7972
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.461899253798688e-05,
+      "loss": 3.2344,
+      "step": 7973
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.461754022904013e-05,
+      "loss": 3.25,
+      "step": 7974
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.461608775460603e-05,
+      "loss": 2.6094,
+      "step": 7975
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.461463511470103e-05,
+      "loss": 4.9375,
+      "step": 7976
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4613182309341555e-05,
+      "loss": 2.6406,
+      "step": 7977
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4611729338544065e-05,
+      "loss": 3.5938,
+      "step": 7978
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.461027620232501e-05,
+      "loss": 1.8125,
+      "step": 7979
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.460882290070083e-05,
+      "loss": 3.0625,
+      "step": 7980
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4607369433687996e-05,
+      "loss": 2.9688,
+      "step": 7981
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4605915801302954e-05,
+      "loss": 2.4375,
+      "step": 7982
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4604462003562146e-05,
+      "loss": 4.8438,
+      "step": 7983
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.460300804048206e-05,
+      "loss": 2.875,
+      "step": 7984
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.460155391207913e-05,
+      "loss": 3.8125,
+      "step": 7985
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4600099618369835e-05,
+      "loss": 4.9062,
+      "step": 7986
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.459864515937064e-05,
+      "loss": 4.4062,
+      "step": 7987
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4597190535097995e-05,
+      "loss": 4.2812,
+      "step": 7988
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.4595735745568375e-05,
+      "loss": 4.7188,
+      "step": 7989
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.459428079079827e-05,
+      "loss": 3.5625,
+      "step": 7990
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.459282567080411e-05,
+      "loss": 3.2188,
+      "step": 7991
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.459137038560241e-05,
+      "loss": 3.6875,
+      "step": 7992
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.458991493520963e-05,
+      "loss": 4.0625,
+      "step": 7993
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.458845931964224e-05,
+      "loss": 3.6875,
+      "step": 7994
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.458700353891673e-05,
+      "loss": 3.6094,
+      "step": 7995
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.4585547593049574e-05,
+      "loss": 3.875,
+      "step": 7996
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.4584091482057254e-05,
+      "loss": 5.2344,
+      "step": 7997
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.458263520595626e-05,
+      "loss": 2.9375,
+      "step": 7998
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.4581178764763077e-05,
+      "loss": 3.3125,
+      "step": 7999
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.45797221584942e-05,
+      "loss": 2.6719,
+      "step": 8000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.457826538716611e-05,
+      "loss": 6.5469,
+      "step": 8001
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.45768084507953e-05,
+      "loss": 5.8906,
+      "step": 8002
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.4575351349398265e-05,
+      "loss": 2.5938,
+      "step": 8003
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.4573894082991504e-05,
+      "loss": 3.5312,
+      "step": 8004
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.457243665159151e-05,
+      "loss": 3.1875,
+      "step": 8005
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.457097905521479e-05,
+      "loss": 3.4219,
+      "step": 8006
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.456952129387784e-05,
+      "loss": 4.7188,
+      "step": 8007
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.456806336759718e-05,
+      "loss": 2.6406,
+      "step": 8008
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.456660527638928e-05,
+      "loss": 2.875,
+      "step": 8009
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.4565147020270685e-05,
+      "loss": 2.8906,
+      "step": 8010
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.4563688599257876e-05,
+      "loss": 1.875,
+      "step": 8011
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.4562230013367374e-05,
+      "loss": 3.5312,
+      "step": 8012
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.4560771262615705e-05,
+      "loss": 6.2969,
+      "step": 8013
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.455931234701936e-05,
+      "loss": 3.4375,
+      "step": 8014
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.455785326659488e-05,
+      "loss": 3.6562,
+      "step": 8015
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.455639402135876e-05,
+      "loss": 4.4688,
+      "step": 8016
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.455493461132754e-05,
+      "loss": 2.9062,
+      "step": 8017
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.455347503651773e-05,
+      "loss": 4.7344,
+      "step": 8018
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.4552015296945856e-05,
+      "loss": 4.4062,
+      "step": 8019
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.455055539262844e-05,
+      "loss": 3.5938,
+      "step": 8020
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.4549095323582026e-05,
+      "loss": 2.75,
+      "step": 8021
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.454763508982313e-05,
+      "loss": 5.1016,
+      "step": 8022
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.454617469136829e-05,
+      "loss": 4.5312,
+      "step": 8023
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.4544714128234025e-05,
+      "loss": 2.4062,
+      "step": 8024
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.454325340043688e-05,
+      "loss": 3.2188,
+      "step": 8025
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.454179250799341e-05,
+      "loss": 4.9531,
+      "step": 8026
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.454033145092013e-05,
+      "loss": 5.2188,
+      "step": 8027
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.4538870229233584e-05,
+      "loss": 6.5938,
+      "step": 8028
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.453740884295032e-05,
+      "loss": 6.0625,
+      "step": 8029
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.453594729208688e-05,
+      "loss": 4.4375,
+      "step": 8030
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.453448557665982e-05,
+      "loss": 3.125,
+      "step": 8031
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.453302369668567e-05,
+      "loss": 3.875,
+      "step": 8032
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.4531561652181e-05,
+      "loss": 2.2812,
+      "step": 8033
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.4530099443162346e-05,
+      "loss": 7.4844,
+      "step": 8034
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.452863706964627e-05,
+      "loss": 2.5625,
+      "step": 8035
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.452717453164932e-05,
+      "loss": 4.0,
+      "step": 8036
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.4525711829188074e-05,
+      "loss": 2.8438,
+      "step": 8037
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.452424896227906e-05,
+      "loss": 4.5469,
+      "step": 8038
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.452278593093887e-05,
+      "loss": 4.2031,
+      "step": 8039
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.452132273518405e-05,
+      "loss": 2.375,
+      "step": 8040
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.4519859375031165e-05,
+      "loss": 2.4688,
+      "step": 8041
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.451839585049679e-05,
+      "loss": 6.25,
+      "step": 8042
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.451693216159749e-05,
+      "loss": 4.8906,
+      "step": 8043
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.4515468308349826e-05,
+      "loss": 4.8125,
+      "step": 8044
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.4514004290770384e-05,
+      "loss": 6.4062,
+      "step": 8045
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.451254010887573e-05,
+      "loss": 3.3125,
+      "step": 8046
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.451107576268245e-05,
+      "loss": 3.4062,
+      "step": 8047
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.450961125220712e-05,
+      "loss": 3.2812,
+      "step": 8048
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.450814657746631e-05,
+      "loss": 4.9688,
+      "step": 8049
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.4506681738476616e-05,
+      "loss": 2.2188,
+      "step": 8050
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.4505216735254604e-05,
+      "loss": 4.8906,
+      "step": 8051
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.450375156781687e-05,
+      "loss": 3.9844,
+      "step": 8052
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4502286236180004e-05,
+      "loss": 5.0938,
+      "step": 8053
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.450082074036059e-05,
+      "loss": 2.75,
+      "step": 8054
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.449935508037522e-05,
+      "loss": 4.5938,
+      "step": 8055
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4497889256240494e-05,
+      "loss": 4.1875,
+      "step": 8056
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4496423267972995e-05,
+      "loss": 2.2812,
+      "step": 8057
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4494957115589324e-05,
+      "loss": 2.4688,
+      "step": 8058
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.449349079910608e-05,
+      "loss": 1.8438,
+      "step": 8059
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.449202431853987e-05,
+      "loss": 2.375,
+      "step": 8060
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.449055767390729e-05,
+      "loss": 2.8594,
+      "step": 8061
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.448909086522493e-05,
+      "loss": 4.3281,
+      "step": 8062
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.448762389250943e-05,
+      "loss": 5.5,
+      "step": 8063
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4486156755777366e-05,
+      "loss": 9.625,
+      "step": 8064
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4484689455045366e-05,
+      "loss": 3.6094,
+      "step": 8065
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.448322199033003e-05,
+      "loss": 3.4531,
+      "step": 8066
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.448175436164799e-05,
+      "loss": 3.5781,
+      "step": 8067
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4480286569015833e-05,
+      "loss": 4.7344,
+      "step": 8068
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.44788186124502e-05,
+      "loss": 6.2969,
+      "step": 8069
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.447735049196769e-05,
+      "loss": 3.2812,
+      "step": 8070
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.447588220758495e-05,
+      "loss": 3.0,
+      "step": 8071
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.447441375931858e-05,
+      "loss": 3.3125,
+      "step": 8072
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.447294514718521e-05,
+      "loss": 4.6719,
+      "step": 8073
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4471476371201474e-05,
+      "loss": 4.3125,
+      "step": 8074
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.447000743138399e-05,
+      "loss": 1.7812,
+      "step": 8075
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.44685383277494e-05,
+      "loss": 4.4688,
+      "step": 8076
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4467069060314325e-05,
+      "loss": 3.1875,
+      "step": 8077
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4465599629095396e-05,
+      "loss": 4.6094,
+      "step": 8078
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.446413003410926e-05,
+      "loss": 2.3125,
+      "step": 8079
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4462660275372554e-05,
+      "loss": 2.7812,
+      "step": 8080
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.446119035290191e-05,
+      "loss": 3.1562,
+      "step": 8081
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.445972026671397e-05,
+      "loss": 2.7344,
+      "step": 8082
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.445825001682538e-05,
+      "loss": 3.6094,
+      "step": 8083
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.445677960325279e-05,
+      "loss": 3.0703,
+      "step": 8084
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4455309026012836e-05,
+      "loss": 4.4688,
+      "step": 8085
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.445383828512217e-05,
+      "loss": 2.25,
+      "step": 8086
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.445236738059745e-05,
+      "loss": 3.1406,
+      "step": 8087
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.445089631245532e-05,
+      "loss": 2.5156,
+      "step": 8088
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.444942508071243e-05,
+      "loss": 4.5312,
+      "step": 8089
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4447953685385455e-05,
+      "loss": 5.5781,
+      "step": 8090
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.444648212649103e-05,
+      "loss": 1.375,
+      "step": 8091
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.444501040404583e-05,
+      "loss": 3.0312,
+      "step": 8092
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.444353851806651e-05,
+      "loss": 1.625,
+      "step": 8093
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4442066468569736e-05,
+      "loss": 3.75,
+      "step": 8094
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4440594255572175e-05,
+      "loss": 5.375,
+      "step": 8095
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.443912187909049e-05,
+      "loss": 2.1719,
+      "step": 8096
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.443764933914136e-05,
+      "loss": 2.2812,
+      "step": 8097
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4436176635741434e-05,
+      "loss": 5.2188,
+      "step": 8098
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4434703768907404e-05,
+      "loss": 3.875,
+      "step": 8099
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.443323073865594e-05,
+      "loss": 4.3281,
+      "step": 8100
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.443175754500372e-05,
+      "loss": 2.2188,
+      "step": 8101
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.443028418796742e-05,
+      "loss": 2.8125,
+      "step": 8102
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.442881066756371e-05,
+      "loss": 4.6875,
+      "step": 8103
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4427336983809286e-05,
+      "loss": 3.4219,
+      "step": 8104
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4425863136720825e-05,
+      "loss": 2.7812,
+      "step": 8105
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.442438912631502e-05,
+      "loss": 1.9062,
+      "step": 8106
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4422914952608554e-05,
+      "loss": 1.9844,
+      "step": 8107
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4421440615618116e-05,
+      "loss": 4.9531,
+      "step": 8108
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.44199661153604e-05,
+      "loss": 3.6562,
+      "step": 8109
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.441849145185209e-05,
+      "loss": 3.4688,
+      "step": 8110
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4417016625109885e-05,
+      "loss": 4.3281,
+      "step": 8111
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.441554163515048e-05,
+      "loss": 2.3594,
+      "step": 8112
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.441406648199059e-05,
+      "loss": 4.0938,
+      "step": 8113
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.441259116564689e-05,
+      "loss": 4.5625,
+      "step": 8114
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4411115686136106e-05,
+      "loss": 2.5,
+      "step": 8115
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.440964004347493e-05,
+      "loss": 5.5156,
+      "step": 8116
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.440816423768007e-05,
+      "loss": 7.0625,
+      "step": 8117
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.440668826876822e-05,
+      "loss": 3.8125,
+      "step": 8118
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.440521213675611e-05,
+      "loss": 4.0938,
+      "step": 8119
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.440373584166045e-05,
+      "loss": 3.2109,
+      "step": 8120
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4402259383497943e-05,
+      "loss": 2.3438,
+      "step": 8121
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.440078276228531e-05,
+      "loss": 5.3125,
+      "step": 8122
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.439930597803926e-05,
+      "loss": 4.3438,
+      "step": 8123
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4397829030776526e-05,
+      "loss": 5.2812,
+      "step": 8124
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.439635192051382e-05,
+      "loss": 3.9375,
+      "step": 8125
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4394874647267864e-05,
+      "loss": 2.7812,
+      "step": 8126
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.439339721105538e-05,
+      "loss": 2.0938,
+      "step": 8127
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.439191961189311e-05,
+      "loss": 3.2812,
+      "step": 8128
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4390441849797765e-05,
+      "loss": 3.2344,
+      "step": 8129
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.438896392478608e-05,
+      "loss": 1.375,
+      "step": 8130
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.438748583687479e-05,
+      "loss": 3.9219,
+      "step": 8131
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.438600758608062e-05,
+      "loss": 4.625,
+      "step": 8132
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.438452917242032e-05,
+      "loss": 3.1094,
+      "step": 8133
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.438305059591061e-05,
+      "loss": 4.9688,
+      "step": 8134
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4381571856568245e-05,
+      "loss": 4.0625,
+      "step": 8135
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4380092954409957e-05,
+      "loss": 4.3125,
+      "step": 8136
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4378613889452485e-05,
+      "loss": 3.125,
+      "step": 8137
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.437713466171259e-05,
+      "loss": 3.0,
+      "step": 8138
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.437565527120701e-05,
+      "loss": 4.75,
+      "step": 8139
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.437417571795248e-05,
+      "loss": 3.6562,
+      "step": 8140
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4372696001965766e-05,
+      "loss": 2.5938,
+      "step": 8141
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.437121612326362e-05,
+      "loss": 2.1875,
+      "step": 8142
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.436973608186279e-05,
+      "loss": 3.3438,
+      "step": 8143
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.436825587778003e-05,
+      "loss": 4.5312,
+      "step": 8144
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.436677551103211e-05,
+      "loss": 4.3125,
+      "step": 8145
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.436529498163578e-05,
+      "loss": 5.8906,
+      "step": 8146
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.436381428960779e-05,
+      "loss": 3.4688,
+      "step": 8147
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.436233343496493e-05,
+      "loss": 2.6562,
+      "step": 8148
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.436085241772394e-05,
+      "loss": 4.7656,
+      "step": 8149
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.43593712379016e-05,
+      "loss": 3.1875,
+      "step": 8150
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4357889895514675e-05,
+      "loss": 6.9688,
+      "step": 8151
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.435640839057994e-05,
+      "loss": 4.7656,
+      "step": 8152
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4354926723114165e-05,
+      "loss": 3.8438,
+      "step": 8153
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4353444893134114e-05,
+      "loss": 3.3125,
+      "step": 8154
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.435196290065658e-05,
+      "loss": 2.75,
+      "step": 8155
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4350480745698324e-05,
+      "loss": 4.5625,
+      "step": 8156
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.434899842827614e-05,
+      "loss": 4.5156,
+      "step": 8157
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.43475159484068e-05,
+      "loss": 3.75,
+      "step": 8158
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4346033306107105e-05,
+      "loss": 3.0312,
+      "step": 8159
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4344550501393814e-05,
+      "loss": 3.8125,
+      "step": 8160
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.434306753428373e-05,
+      "loss": 3.5625,
+      "step": 8161
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4341584404793637e-05,
+      "loss": 2.2188,
+      "step": 8162
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.434010111294033e-05,
+      "loss": 4.1875,
+      "step": 8163
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.433861765874059e-05,
+      "loss": 2.4375,
+      "step": 8164
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.433713404221123e-05,
+      "loss": 6.7188,
+      "step": 8165
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.433565026336903e-05,
+      "loss": 4.4688,
+      "step": 8166
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4334166322230796e-05,
+      "loss": 7.5938,
+      "step": 8167
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4332682218813325e-05,
+      "loss": 3.5781,
+      "step": 8168
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.433119795313342e-05,
+      "loss": 3.3906,
+      "step": 8169
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.432971352520789e-05,
+      "loss": 5.1719,
+      "step": 8170
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.432822893505353e-05,
+      "loss": 3.7188,
+      "step": 8171
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4326744182687156e-05,
+      "loss": 2.4375,
+      "step": 8172
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.4325259268125565e-05,
+      "loss": 3.3438,
+      "step": 8173
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.432377419138559e-05,
+      "loss": 2.4688,
+      "step": 8174
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4322288952484025e-05,
+      "loss": 5.3438,
+      "step": 8175
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4320803551437686e-05,
+      "loss": 2.6875,
+      "step": 8176
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4319317988263394e-05,
+      "loss": 4.0625,
+      "step": 8177
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.431783226297797e-05,
+      "loss": 2.6406,
+      "step": 8178
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.431634637559823e-05,
+      "loss": 2.5469,
+      "step": 8179
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4314860326141e-05,
+      "loss": 3.6562,
+      "step": 8180
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.431337411462309e-05,
+      "loss": 3.7969,
+      "step": 8181
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4311887741061337e-05,
+      "loss": 3.8281,
+      "step": 8182
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4310401205472573e-05,
+      "loss": 5.75,
+      "step": 8183
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.430891450787362e-05,
+      "loss": 4.4531,
+      "step": 8184
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.430742764828131e-05,
+      "loss": 4.6562,
+      "step": 8185
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4305940626712484e-05,
+      "loss": 2.4844,
+      "step": 8186
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.430445344318397e-05,
+      "loss": 2.75,
+      "step": 8187
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4302966097712596e-05,
+      "loss": 4.8438,
+      "step": 8188
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.430147859031521e-05,
+      "loss": 4.5469,
+      "step": 8189
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.429999092100866e-05,
+      "loss": 3.8438,
+      "step": 8190
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4298503089809764e-05,
+      "loss": 5.25,
+      "step": 8191
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4297015096735386e-05,
+      "loss": 3.6875,
+      "step": 8192
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.429552694180237e-05,
+      "loss": 3.1562,
+      "step": 8193
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.429403862502756e-05,
+      "loss": 3.8125,
+      "step": 8194
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.429255014642781e-05,
+      "loss": 5.2031,
+      "step": 8195
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.429106150601996e-05,
+      "loss": 3.75,
+      "step": 8196
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.428957270382087e-05,
+      "loss": 2.8594,
+      "step": 8197
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.428808373984741e-05,
+      "loss": 4.4062,
+      "step": 8198
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4286594614116405e-05,
+      "loss": 2.3594,
+      "step": 8199
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.428510532664474e-05,
+      "loss": 3.3125,
+      "step": 8200
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4283615877449256e-05,
+      "loss": 2.0,
+      "step": 8201
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4282126266546834e-05,
+      "loss": 4.7656,
+      "step": 8202
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4280636493954334e-05,
+      "loss": 4.0938,
+      "step": 8203
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.427914655968861e-05,
+      "loss": 3.2812,
+      "step": 8204
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4277656463766536e-05,
+      "loss": 3.2969,
+      "step": 8205
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.427616620620499e-05,
+      "loss": 3.9062,
+      "step": 8206
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.427467578702083e-05,
+      "loss": 9.2188,
+      "step": 8207
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4273185206230935e-05,
+      "loss": 5.2031,
+      "step": 8208
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4271694463852185e-05,
+      "loss": 2.6875,
+      "step": 8209
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.427020355990145e-05,
+      "loss": 4.25,
+      "step": 8210
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4268712494395614e-05,
+      "loss": 4.9219,
+      "step": 8211
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.426722126735155e-05,
+      "loss": 5.1875,
+      "step": 8212
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.426572987878615e-05,
+      "loss": 4.9844,
+      "step": 8213
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.426423832871629e-05,
+      "loss": 2.9375,
+      "step": 8214
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.426274661715886e-05,
+      "loss": 2.7812,
+      "step": 8215
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4261254744130754e-05,
+      "loss": 4.5,
+      "step": 8216
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.425976270964885e-05,
+      "loss": 2.9844,
+      "step": 8217
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.425827051373004e-05,
+      "loss": 2.1406,
+      "step": 8218
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.425677815639123e-05,
+      "loss": 6.4375,
+      "step": 8219
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4255285637649306e-05,
+      "loss": 3.5625,
+      "step": 8220
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.425379295752116e-05,
+      "loss": 3.2969,
+      "step": 8221
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.42523001160237e-05,
+      "loss": 2.9844,
+      "step": 8222
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.425080711317383e-05,
+      "loss": 5.1406,
+      "step": 8223
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4249313948988446e-05,
+      "loss": 2.9688,
+      "step": 8224
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.424782062348445e-05,
+      "loss": 3.8125,
+      "step": 8225
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4246327136678746e-05,
+      "loss": 4.1875,
+      "step": 8226
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.424483348858825e-05,
+      "loss": 3.5,
+      "step": 8227
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.424333967922987e-05,
+      "loss": 2.7031,
+      "step": 8228
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.424184570862052e-05,
+      "loss": 3.25,
+      "step": 8229
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4240351576777095e-05,
+      "loss": 5.8438,
+      "step": 8230
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4238857283716536e-05,
+      "loss": 2.9062,
+      "step": 8231
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4237362829455754e-05,
+      "loss": 4.6406,
+      "step": 8232
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.423586821401165e-05,
+      "loss": 3.5312,
+      "step": 8233
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.423437343740117e-05,
+      "loss": 2.0156,
+      "step": 8234
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.423287849964122e-05,
+      "loss": 4.6719,
+      "step": 8235
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4231383400748735e-05,
+      "loss": 3.375,
+      "step": 8236
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.422988814074062e-05,
+      "loss": 4.1406,
+      "step": 8237
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.422839271963383e-05,
+      "loss": 3.875,
+      "step": 8238
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.422689713744528e-05,
+      "loss": 4.0,
+      "step": 8239
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.422540139419191e-05,
+      "loss": 4.3281,
+      "step": 8240
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4223905489890636e-05,
+      "loss": 2.125,
+      "step": 8241
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.422240942455841e-05,
+      "loss": 1.875,
+      "step": 8242
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.422091319821217e-05,
+      "loss": 1.5625,
+      "step": 8243
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4219416810868845e-05,
+      "loss": 3.4844,
+      "step": 8244
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.421792026254539e-05,
+      "loss": 2.3438,
+      "step": 8245
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.421642355325873e-05,
+      "loss": 3.8438,
+      "step": 8246
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.421492668302582e-05,
+      "loss": 2.375,
+      "step": 8247
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4213429651863606e-05,
+      "loss": 3.0469,
+      "step": 8248
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.421193245978903e-05,
+      "loss": 3.6875,
+      "step": 8249
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.421043510681905e-05,
+      "loss": 3.9844,
+      "step": 8250
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4208937592970604e-05,
+      "loss": 3.75,
+      "step": 8251
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4207439918260667e-05,
+      "loss": 2.8125,
+      "step": 8252
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.420594208270618e-05,
+      "loss": 3.1406,
+      "step": 8253
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4204444086324103e-05,
+      "loss": 4.8125,
+      "step": 8254
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.420294592913139e-05,
+      "loss": 3.5312,
+      "step": 8255
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.420144761114501e-05,
+      "loss": 3.0625,
+      "step": 8256
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.419994913238193e-05,
+      "loss": 1.9688,
+      "step": 8257
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.41984504928591e-05,
+      "loss": 3.1875,
+      "step": 8258
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.4196951692593495e-05,
+      "loss": 2.7344,
+      "step": 8259
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.419545273160208e-05,
+      "loss": 3.0,
+      "step": 8260
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.419395360990183e-05,
+      "loss": 4.0,
+      "step": 8261
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.419245432750971e-05,
+      "loss": 5.0,
+      "step": 8262
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.41909548844427e-05,
+      "loss": 6.5,
+      "step": 8263
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.418945528071777e-05,
+      "loss": 5.0,
+      "step": 8264
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.41879555163519e-05,
+      "loss": 4.1406,
+      "step": 8265
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.418645559136207e-05,
+      "loss": 3.8438,
+      "step": 8266
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.418495550576525e-05,
+      "loss": 2.0938,
+      "step": 8267
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.418345525957844e-05,
+      "loss": 4.75,
+      "step": 8268
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.418195485281862e-05,
+      "loss": 2.1875,
+      "step": 8269
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.418045428550278e-05,
+      "loss": 3.375,
+      "step": 8270
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.4178953557647885e-05,
+      "loss": 2.3281,
+      "step": 8271
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.417745266927095e-05,
+      "loss": 4.6875,
+      "step": 8272
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.417595162038895e-05,
+      "loss": 3.4688,
+      "step": 8273
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.4174450411018895e-05,
+      "loss": 2.5156,
+      "step": 8274
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.417294904117777e-05,
+      "loss": 2.5469,
+      "step": 8275
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.4171447510882574e-05,
+      "loss": 3.7812,
+      "step": 8276
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.416994582015031e-05,
+      "loss": 4.75,
+      "step": 8277
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.4168443968997974e-05,
+      "loss": 5.0938,
+      "step": 8278
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.4166941957442566e-05,
+      "loss": 4.8281,
+      "step": 8279
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.416543978550109e-05,
+      "loss": 2.875,
+      "step": 8280
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.416393745319056e-05,
+      "loss": 5.7812,
+      "step": 8281
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.4162434960527986e-05,
+      "loss": 3.9375,
+      "step": 8282
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.416093230753036e-05,
+      "loss": 6.25,
+      "step": 8283
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.4159429494214725e-05,
+      "loss": 3.7812,
+      "step": 8284
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.415792652059806e-05,
+      "loss": 4.4844,
+      "step": 8285
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.415642338669741e-05,
+      "loss": 3.1562,
+      "step": 8286
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.4154920092529766e-05,
+      "loss": 3.4688,
+      "step": 8287
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.4153416638112164e-05,
+      "loss": 3.9688,
+      "step": 8288
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.415191302346163e-05,
+      "loss": 4.0312,
+      "step": 8289
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.4150409248595167e-05,
+      "loss": 3.8438,
+      "step": 8290
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.414890531352981e-05,
+      "loss": 3.625,
+      "step": 8291
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.4147401218282586e-05,
+      "loss": 5.6406,
+      "step": 8292
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.414589696287053e-05,
+      "loss": 5.5469,
+      "step": 8293
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.414439254731065e-05,
+      "loss": 3.0625,
+      "step": 8294
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.414288797162e-05,
+      "loss": 4.875,
+      "step": 8295
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.4141383235815604e-05,
+      "loss": 4.625,
+      "step": 8296
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.41398783399145e-05,
+      "loss": 4.4219,
+      "step": 8297
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.4138373283933715e-05,
+      "loss": 4.7188,
+      "step": 8298
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.41368680678903e-05,
+      "loss": 3.0312,
+      "step": 8299
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.41353626918013e-05,
+      "loss": 2.3125,
+      "step": 8300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.4133857155683745e-05,
+      "loss": 4.0625,
+      "step": 8301
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.413235145955468e-05,
+      "loss": 5.0,
+      "step": 8302
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.413084560343116e-05,
+      "loss": 4.3438,
+      "step": 8303
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.412933958733023e-05,
+      "loss": 4.4531,
+      "step": 8304
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.412783341126894e-05,
+      "loss": 3.4375,
+      "step": 8305
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.412632707526434e-05,
+      "loss": 1.8281,
+      "step": 8306
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.4124820579333473e-05,
+      "loss": 6.0,
+      "step": 8307
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.4123313923493407e-05,
+      "loss": 2.5625,
+      "step": 8308
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.41218071077612e-05,
+      "loss": 3.0312,
+      "step": 8309
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.412030013215392e-05,
+      "loss": 2.6094,
+      "step": 8310
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.41187929966886e-05,
+      "loss": 4.375,
+      "step": 8311
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.4117285701382324e-05,
+      "loss": 3.9688,
+      "step": 8312
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.411577824625215e-05,
+      "loss": 4.3594,
+      "step": 8313
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.411427063131514e-05,
+      "loss": 6.2969,
+      "step": 8314
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.4112762856588366e-05,
+      "loss": 3.9688,
+      "step": 8315
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.41112549220889e-05,
+      "loss": 3.1406,
+      "step": 8316
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.4109746827833816e-05,
+      "loss": 4.0312,
+      "step": 8317
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.410823857384018e-05,
+      "loss": 3.4062,
+      "step": 8318
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.4106730160125065e-05,
+      "loss": 3.5625,
+      "step": 8319
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.410522158670556e-05,
+      "loss": 3.6719,
+      "step": 8320
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.4103712853598736e-05,
+      "loss": 3.2188,
+      "step": 8321
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.4102203960821675e-05,
+      "loss": 10.3438,
+      "step": 8322
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.410069490839146e-05,
+      "loss": 5.875,
+      "step": 8323
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.4099185696325165e-05,
+      "loss": 1.8438,
+      "step": 8324
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.40976763246399e-05,
+      "loss": 8.4688,
+      "step": 8325
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.409616679335272e-05,
+      "loss": 5.2969,
+      "step": 8326
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.409465710248074e-05,
+      "loss": 4.7812,
+      "step": 8327
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.4093147252041046e-05,
+      "loss": 3.8438,
+      "step": 8328
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.4091637242050736e-05,
+      "loss": 6.8594,
+      "step": 8329
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.4090127072526886e-05,
+      "loss": 4.1094,
+      "step": 8330
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.408861674348661e-05,
+      "loss": 4.625,
+      "step": 8331
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.4087106254947e-05,
+      "loss": 5.3906,
+      "step": 8332
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.408559560692516e-05,
+      "loss": 4.0938,
+      "step": 8333
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.408408479943819e-05,
+      "loss": 3.125,
+      "step": 8334
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.40825738325032e-05,
+      "loss": 3.6562,
+      "step": 8335
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.408106270613728e-05,
+      "loss": 4.0,
+      "step": 8336
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.407955142035755e-05,
+      "loss": 3.5,
+      "step": 8337
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.4078039975181126e-05,
+      "loss": 4.2969,
+      "step": 8338
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.407652837062511e-05,
+      "loss": 3.6875,
+      "step": 8339
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.4075016606706605e-05,
+      "loss": 3.875,
+      "step": 8340
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.4073504683442746e-05,
+      "loss": 4.9688,
+      "step": 8341
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.407199260085064e-05,
+      "loss": 3.75,
+      "step": 8342
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.407048035894741e-05,
+      "loss": 6.9688,
+      "step": 8343
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.406896795775016e-05,
+      "loss": 5.4062,
+      "step": 8344
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.406745539727603e-05,
+      "loss": 3.3281,
+      "step": 8345
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.4065942677542146e-05,
+      "loss": 1.9688,
+      "step": 8346
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.406442979856562e-05,
+      "loss": 3.375,
+      "step": 8347
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.406291676036358e-05,
+      "loss": 2.625,
+      "step": 8348
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.4061403562953174e-05,
+      "loss": 4.0938,
+      "step": 8349
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.405989020635151e-05,
+      "loss": 2.4844,
+      "step": 8350
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.4058376690575737e-05,
+      "loss": 4.6875,
+      "step": 8351
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.405686301564298e-05,
+      "loss": 3.6875,
+      "step": 8352
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.405534918157038e-05,
+      "loss": 3.9844,
+      "step": 8353
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.405383518837508e-05,
+      "loss": 4.1094,
+      "step": 8354
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.405232103607421e-05,
+      "loss": 2.4219,
+      "step": 8355
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.405080672468492e-05,
+      "loss": 4.1094,
+      "step": 8356
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4049292254224346e-05,
+      "loss": 3.4219,
+      "step": 8357
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.404777762470964e-05,
+      "loss": 3.4688,
+      "step": 8358
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.404626283615795e-05,
+      "loss": 5.4375,
+      "step": 8359
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4044747888586417e-05,
+      "loss": 3.2812,
+      "step": 8360
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4043232782012205e-05,
+      "loss": 5.4688,
+      "step": 8361
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.404171751645246e-05,
+      "loss": 3.2656,
+      "step": 8362
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4040202091924324e-05,
+      "loss": 4.75,
+      "step": 8363
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4038686508444976e-05,
+      "loss": 2.8594,
+      "step": 8364
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.403717076603156e-05,
+      "loss": 3.2188,
+      "step": 8365
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.403565486470123e-05,
+      "loss": 4.0625,
+      "step": 8366
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.403413880447117e-05,
+      "loss": 2.5312,
+      "step": 8367
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4032622585358534e-05,
+      "loss": 2.75,
+      "step": 8368
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4031106207380476e-05,
+      "loss": 3.1875,
+      "step": 8369
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4029589670554175e-05,
+      "loss": 3.625,
+      "step": 8370
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.402807297489679e-05,
+      "loss": 3.7969,
+      "step": 8371
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.40265561204255e-05,
+      "loss": 3.0312,
+      "step": 8372
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.402503910715748e-05,
+      "loss": 4.0781,
+      "step": 8373
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4023521935109894e-05,
+      "loss": 4.0,
+      "step": 8374
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.402200460429993e-05,
+      "loss": 6.1406,
+      "step": 8375
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.402048711474475e-05,
+      "loss": 2.8438,
+      "step": 8376
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.401896946646155e-05,
+      "loss": 2.9062,
+      "step": 8377
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4017451659467506e-05,
+      "loss": 2.5781,
+      "step": 8378
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4015933693779795e-05,
+      "loss": 4.1562,
+      "step": 8379
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4014415569415614e-05,
+      "loss": 5.0312,
+      "step": 8380
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.401289728639214e-05,
+      "loss": 4.4375,
+      "step": 8381
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.401137884472657e-05,
+      "loss": 4.375,
+      "step": 8382
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.400986024443608e-05,
+      "loss": 4.2188,
+      "step": 8383
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4008341485537876e-05,
+      "loss": 3.75,
+      "step": 8384
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.400682256804915e-05,
+      "loss": 2.0938,
+      "step": 8385
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.400530349198709e-05,
+      "loss": 4.7812,
+      "step": 8386
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4003784257368905e-05,
+      "loss": 3.75,
+      "step": 8387
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.400226486421179e-05,
+      "loss": 2.2812,
+      "step": 8388
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.400074531253294e-05,
+      "loss": 3.1875,
+      "step": 8389
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3999225602349564e-05,
+      "loss": 7.125,
+      "step": 8390
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3997705733678867e-05,
+      "loss": 3.6875,
+      "step": 8391
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.399618570653806e-05,
+      "loss": 3.5,
+      "step": 8392
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3994665520944344e-05,
+      "loss": 2.6562,
+      "step": 8393
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.399314517691493e-05,
+      "loss": 5.6406,
+      "step": 8394
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3991624674467034e-05,
+      "loss": 3.9375,
+      "step": 8395
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3990104013617867e-05,
+      "loss": 4.5,
+      "step": 8396
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3988583194384644e-05,
+      "loss": 5.2188,
+      "step": 8397
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.398706221678458e-05,
+      "loss": 3.5156,
+      "step": 8398
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3985541080834896e-05,
+      "loss": 2.6719,
+      "step": 8399
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.398401978655282e-05,
+      "loss": 6.1875,
+      "step": 8400
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.398249833395557e-05,
+      "loss": 4.0625,
+      "step": 8401
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.398097672306037e-05,
+      "loss": 5.9062,
+      "step": 8402
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.397945495388445e-05,
+      "loss": 2.5781,
+      "step": 8403
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.397793302644503e-05,
+      "loss": 2.25,
+      "step": 8404
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.397641094075934e-05,
+      "loss": 2.9531,
+      "step": 8405
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.397488869684462e-05,
+      "loss": 5.6719,
+      "step": 8406
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.39733662947181e-05,
+      "loss": 2.3125,
+      "step": 8407
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3971843734397016e-05,
+      "loss": 4.0625,
+      "step": 8408
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3970321015898605e-05,
+      "loss": 3.1562,
+      "step": 8409
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.396879813924011e-05,
+      "loss": 3.3906,
+      "step": 8410
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3967275104438763e-05,
+      "loss": 4.8438,
+      "step": 8411
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.396575191151181e-05,
+      "loss": 3.6875,
+      "step": 8412
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3964228560476496e-05,
+      "loss": 4.125,
+      "step": 8413
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3962705051350064e-05,
+      "loss": 3.9062,
+      "step": 8414
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.396118138414976e-05,
+      "loss": 2.75,
+      "step": 8415
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.395965755889285e-05,
+      "loss": 3.2656,
+      "step": 8416
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.395813357559657e-05,
+      "loss": 4.3125,
+      "step": 8417
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.3956609434278174e-05,
+      "loss": 3.5,
+      "step": 8418
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.395508513495493e-05,
+      "loss": 3.4688,
+      "step": 8419
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.3953560677644075e-05,
+      "loss": 3.6875,
+      "step": 8420
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.395203606236288e-05,
+      "loss": 2.4375,
+      "step": 8421
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.39505112891286e-05,
+      "loss": 2.8281,
+      "step": 8422
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.394898635795851e-05,
+      "loss": 3.1875,
+      "step": 8423
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.394746126886985e-05,
+      "loss": 3.6562,
+      "step": 8424
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.3945936021879906e-05,
+      "loss": 2.4844,
+      "step": 8425
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.3944410617005945e-05,
+      "loss": 4.5625,
+      "step": 8426
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.394288505426523e-05,
+      "loss": 2.5781,
+      "step": 8427
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.394135933367503e-05,
+      "loss": 3.25,
+      "step": 8428
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.393983345525262e-05,
+      "loss": 1.7812,
+      "step": 8429
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.393830741901527e-05,
+      "loss": 3.8125,
+      "step": 8430
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.393678122498028e-05,
+      "loss": 3.375,
+      "step": 8431
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.393525487316489e-05,
+      "loss": 1.9375,
+      "step": 8432
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.393372836358641e-05,
+      "loss": 3.0781,
+      "step": 8433
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.3932201696262116e-05,
+      "loss": 3.2188,
+      "step": 8434
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.393067487120928e-05,
+      "loss": 3.5469,
+      "step": 8435
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.3929147888445196e-05,
+      "loss": 1.6562,
+      "step": 8436
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.392762074798716e-05,
+      "loss": 2.8438,
+      "step": 8437
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.3926093449852444e-05,
+      "loss": 2.5,
+      "step": 8438
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.3924565994058346e-05,
+      "loss": 4.1875,
+      "step": 8439
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.3923038380622154e-05,
+      "loss": 4.4375,
+      "step": 8440
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.392151060956118e-05,
+      "loss": 4.25,
+      "step": 8441
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.391998268089269e-05,
+      "loss": 4.0938,
+      "step": 8442
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.3918454594634013e-05,
+      "loss": 3.1875,
+      "step": 8443
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.3916926350802434e-05,
+      "loss": 3.0,
+      "step": 8444
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.391539794941525e-05,
+      "loss": 4.6094,
+      "step": 8445
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.3913869390489774e-05,
+      "loss": 4.7812,
+      "step": 8446
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.39123406740433e-05,
+      "loss": 3.3125,
+      "step": 8447
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.391081180009315e-05,
+      "loss": 3.9375,
+      "step": 8448
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.390928276865662e-05,
+      "loss": 2.7969,
+      "step": 8449
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.390775357975102e-05,
+      "loss": 4.2188,
+      "step": 8450
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.390622423339368e-05,
+      "loss": 2.3438,
+      "step": 8451
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.390469472960188e-05,
+      "loss": 4.0312,
+      "step": 8452
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.390316506839297e-05,
+      "loss": 3.25,
+      "step": 8453
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.390163524978426e-05,
+      "loss": 4.0312,
+      "step": 8454
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.390010527379305e-05,
+      "loss": 2.8125,
+      "step": 8455
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.389857514043667e-05,
+      "loss": 2.3125,
+      "step": 8456
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.3897044849732466e-05,
+      "loss": 2.9375,
+      "step": 8457
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.389551440169773e-05,
+      "loss": 3.9375,
+      "step": 8458
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.389398379634981e-05,
+      "loss": 5.2031,
+      "step": 8459
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.389245303370602e-05,
+      "loss": 2.7812,
+      "step": 8460
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.38909221137837e-05,
+      "loss": 3.75,
+      "step": 8461
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.3889391036600175e-05,
+      "loss": 4.3594,
+      "step": 8462
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.388785980217279e-05,
+      "loss": 5.8594,
+      "step": 8463
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.388632841051886e-05,
+      "loss": 3.8438,
+      "step": 8464
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.388479686165574e-05,
+      "loss": 3.7812,
+      "step": 8465
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.388326515560076e-05,
+      "loss": 4.5781,
+      "step": 8466
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.3881733292371265e-05,
+      "loss": 3.25,
+      "step": 8467
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.38802012719846e-05,
+      "loss": 3.1875,
+      "step": 8468
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.3878669094458106e-05,
+      "loss": 4.5781,
+      "step": 8469
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.387713675980912e-05,
+      "loss": 5.5312,
+      "step": 8470
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.387560426805501e-05,
+      "loss": 6.1406,
+      "step": 8471
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.3874071619213104e-05,
+      "loss": 4.6719,
+      "step": 8472
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.387253881330077e-05,
+      "loss": 6.0625,
+      "step": 8473
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.387100585033535e-05,
+      "loss": 3.3125,
+      "step": 8474
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.3869472730334204e-05,
+      "loss": 3.7969,
+      "step": 8475
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.386793945331469e-05,
+      "loss": 2.375,
+      "step": 8476
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.386640601929416e-05,
+      "loss": 2.8438,
+      "step": 8477
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.386487242828998e-05,
+      "loss": 3.8906,
+      "step": 8478
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.386333868031951e-05,
+      "loss": 4.4688,
+      "step": 8479
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.3861804775400115e-05,
+      "loss": 5.2812,
+      "step": 8480
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.386027071354916e-05,
+      "loss": 2.0938,
+      "step": 8481
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.385873649478401e-05,
+      "loss": 1.9531,
+      "step": 8482
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.385720211912204e-05,
+      "loss": 2.375,
+      "step": 8483
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.385566758658062e-05,
+      "loss": 3.9219,
+      "step": 8484
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.3854132897177114e-05,
+      "loss": 3.8125,
+      "step": 8485
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.38525980509289e-05,
+      "loss": 4.8438,
+      "step": 8486
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.385106304785337e-05,
+      "loss": 3.7188,
+      "step": 8487
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.384952788796788e-05,
+      "loss": 4.0312,
+      "step": 8488
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.384799257128981e-05,
+      "loss": 3.4062,
+      "step": 8489
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.384645709783656e-05,
+      "loss": 2.8594,
+      "step": 8490
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.38449214676255e-05,
+      "loss": 3.5625,
+      "step": 8491
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.384338568067402e-05,
+      "loss": 3.8125,
+      "step": 8492
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.3841849736999504e-05,
+      "loss": 5.4375,
+      "step": 8493
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.384031363661934e-05,
+      "loss": 3.8125,
+      "step": 8494
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.3838777379550926e-05,
+      "loss": 5.5156,
+      "step": 8495
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.3837240965811644e-05,
+      "loss": 3.875,
+      "step": 8496
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.3835704395418896e-05,
+      "loss": 3.0469,
+      "step": 8497
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.383416766839007e-05,
+      "loss": 2.1875,
+      "step": 8498
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.383263078474257e-05,
+      "loss": 3.75,
+      "step": 8499
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.38310937444938e-05,
+      "loss": 1.9688,
+      "step": 8500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.382955654766114e-05,
+      "loss": 2.7969,
+      "step": 8501
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.382801919426202e-05,
+      "loss": 1.8906,
+      "step": 8502
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.382648168431382e-05,
+      "loss": 4.5938,
+      "step": 8503
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.382494401783396e-05,
+      "loss": 6.0781,
+      "step": 8504
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.382340619483985e-05,
+      "loss": 3.8906,
+      "step": 8505
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.382186821534889e-05,
+      "loss": 3.0,
+      "step": 8506
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.382033007937851e-05,
+      "loss": 1.9688,
+      "step": 8507
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.38187917869461e-05,
+      "loss": 4.9219,
+      "step": 8508
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.381725333806909e-05,
+      "loss": 4.375,
+      "step": 8509
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3815714732764896e-05,
+      "loss": 1.7812,
+      "step": 8510
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.381417597105093e-05,
+      "loss": 4.1094,
+      "step": 8511
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.381263705294462e-05,
+      "loss": 4.3281,
+      "step": 8512
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.381109797846338e-05,
+      "loss": 2.1875,
+      "step": 8513
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.380955874762464e-05,
+      "loss": 5.4062,
+      "step": 8514
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.380801936044583e-05,
+      "loss": 4.4062,
+      "step": 8515
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.380647981694437e-05,
+      "loss": 3.7656,
+      "step": 8516
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.380494011713769e-05,
+      "loss": 3.375,
+      "step": 8517
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.380340026104323e-05,
+      "loss": 6.0156,
+      "step": 8518
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.380186024867841e-05,
+      "loss": 6.7812,
+      "step": 8519
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.380032008006067e-05,
+      "loss": 4.3125,
+      "step": 8520
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.379877975520745e-05,
+      "loss": 3.375,
+      "step": 8521
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3797239274136185e-05,
+      "loss": 5.0938,
+      "step": 8522
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3795698636864314e-05,
+      "loss": 2.75,
+      "step": 8523
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.379415784340928e-05,
+      "loss": 2.4688,
+      "step": 8524
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3792616893788525e-05,
+      "loss": 3.1719,
+      "step": 8525
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.37910757880195e-05,
+      "loss": 4.5,
+      "step": 8526
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3789534526119653e-05,
+      "loss": 2.2812,
+      "step": 8527
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3787993108106425e-05,
+      "loss": 2.5312,
+      "step": 8528
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.378645153399727e-05,
+      "loss": 4.7969,
+      "step": 8529
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.378490980380963e-05,
+      "loss": 3.9688,
+      "step": 8530
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.378336791756098e-05,
+      "loss": 1.9688,
+      "step": 8531
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.378182587526876e-05,
+      "loss": 3.125,
+      "step": 8532
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.378028367695044e-05,
+      "loss": 3.2812,
+      "step": 8533
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.377874132262347e-05,
+      "loss": 3.25,
+      "step": 8534
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.377719881230531e-05,
+      "loss": 3.9375,
+      "step": 8535
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3775656146013425e-05,
+      "loss": 3.4688,
+      "step": 8536
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.377411332376529e-05,
+      "loss": 2.9062,
+      "step": 8537
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.377257034557835e-05,
+      "loss": 3.6875,
+      "step": 8538
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3771027211470096e-05,
+      "loss": 3.4062,
+      "step": 8539
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.376948392145798e-05,
+      "loss": 2.5,
+      "step": 8540
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.376794047555949e-05,
+      "loss": 3.1719,
+      "step": 8541
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.376639687379209e-05,
+      "loss": 6.2344,
+      "step": 8542
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3764853116173256e-05,
+      "loss": 4.0312,
+      "step": 8543
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.376330920272047e-05,
+      "loss": 3.8438,
+      "step": 8544
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.37617651334512e-05,
+      "loss": 4.0938,
+      "step": 8545
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.376022090838294e-05,
+      "loss": 6.9062,
+      "step": 8546
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.375867652753316e-05,
+      "loss": 3.1562,
+      "step": 8547
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3757131990919356e-05,
+      "loss": 4.6562,
+      "step": 8548
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3755587298559006e-05,
+      "loss": 3.0312,
+      "step": 8549
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.37540424504696e-05,
+      "loss": 2.6875,
+      "step": 8550
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3752497446668626e-05,
+      "loss": 5.9375,
+      "step": 8551
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3750952287173576e-05,
+      "loss": 5.9062,
+      "step": 8552
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.374940697200194e-05,
+      "loss": 2.9062,
+      "step": 8553
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.374786150117123e-05,
+      "loss": 3.9375,
+      "step": 8554
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3746315874698914e-05,
+      "loss": 4.625,
+      "step": 8555
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3744770092602514e-05,
+      "loss": 3.8438,
+      "step": 8556
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3743224154899515e-05,
+      "loss": 3.25,
+      "step": 8557
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3741678061607426e-05,
+      "loss": 3.8281,
+      "step": 8558
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.374013181274375e-05,
+      "loss": 3.0,
+      "step": 8559
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3738585408325996e-05,
+      "loss": 4.2969,
+      "step": 8560
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.373703884837166e-05,
+      "loss": 6.9844,
+      "step": 8561
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.373549213289826e-05,
+      "loss": 3.2188,
+      "step": 8562
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3733945261923304e-05,
+      "loss": 4.0156,
+      "step": 8563
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.37323982354643e-05,
+      "loss": 4.2812,
+      "step": 8564
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3730851053538776e-05,
+      "loss": 3.125,
+      "step": 8565
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.372930371616423e-05,
+      "loss": 3.375,
+      "step": 8566
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.3727756223358196e-05,
+      "loss": 1.7188,
+      "step": 8567
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.372620857513818e-05,
+      "loss": 4.4688,
+      "step": 8568
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.372466077152171e-05,
+      "loss": 5.0625,
+      "step": 8569
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.372311281252631e-05,
+      "loss": 2.8438,
+      "step": 8570
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.37215646981695e-05,
+      "loss": 5.0625,
+      "step": 8571
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.372001642846881e-05,
+      "loss": 4.0469,
+      "step": 8572
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3718468003441764e-05,
+      "loss": 3.6875,
+      "step": 8573
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.37169194231059e-05,
+      "loss": 1.9688,
+      "step": 8574
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.371537068747874e-05,
+      "loss": 4.2188,
+      "step": 8575
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.371382179657783e-05,
+      "loss": 3.6406,
+      "step": 8576
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.371227275042069e-05,
+      "loss": 3.6562,
+      "step": 8577
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.371072354902486e-05,
+      "loss": 2.8438,
+      "step": 8578
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3709174192407886e-05,
+      "loss": 3.7188,
+      "step": 8579
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.370762468058731e-05,
+      "loss": 2.8438,
+      "step": 8580
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3706075013580666e-05,
+      "loss": 2.1406,
+      "step": 8581
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.37045251914055e-05,
+      "loss": 4.5,
+      "step": 8582
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.370297521407937e-05,
+      "loss": 5.7031,
+      "step": 8583
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.37014250816198e-05,
+      "loss": 2.8125,
+      "step": 8584
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.369987479404436e-05,
+      "loss": 2.7812,
+      "step": 8585
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.369832435137059e-05,
+      "loss": 2.9375,
+      "step": 8586
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.369677375361605e-05,
+      "loss": 2.0312,
+      "step": 8587
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.369522300079829e-05,
+      "loss": 3.6406,
+      "step": 8588
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3693672092934865e-05,
+      "loss": 2.9062,
+      "step": 8589
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.369212103004333e-05,
+      "loss": 4.6094,
+      "step": 8590
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3690569812141255e-05,
+      "loss": 4.1094,
+      "step": 8591
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.36890184392462e-05,
+      "loss": 4.875,
+      "step": 8592
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3687466911375724e-05,
+      "loss": 2.7188,
+      "step": 8593
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.368591522854739e-05,
+      "loss": 3.0938,
+      "step": 8594
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.368436339077876e-05,
+      "loss": 3.4062,
+      "step": 8595
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3682811398087425e-05,
+      "loss": 2.375,
+      "step": 8596
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3681259250490926e-05,
+      "loss": 3.0312,
+      "step": 8597
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.367970694800686e-05,
+      "loss": 2.7188,
+      "step": 8598
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3678154490652786e-05,
+      "loss": 5.2969,
+      "step": 8599
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.367660187844629e-05,
+      "loss": 2.2188,
+      "step": 8600
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.3675049111404944e-05,
+      "loss": 4.0938,
+      "step": 8601
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.367349618954632e-05,
+      "loss": 4.2031,
+      "step": 8602
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.367194311288801e-05,
+      "loss": 2.7969,
+      "step": 8603
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.367038988144758e-05,
+      "loss": 2.6719,
+      "step": 8604
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.366883649524264e-05,
+      "loss": 3.3125,
+      "step": 8605
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.366728295429076e-05,
+      "loss": 2.25,
+      "step": 8606
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.366572925860953e-05,
+      "loss": 2.7891,
+      "step": 8607
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.366417540821654e-05,
+      "loss": 6.0156,
+      "step": 8608
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.366262140312938e-05,
+      "loss": 4.3438,
+      "step": 8609
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.366106724336565e-05,
+      "loss": 2.0312,
+      "step": 8610
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.365951292894293e-05,
+      "loss": 7.0,
+      "step": 8611
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.365795845987883e-05,
+      "loss": 5.3438,
+      "step": 8612
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.365640383619095e-05,
+      "loss": 5.9688,
+      "step": 8613
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.365484905789687e-05,
+      "loss": 5.5469,
+      "step": 8614
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.3653294125014225e-05,
+      "loss": 3.6719,
+      "step": 8615
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.3651739037560594e-05,
+      "loss": 4.3906,
+      "step": 8616
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.365018379555358e-05,
+      "loss": 2.0312,
+      "step": 8617
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.3648628399010806e-05,
+      "loss": 5.8125,
+      "step": 8618
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.364707284794987e-05,
+      "loss": 3.7188,
+      "step": 8619
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.364551714238839e-05,
+      "loss": 2.9375,
+      "step": 8620
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.364396128234397e-05,
+      "loss": 2.1875,
+      "step": 8621
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.364240526783423e-05,
+      "loss": 3.7656,
+      "step": 8622
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.364084909887679e-05,
+      "loss": 4.9375,
+      "step": 8623
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.363929277548925e-05,
+      "loss": 1.875,
+      "step": 8624
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.363773629768925e-05,
+      "loss": 2.625,
+      "step": 8625
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.3636179665494404e-05,
+      "loss": 1.4688,
+      "step": 8626
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.363462287892233e-05,
+      "loss": 5.4375,
+      "step": 8627
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.363306593799066e-05,
+      "loss": 3.375,
+      "step": 8628
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.3631508842717016e-05,
+      "loss": 2.75,
+      "step": 8629
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.362995159311902e-05,
+      "loss": 4.1094,
+      "step": 8630
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3628394189214314e-05,
+      "loss": 2.9219,
+      "step": 8631
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3626836631020525e-05,
+      "loss": 4.4375,
+      "step": 8632
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.362527891855528e-05,
+      "loss": 3.0938,
+      "step": 8633
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.362372105183622e-05,
+      "loss": 7.9688,
+      "step": 8634
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.362216303088099e-05,
+      "loss": 2.1562,
+      "step": 8635
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.362060485570721e-05,
+      "loss": 2.8438,
+      "step": 8636
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.361904652633253e-05,
+      "loss": 3.8125,
+      "step": 8637
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3617488042774596e-05,
+      "loss": 3.0156,
+      "step": 8638
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3615929405051044e-05,
+      "loss": 3.3594,
+      "step": 8639
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.361437061317953e-05,
+      "loss": 3.0,
+      "step": 8640
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.36128116671777e-05,
+      "loss": 3.7969,
+      "step": 8641
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3611252567063184e-05,
+      "loss": 3.1094,
+      "step": 8642
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.360969331285365e-05,
+      "loss": 3.6875,
+      "step": 8643
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.360813390456675e-05,
+      "loss": 3.4844,
+      "step": 8644
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3606574342220135e-05,
+      "loss": 4.3438,
+      "step": 8645
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3605014625831465e-05,
+      "loss": 3.1875,
+      "step": 8646
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.360345475541839e-05,
+      "loss": 4.375,
+      "step": 8647
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.360189473099857e-05,
+      "loss": 4.7656,
+      "step": 8648
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.360033455258968e-05,
+      "loss": 3.5781,
+      "step": 8649
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.359877422020938e-05,
+      "loss": 3.9375,
+      "step": 8650
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3597213733875315e-05,
+      "loss": 3.8125,
+      "step": 8651
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.359565309360516e-05,
+      "loss": 7.9844,
+      "step": 8652
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3594092299416595e-05,
+      "loss": 3.2812,
+      "step": 8653
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.359253135132729e-05,
+      "loss": 5.2344,
+      "step": 8654
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3590970249354894e-05,
+      "loss": 3.1406,
+      "step": 8655
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3589408993517107e-05,
+      "loss": 3.0312,
+      "step": 8656
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.358784758383159e-05,
+      "loss": 3.9844,
+      "step": 8657
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3586286020316027e-05,
+      "loss": 4.375,
+      "step": 8658
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3584724302988086e-05,
+      "loss": 2.5,
+      "step": 8659
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.3583162431865464e-05,
+      "loss": 4.2969,
+      "step": 8660
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.358160040696582e-05,
+      "loss": 4.0625,
+      "step": 8661
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.358003822830686e-05,
+      "loss": 3.875,
+      "step": 8662
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.357847589590626e-05,
+      "loss": 3.0,
+      "step": 8663
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.35769134097817e-05,
+      "loss": 3.25,
+      "step": 8664
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.357535076995089e-05,
+      "loss": 2.3438,
+      "step": 8665
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.3573787976431506e-05,
+      "loss": 3.7656,
+      "step": 8666
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.357222502924124e-05,
+      "loss": 5.8281,
+      "step": 8667
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.3570661928397794e-05,
+      "loss": 2.125,
+      "step": 8668
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.356909867391886e-05,
+      "loss": 2.0781,
+      "step": 8669
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.356753526582212e-05,
+      "loss": 4.9844,
+      "step": 8670
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.356597170412531e-05,
+      "loss": 6.125,
+      "step": 8671
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.3564407988846096e-05,
+      "loss": 1.8438,
+      "step": 8672
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.35628441200022e-05,
+      "loss": 2.125,
+      "step": 8673
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.3561280097611317e-05,
+      "loss": 4.4688,
+      "step": 8674
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.355971592169117e-05,
+      "loss": 6.25,
+      "step": 8675
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.355815159225944e-05,
+      "loss": 5.5312,
+      "step": 8676
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.355658710933387e-05,
+      "loss": 2.9375,
+      "step": 8677
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.355502247293214e-05,
+      "loss": 6.9375,
+      "step": 8678
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.3553457683071986e-05,
+      "loss": 3.0781,
+      "step": 8679
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.3551892739771117e-05,
+      "loss": 3.2812,
+      "step": 8680
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.3550327643047244e-05,
+      "loss": 4.4375,
+      "step": 8681
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.354876239291809e-05,
+      "loss": 5.7656,
+      "step": 8682
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.3547196989401376e-05,
+      "loss": 3.1875,
+      "step": 8683
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.354563143251483e-05,
+      "loss": 3.9219,
+      "step": 8684
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.3544065722276155e-05,
+      "loss": 4.5156,
+      "step": 8685
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.3542499858703106e-05,
+      "loss": 5.1562,
+      "step": 8686
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.354093384181338e-05,
+      "loss": 3.8906,
+      "step": 8687
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.353936767162473e-05,
+      "loss": 3.4688,
+      "step": 8688
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.353780134815488e-05,
+      "loss": 2.875,
+      "step": 8689
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.3536234871421566e-05,
+      "loss": 4.0312,
+      "step": 8690
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.353466824144251e-05,
+      "loss": 4.1875,
+      "step": 8691
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.353310145823546e-05,
+      "loss": 4.4688,
+      "step": 8692
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.3531534521818146e-05,
+      "loss": 7.7031,
+      "step": 8693
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.352996743220831e-05,
+      "loss": 1.9062,
+      "step": 8694
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.3528400189423695e-05,
+      "loss": 5.2656,
+      "step": 8695
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.352683279348204e-05,
+      "loss": 2.2344,
+      "step": 8696
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.35252652444011e-05,
+      "loss": 6.3438,
+      "step": 8697
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.352369754219861e-05,
+      "loss": 3.4062,
+      "step": 8698
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.352212968689232e-05,
+      "loss": 4.0938,
+      "step": 8699
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.3520561678499986e-05,
+      "loss": 2.5938,
+      "step": 8700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.351899351703935e-05,
+      "loss": 4.4062,
+      "step": 8701
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.351742520252818e-05,
+      "loss": 4.1719,
+      "step": 8702
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.351585673498422e-05,
+      "loss": 3.0938,
+      "step": 8703
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.351428811442522e-05,
+      "loss": 2.5,
+      "step": 8704
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.3512719340868955e-05,
+      "loss": 4.0938,
+      "step": 8705
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.351115041433318e-05,
+      "loss": 3.8594,
+      "step": 8706
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.350958133483565e-05,
+      "loss": 3.0625,
+      "step": 8707
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.3508012102394145e-05,
+      "loss": 3.9688,
+      "step": 8708
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.350644271702641e-05,
+      "loss": 4.4844,
+      "step": 8709
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.350487317875022e-05,
+      "loss": 2.4375,
+      "step": 8710
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.350330348758335e-05,
+      "loss": 4.0312,
+      "step": 8711
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.350173364354356e-05,
+      "loss": 2.75,
+      "step": 8712
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.350016364664864e-05,
+      "loss": 2.8906,
+      "step": 8713
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.349859349691634e-05,
+      "loss": 4.8125,
+      "step": 8714
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.349702319436447e-05,
+      "loss": 6.0312,
+      "step": 8715
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.3495452739010775e-05,
+      "loss": 3.4375,
+      "step": 8716
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.349388213087304e-05,
+      "loss": 2.25,
+      "step": 8717
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.349231136996905e-05,
+      "loss": 3.4688,
+      "step": 8718
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.34907404563166e-05,
+      "loss": 1.3594,
+      "step": 8719
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.348916938993346e-05,
+      "loss": 4.2031,
+      "step": 8720
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.348759817083742e-05,
+      "loss": 1.1406,
+      "step": 8721
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.348602679904627e-05,
+      "loss": 3.7188,
+      "step": 8722
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.3484455274577796e-05,
+      "loss": 3.0312,
+      "step": 8723
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.34828835974498e-05,
+      "loss": 3.9062,
+      "step": 8724
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.3481311767680066e-05,
+      "loss": 4.1562,
+      "step": 8725
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.347973978528638e-05,
+      "loss": 3.5781,
+      "step": 8726
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.347816765028655e-05,
+      "loss": 2.7344,
+      "step": 8727
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.347659536269838e-05,
+      "loss": 3.5938,
+      "step": 8728
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.347502292253966e-05,
+      "loss": 2.1562,
+      "step": 8729
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.347345032982819e-05,
+      "loss": 3.125,
+      "step": 8730
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.3471877584581785e-05,
+      "loss": 3.7656,
+      "step": 8731
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.3470304686818245e-05,
+      "loss": 3.4844,
+      "step": 8732
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.3468731636555375e-05,
+      "loss": 3.7812,
+      "step": 8733
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.346715843381098e-05,
+      "loss": 2.7188,
+      "step": 8734
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.346558507860288e-05,
+      "loss": 3.125,
+      "step": 8735
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.346401157094888e-05,
+      "loss": 3.4062,
+      "step": 8736
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.3462437910866795e-05,
+      "loss": 4.875,
+      "step": 8737
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.3460864098374445e-05,
+      "loss": 3.1875,
+      "step": 8738
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.345929013348964e-05,
+      "loss": 5.6562,
+      "step": 8739
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.345771601623021e-05,
+      "loss": 3.3594,
+      "step": 8740
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.3456141746613964e-05,
+      "loss": 3.2812,
+      "step": 8741
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.345456732465873e-05,
+      "loss": 2.8438,
+      "step": 8742
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.3452992750382325e-05,
+      "loss": 3.1562,
+      "step": 8743
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.345141802380259e-05,
+      "loss": 4.0781,
+      "step": 8744
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.344984314493734e-05,
+      "loss": 3.9688,
+      "step": 8745
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.3448268113804415e-05,
+      "loss": 4.5469,
+      "step": 8746
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.344669293042163e-05,
+      "loss": 3.1719,
+      "step": 8747
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.344511759480684e-05,
+      "loss": 3.3906,
+      "step": 8748
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.344354210697786e-05,
+      "loss": 4.8125,
+      "step": 8749
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.344196646695254e-05,
+      "loss": 4.3438,
+      "step": 8750
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.344039067474871e-05,
+      "loss": 3.5,
+      "step": 8751
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3438814730384205e-05,
+      "loss": 3.4375,
+      "step": 8752
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3437238633876885e-05,
+      "loss": 3.8438,
+      "step": 8753
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3435662385244576e-05,
+      "loss": 4.75,
+      "step": 8754
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3434085984505125e-05,
+      "loss": 2.1562,
+      "step": 8755
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.343250943167638e-05,
+      "loss": 5.7656,
+      "step": 8756
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3430932726776196e-05,
+      "loss": 2.8125,
+      "step": 8757
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.342935586982242e-05,
+      "loss": 1.875,
+      "step": 8758
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.34277788608329e-05,
+      "loss": 9.0938,
+      "step": 8759
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.34262016998255e-05,
+      "loss": 6.6719,
+      "step": 8760
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3424624386818064e-05,
+      "loss": 8.0781,
+      "step": 8761
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3423046921828444e-05,
+      "loss": 9.0625,
+      "step": 8762
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.342146930487451e-05,
+      "loss": 2.4375,
+      "step": 8763
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.341989153597412e-05,
+      "loss": 5.9375,
+      "step": 8764
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.341831361514514e-05,
+      "loss": 3.9062,
+      "step": 8765
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3416735542405426e-05,
+      "loss": 6.3438,
+      "step": 8766
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.341515731777285e-05,
+      "loss": 4.6875,
+      "step": 8767
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.341357894126527e-05,
+      "loss": 5.8594,
+      "step": 8768
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.341200041290057e-05,
+      "loss": 4.9844,
+      "step": 8769
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.34104217326966e-05,
+      "loss": 2.8125,
+      "step": 8770
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3408842900671254e-05,
+      "loss": 3.4375,
+      "step": 8771
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3407263916842394e-05,
+      "loss": 4.3906,
+      "step": 8772
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.340568478122791e-05,
+      "loss": 4.1875,
+      "step": 8773
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3404105493845654e-05,
+      "loss": 2.9375,
+      "step": 8774
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.340252605471352e-05,
+      "loss": 5.2812,
+      "step": 8775
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.340094646384939e-05,
+      "loss": 2.5625,
+      "step": 8776
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.339936672127115e-05,
+      "loss": 4.2031,
+      "step": 8777
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3397786826996683e-05,
+      "loss": 4.4375,
+      "step": 8778
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.339620678104387e-05,
+      "loss": 5.9688,
+      "step": 8779
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3394626583430593e-05,
+      "loss": 3.0625,
+      "step": 8780
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.3393046234174757e-05,
+      "loss": 2.5469,
+      "step": 8781
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.339146573329425e-05,
+      "loss": 3.8438,
+      "step": 8782
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.338988508080696e-05,
+      "loss": 4.125,
+      "step": 8783
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.338830427673077e-05,
+      "loss": 3.2812,
+      "step": 8784
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.338672332108361e-05,
+      "loss": 2.6562,
+      "step": 8785
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3385142213883336e-05,
+      "loss": 2.4375,
+      "step": 8786
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.338356095514789e-05,
+      "loss": 6.4062,
+      "step": 8787
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.338197954489514e-05,
+      "loss": 2.7812,
+      "step": 8788
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.338039798314301e-05,
+      "loss": 2.7656,
+      "step": 8789
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.337881626990939e-05,
+      "loss": 3.4375,
+      "step": 8790
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.33772344052122e-05,
+      "loss": 4.4219,
+      "step": 8791
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.337565238906935e-05,
+      "loss": 6.7812,
+      "step": 8792
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.337407022149874e-05,
+      "loss": 4.3438,
+      "step": 8793
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.337248790251828e-05,
+      "loss": 2.4062,
+      "step": 8794
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.337090543214589e-05,
+      "loss": 3.1719,
+      "step": 8795
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.336932281039949e-05,
+      "loss": 5.7031,
+      "step": 8796
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.336774003729698e-05,
+      "loss": 1.6875,
+      "step": 8797
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3366157112856307e-05,
+      "loss": 2.5625,
+      "step": 8798
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3364574037095365e-05,
+      "loss": 4.2188,
+      "step": 8799
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3362990810032087e-05,
+      "loss": 3.4062,
+      "step": 8800
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.33614074316844e-05,
+      "loss": 1.5938,
+      "step": 8801
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.335982390207022e-05,
+      "loss": 3.0625,
+      "step": 8802
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.335824022120749e-05,
+      "loss": 2.7969,
+      "step": 8803
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.335665638911412e-05,
+      "loss": 4.3125,
+      "step": 8804
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.335507240580805e-05,
+      "loss": 4.5625,
+      "step": 8805
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.335348827130722e-05,
+      "loss": 7.2969,
+      "step": 8806
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.335190398562955e-05,
+      "loss": 2.3125,
+      "step": 8807
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.335031954879298e-05,
+      "loss": 3.0625,
+      "step": 8808
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.334873496081546e-05,
+      "loss": 5.6094,
+      "step": 8809
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3347150221714914e-05,
+      "loss": 3.375,
+      "step": 8810
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3345565331509294e-05,
+      "loss": 2.4531,
+      "step": 8811
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.334398029021653e-05,
+      "loss": 4.625,
+      "step": 8812
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.334239509785458e-05,
+      "loss": 3.5312,
+      "step": 8813
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3340809754441386e-05,
+      "loss": 4.5781,
+      "step": 8814
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.333922425999489e-05,
+      "loss": 3.3125,
+      "step": 8815
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.333763861453305e-05,
+      "loss": 4.5,
+      "step": 8816
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.333605281807381e-05,
+      "loss": 2.5312,
+      "step": 8817
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3334466870635125e-05,
+      "loss": 4.9688,
+      "step": 8818
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.333288077223495e-05,
+      "loss": 3.6094,
+      "step": 8819
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.333129452289124e-05,
+      "loss": 2.1562,
+      "step": 8820
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.332970812262197e-05,
+      "loss": 5.5156,
+      "step": 8821
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.332812157144507e-05,
+      "loss": 3.8125,
+      "step": 8822
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.332653486937852e-05,
+      "loss": 4.0156,
+      "step": 8823
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.332494801644028e-05,
+      "loss": 2.5312,
+      "step": 8824
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.332336101264832e-05,
+      "loss": 5.625,
+      "step": 8825
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.332177385802061e-05,
+      "loss": 3.1562,
+      "step": 8826
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.33201865525751e-05,
+      "loss": 2.875,
+      "step": 8827
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.331859909632977e-05,
+      "loss": 3.0781,
+      "step": 8828
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.331701148930259e-05,
+      "loss": 2.9531,
+      "step": 8829
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.331542373151154e-05,
+      "loss": 5.0625,
+      "step": 8830
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.331383582297459e-05,
+      "loss": 2.0938,
+      "step": 8831
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.331224776370972e-05,
+      "loss": 2.5938,
+      "step": 8832
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.331065955373491e-05,
+      "loss": 4.5781,
+      "step": 8833
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.330907119306813e-05,
+      "loss": 3.0469,
+      "step": 8834
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3307482681727376e-05,
+      "loss": 2.9375,
+      "step": 8835
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.330589401973062e-05,
+      "loss": 3.625,
+      "step": 8836
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3304305207095855e-05,
+      "loss": 4.6875,
+      "step": 8837
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.330271624384106e-05,
+      "loss": 5.0625,
+      "step": 8838
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.330112712998424e-05,
+      "loss": 2.125,
+      "step": 8839
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3299537865543365e-05,
+      "loss": 2.0938,
+      "step": 8840
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.329794845053645e-05,
+      "loss": 6.1406,
+      "step": 8841
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.329635888498146e-05,
+      "loss": 2.3906,
+      "step": 8842
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.329476916889642e-05,
+      "loss": 3.5625,
+      "step": 8843
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3293179302299314e-05,
+      "loss": 6.25,
+      "step": 8844
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.329158928520813e-05,
+      "loss": 5.0156,
+      "step": 8845
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.328999911764089e-05,
+      "loss": 3.125,
+      "step": 8846
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.328840879961559e-05,
+      "loss": 6.4844,
+      "step": 8847
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3286818331150225e-05,
+      "loss": 1.7812,
+      "step": 8848
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.328522771226281e-05,
+      "loss": 4.625,
+      "step": 8849
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.328363694297135e-05,
+      "loss": 6.5938,
+      "step": 8850
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.328204602329385e-05,
+      "loss": 4.0781,
+      "step": 8851
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3280454953248326e-05,
+      "loss": 3.0,
+      "step": 8852
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3278863732852794e-05,
+      "loss": 2.9219,
+      "step": 8853
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3277272362125265e-05,
+      "loss": 3.0938,
+      "step": 8854
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.327568084108375e-05,
+      "loss": 3.125,
+      "step": 8855
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3274089169746274e-05,
+      "loss": 7.9844,
+      "step": 8856
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3272497348130855e-05,
+      "loss": 4.6016,
+      "step": 8857
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.327090537625551e-05,
+      "loss": 3.7969,
+      "step": 8858
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.326931325413827e-05,
+      "loss": 3.4375,
+      "step": 8859
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.326772098179715e-05,
+      "loss": 1.6562,
+      "step": 8860
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.326612855925018e-05,
+      "loss": 7.2656,
+      "step": 8861
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.326453598651539e-05,
+      "loss": 4.0938,
+      "step": 8862
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3262943263610816e-05,
+      "loss": 3.7031,
+      "step": 8863
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.326135039055448e-05,
+      "loss": 2.4688,
+      "step": 8864
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.325975736736441e-05,
+      "loss": 2.5156,
+      "step": 8865
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.325816419405866e-05,
+      "loss": 3.9531,
+      "step": 8866
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.325657087065525e-05,
+      "loss": 4.5,
+      "step": 8867
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3254977397172214e-05,
+      "loss": 2.875,
+      "step": 8868
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.325338377362761e-05,
+      "loss": 6.4531,
+      "step": 8869
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.325179000003947e-05,
+      "loss": 4.5312,
+      "step": 8870
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3250196076425836e-05,
+      "loss": 3.0625,
+      "step": 8871
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.324860200280476e-05,
+      "loss": 5.4375,
+      "step": 8872
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.3247007779194276e-05,
+      "loss": 4.1875,
+      "step": 8873
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.324541340561245e-05,
+      "loss": 3.8438,
+      "step": 8874
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3243818882077314e-05,
+      "loss": 2.7969,
+      "step": 8875
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.324222420860693e-05,
+      "loss": 3.5,
+      "step": 8876
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.324062938521936e-05,
+      "loss": 3.3438,
+      "step": 8877
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.323903441193264e-05,
+      "loss": 4.2031,
+      "step": 8878
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3237439288764836e-05,
+      "loss": 2.875,
+      "step": 8879
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.323584401573401e-05,
+      "loss": 2.875,
+      "step": 8880
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.323424859285821e-05,
+      "loss": 5.2031,
+      "step": 8881
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.323265302015552e-05,
+      "loss": 2.6094,
+      "step": 8882
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3231057297643986e-05,
+      "loss": 4.2188,
+      "step": 8883
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.322946142534168e-05,
+      "loss": 4.6406,
+      "step": 8884
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.322786540326666e-05,
+      "loss": 4.0312,
+      "step": 8885
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.322626923143701e-05,
+      "loss": 7.4375,
+      "step": 8886
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.322467290987079e-05,
+      "loss": 3.8594,
+      "step": 8887
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.322307643858607e-05,
+      "loss": 5.7188,
+      "step": 8888
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3221479817600934e-05,
+      "loss": 3.0,
+      "step": 8889
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.321988304693345e-05,
+      "loss": 3.1562,
+      "step": 8890
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.32182861266017e-05,
+      "loss": 4.9375,
+      "step": 8891
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.321668905662375e-05,
+      "loss": 3.6719,
+      "step": 8892
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3215091837017706e-05,
+      "loss": 5.5938,
+      "step": 8893
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.321349446780163e-05,
+      "loss": 4.1719,
+      "step": 8894
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.321189694899361e-05,
+      "loss": 5.3906,
+      "step": 8895
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.321029928061174e-05,
+      "loss": 4.2031,
+      "step": 8896
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.320870146267409e-05,
+      "loss": 2.9375,
+      "step": 8897
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3207103495198776e-05,
+      "loss": 3.8438,
+      "step": 8898
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3205505378203856e-05,
+      "loss": 4.4844,
+      "step": 8899
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3203907111707445e-05,
+      "loss": 2.0625,
+      "step": 8900
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.320230869572763e-05,
+      "loss": 4.1094,
+      "step": 8901
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.320071013028251e-05,
+      "loss": 3.4688,
+      "step": 8902
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3199111415390184e-05,
+      "loss": 2.5312,
+      "step": 8903
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.319751255106875e-05,
+      "loss": 1.6875,
+      "step": 8904
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.31959135373363e-05,
+      "loss": 4.625,
+      "step": 8905
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.319431437421095e-05,
+      "loss": 4.7969,
+      "step": 8906
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3192715061710795e-05,
+      "loss": 3.3906,
+      "step": 8907
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3191115599853936e-05,
+      "loss": 3.3438,
+      "step": 8908
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.31895159886585e-05,
+      "loss": 3.8125,
+      "step": 8909
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.318791622814258e-05,
+      "loss": 1.875,
+      "step": 8910
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.31863163183243e-05,
+      "loss": 3.0312,
+      "step": 8911
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.318471625922176e-05,
+      "loss": 4.3125,
+      "step": 8912
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.318311605085309e-05,
+      "loss": 3.5469,
+      "step": 8913
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3181515693236384e-05,
+      "loss": 2.125,
+      "step": 8914
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3179915186389774e-05,
+      "loss": 4.7969,
+      "step": 8915
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3178314530331386e-05,
+      "loss": 3.125,
+      "step": 8916
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3176713725079325e-05,
+      "loss": 2.6094,
+      "step": 8917
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.317511277065172e-05,
+      "loss": 2.7422,
+      "step": 8918
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3173511667066696e-05,
+      "loss": 2.9062,
+      "step": 8919
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.317191041434239e-05,
+      "loss": 2.6562,
+      "step": 8920
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.317030901249692e-05,
+      "loss": 5.6406,
+      "step": 8921
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.316870746154841e-05,
+      "loss": 3.9375,
+      "step": 8922
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3167105761515e-05,
+      "loss": 2.5312,
+      "step": 8923
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3165503912414825e-05,
+      "loss": 4.0625,
+      "step": 8924
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.316390191426601e-05,
+      "loss": 3.5938,
+      "step": 8925
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.31622997670867e-05,
+      "loss": 4.3906,
+      "step": 8926
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.316069747089503e-05,
+      "loss": 4.6875,
+      "step": 8927
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.315909502570914e-05,
+      "loss": 3.0312,
+      "step": 8928
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.315749243154718e-05,
+      "loss": 3.7031,
+      "step": 8929
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.315588968842727e-05,
+      "loss": 4.2188,
+      "step": 8930
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.315428679636758e-05,
+      "loss": 4.7812,
+      "step": 8931
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.315268375538625e-05,
+      "loss": 4.8438,
+      "step": 8932
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.3151080565501415e-05,
+      "loss": 4.2812,
+      "step": 8933
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.314947722673124e-05,
+      "loss": 5.4375,
+      "step": 8934
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.314787373909387e-05,
+      "loss": 3.5469,
+      "step": 8935
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.314627010260746e-05,
+      "loss": 2.9375,
+      "step": 8936
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.314466631729016e-05,
+      "loss": 3.9531,
+      "step": 8937
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.314306238316014e-05,
+      "loss": 3.1875,
+      "step": 8938
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.3141458300235544e-05,
+      "loss": 2.9062,
+      "step": 8939
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.313985406853454e-05,
+      "loss": 3.9688,
+      "step": 8940
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.313824968807529e-05,
+      "loss": 4.1875,
+      "step": 8941
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.313664515887595e-05,
+      "loss": 4.8438,
+      "step": 8942
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.313504048095469e-05,
+      "loss": 2.5,
+      "step": 8943
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.313343565432968e-05,
+      "loss": 2.75,
+      "step": 8944
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.3131830679019086e-05,
+      "loss": 2.5625,
+      "step": 8945
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.313022555504108e-05,
+      "loss": 4.0469,
+      "step": 8946
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.3128620282413823e-05,
+      "loss": 0.7812,
+      "step": 8947
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.31270148611555e-05,
+      "loss": 3.5,
+      "step": 8948
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.312540929128429e-05,
+      "loss": 3.2656,
+      "step": 8949
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.3123803572818354e-05,
+      "loss": 4.6094,
+      "step": 8950
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.312219770577588e-05,
+      "loss": 5.3438,
+      "step": 8951
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.312059169017505e-05,
+      "loss": 4.8125,
+      "step": 8952
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.311898552603404e-05,
+      "loss": 3.0,
+      "step": 8953
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.311737921337105e-05,
+      "loss": 3.875,
+      "step": 8954
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.3115772752204243e-05,
+      "loss": 3.0781,
+      "step": 8955
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.311416614255182e-05,
+      "loss": 2.9375,
+      "step": 8956
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.311255938443196e-05,
+      "loss": 2.5625,
+      "step": 8957
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.3110952477862864e-05,
+      "loss": 3.5312,
+      "step": 8958
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.3109345422862715e-05,
+      "loss": 3.6719,
+      "step": 8959
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.3107738219449715e-05,
+      "loss": 4.3281,
+      "step": 8960
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.310613086764205e-05,
+      "loss": 5.6875,
+      "step": 8961
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.310452336745792e-05,
+      "loss": 5.2188,
+      "step": 8962
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.310291571891553e-05,
+      "loss": 3.2812,
+      "step": 8963
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.310130792203308e-05,
+      "loss": 2.875,
+      "step": 8964
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.309969997682876e-05,
+      "loss": 2.9531,
+      "step": 8965
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3098091883320783e-05,
+      "loss": 3.5469,
+      "step": 8966
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3096483641527356e-05,
+      "loss": 3.8906,
+      "step": 8967
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3094875251466686e-05,
+      "loss": 2.375,
+      "step": 8968
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.309326671315698e-05,
+      "loss": 3.9688,
+      "step": 8969
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.309165802661645e-05,
+      "loss": 5.1094,
+      "step": 8970
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.30900491918633e-05,
+      "loss": 3.2344,
+      "step": 8971
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3088440208915746e-05,
+      "loss": 3.9844,
+      "step": 8972
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3086831077792015e-05,
+      "loss": 2.625,
+      "step": 8973
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3085221798510316e-05,
+      "loss": 7.6094,
+      "step": 8974
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3083612371088864e-05,
+      "loss": 4.9219,
+      "step": 8975
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.308200279554589e-05,
+      "loss": 3.875,
+      "step": 8976
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.308039307189961e-05,
+      "loss": 1.7188,
+      "step": 8977
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3078783200168244e-05,
+      "loss": 4.4531,
+      "step": 8978
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.307717318037002e-05,
+      "loss": 5.375,
+      "step": 8979
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3075563012523176e-05,
+      "loss": 4.5312,
+      "step": 8980
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.307395269664593e-05,
+      "loss": 3.75,
+      "step": 8981
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3072342232756516e-05,
+      "loss": 4.0,
+      "step": 8982
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.307073162087316e-05,
+      "loss": 1.5312,
+      "step": 8983
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.30691208610141e-05,
+      "loss": 3.3438,
+      "step": 8984
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3067509953197576e-05,
+      "loss": 3.2812,
+      "step": 8985
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.306589889744182e-05,
+      "loss": 4.0781,
+      "step": 8986
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.306428769376508e-05,
+      "loss": 7.8594,
+      "step": 8987
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.306267634218558e-05,
+      "loss": 4.1562,
+      "step": 8988
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.306106484272158e-05,
+      "loss": 4.7188,
+      "step": 8989
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3059453195391306e-05,
+      "loss": 2.4062,
+      "step": 8990
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3057841400213024e-05,
+      "loss": 4.4375,
+      "step": 8991
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3056229457204976e-05,
+      "loss": 2.6875,
+      "step": 8992
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.305461736638539e-05,
+      "loss": 5.5156,
+      "step": 8993
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.305300512777254e-05,
+      "loss": 2.2188,
+      "step": 8994
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.305139274138468e-05,
+      "loss": 4.9688,
+      "step": 8995
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.304978020724004e-05,
+      "loss": 4.7969,
+      "step": 8996
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.304816752535691e-05,
+      "loss": 3.5156,
+      "step": 8997
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.304655469575351e-05,
+      "loss": 5.25,
+      "step": 8998
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3044941718448126e-05,
+      "loss": 3.0938,
+      "step": 8999
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.304332859345901e-05,
+      "loss": 4.3125,
+      "step": 9000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3041715320804425e-05,
+      "loss": 3.125,
+      "step": 9001
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.304010190050263e-05,
+      "loss": 4.7656,
+      "step": 9002
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.303848833257191e-05,
+      "loss": 3.7656,
+      "step": 9003
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3036874617030506e-05,
+      "loss": 3.5781,
+      "step": 9004
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3035260753896704e-05,
+      "loss": 4.875,
+      "step": 9005
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.303364674318876e-05,
+      "loss": 3.6562,
+      "step": 9006
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.303203258492497e-05,
+      "loss": 4.4531,
+      "step": 9007
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.303041827912359e-05,
+      "loss": 4.4844,
+      "step": 9008
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.302880382580289e-05,
+      "loss": 3.6406,
+      "step": 9009
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.302718922498117e-05,
+      "loss": 1.7812,
+      "step": 9010
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3025574476676695e-05,
+      "loss": 3.25,
+      "step": 9011
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3023959580907754e-05,
+      "loss": 3.2812,
+      "step": 9012
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3022344537692616e-05,
+      "loss": 4.0156,
+      "step": 9013
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.302072934704957e-05,
+      "loss": 6.1406,
+      "step": 9014
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.301911400899691e-05,
+      "loss": 2.4531,
+      "step": 9015
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3017498523552916e-05,
+      "loss": 2.5,
+      "step": 9016
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.301588289073587e-05,
+      "loss": 3.3906,
+      "step": 9017
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.301426711056408e-05,
+      "loss": 5.0312,
+      "step": 9018
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.301265118305583e-05,
+      "loss": 4.9375,
+      "step": 9019
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.301103510822942e-05,
+      "loss": 2.7656,
+      "step": 9020
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3009418886103133e-05,
+      "loss": 4.2031,
+      "step": 9021
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.300780251669528e-05,
+      "loss": 2.2031,
+      "step": 9022
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3006186000024155e-05,
+      "loss": 2.3281,
+      "step": 9023
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.300456933610805e-05,
+      "loss": 4.7031,
+      "step": 9024
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3002952524965275e-05,
+      "loss": 2.7188,
+      "step": 9025
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.300133556661413e-05,
+      "loss": 3.1562,
+      "step": 9026
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2999718461072936e-05,
+      "loss": 4.1406,
+      "step": 9027
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.299810120835999e-05,
+      "loss": 3.5938,
+      "step": 9028
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2996483808493594e-05,
+      "loss": 4.0625,
+      "step": 9029
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2994866261492066e-05,
+      "loss": 2.1719,
+      "step": 9030
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.299324856737372e-05,
+      "loss": 3.25,
+      "step": 9031
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.299163072615687e-05,
+      "loss": 2.5938,
+      "step": 9032
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2990012737859827e-05,
+      "loss": 5.2969,
+      "step": 9033
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.298839460250092e-05,
+      "loss": 5.5781,
+      "step": 9034
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.298677632009844e-05,
+      "loss": 3.5312,
+      "step": 9035
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.298515789067075e-05,
+      "loss": 6.0312,
+      "step": 9036
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.298353931423614e-05,
+      "loss": 3.1875,
+      "step": 9037
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.298192059081294e-05,
+      "loss": 6.6719,
+      "step": 9038
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.298030172041949e-05,
+      "loss": 5.7812,
+      "step": 9039
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.29786827030741e-05,
+      "loss": 4.1406,
+      "step": 9040
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.297706353879511e-05,
+      "loss": 4.375,
+      "step": 9041
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2975444227600846e-05,
+      "loss": 4.7188,
+      "step": 9042
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2973824769509644e-05,
+      "loss": 3.7031,
+      "step": 9043
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.297220516453984e-05,
+      "loss": 3.3438,
+      "step": 9044
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2970585412709754e-05,
+      "loss": 5.5,
+      "step": 9045
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2968965514037746e-05,
+      "loss": 3.1406,
+      "step": 9046
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.296734546854214e-05,
+      "loss": 2.875,
+      "step": 9047
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.296572527624127e-05,
+      "loss": 3.7656,
+      "step": 9048
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.296410493715351e-05,
+      "loss": 3.5625,
+      "step": 9049
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2962484451297173e-05,
+      "loss": 4.125,
+      "step": 9050
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.296086381869061e-05,
+      "loss": 4.2656,
+      "step": 9051
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.295924303935219e-05,
+      "loss": 5.9844,
+      "step": 9052
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2957622113300236e-05,
+      "loss": 4.1875,
+      "step": 9053
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.295600104055311e-05,
+      "loss": 5.9062,
+      "step": 9054
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.295437982112916e-05,
+      "loss": 6.6406,
+      "step": 9055
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2952758455046746e-05,
+      "loss": 2.4688,
+      "step": 9056
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.295113694232422e-05,
+      "loss": 3.6562,
+      "step": 9057
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.294951528297994e-05,
+      "loss": 8.7344,
+      "step": 9058
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2947893477032264e-05,
+      "loss": 1.8438,
+      "step": 9059
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.294627152449956e-05,
+      "loss": 3.2188,
+      "step": 9060
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.294464942540018e-05,
+      "loss": 2.0156,
+      "step": 9061
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2943027179752494e-05,
+      "loss": 2.5312,
+      "step": 9062
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.294140478757487e-05,
+      "loss": 2.25,
+      "step": 9063
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.293978224888566e-05,
+      "loss": 4.2344,
+      "step": 9064
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.293815956370325e-05,
+      "loss": 2.6875,
+      "step": 9065
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2936536732046006e-05,
+      "loss": 1.5938,
+      "step": 9066
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.29349137539323e-05,
+      "loss": 3.9219,
+      "step": 9067
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2933290629380504e-05,
+      "loss": 3.0156,
+      "step": 9068
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2931667358409e-05,
+      "loss": 3.0,
+      "step": 9069
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.293004394103616e-05,
+      "loss": 2.8438,
+      "step": 9070
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2928420377280356e-05,
+      "loss": 3.8125,
+      "step": 9071
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.292679666715998e-05,
+      "loss": 4.4219,
+      "step": 9072
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.292517281069341e-05,
+      "loss": 3.2031,
+      "step": 9073
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.292354880789903e-05,
+      "loss": 2.9844,
+      "step": 9074
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.292192465879523e-05,
+      "loss": 1.7188,
+      "step": 9075
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.292030036340038e-05,
+      "loss": 3.4375,
+      "step": 9076
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2918675921732904e-05,
+      "loss": 2.8906,
+      "step": 9077
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.291705133381115e-05,
+      "loss": 3.4297,
+      "step": 9078
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.291542659965354e-05,
+      "loss": 4.1875,
+      "step": 9079
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.291380171927845e-05,
+      "loss": 5.1562,
+      "step": 9080
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.291217669270429e-05,
+      "loss": 1.4688,
+      "step": 9081
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2910551519949456e-05,
+      "loss": 4.5625,
+      "step": 9082
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.290892620103234e-05,
+      "loss": 2.1562,
+      "step": 9083
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.290730073597133e-05,
+      "loss": 4.5,
+      "step": 9084
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.290567512478486e-05,
+      "loss": 3.2812,
+      "step": 9085
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.290404936749131e-05,
+      "loss": 3.2188,
+      "step": 9086
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.290242346410909e-05,
+      "loss": 3.1562,
+      "step": 9087
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.290079741465661e-05,
+      "loss": 3.5625,
+      "step": 9088
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2899171219152285e-05,
+      "loss": 5.9062,
+      "step": 9089
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2897544877614504e-05,
+      "loss": 7.2188,
+      "step": 9090
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.28959183900617e-05,
+      "loss": 4.375,
+      "step": 9091
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.289429175651228e-05,
+      "loss": 5.5938,
+      "step": 9092
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.289266497698466e-05,
+      "loss": 4.9844,
+      "step": 9093
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.289103805149726e-05,
+      "loss": 2.6875,
+      "step": 9094
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.288941098006849e-05,
+      "loss": 3.9688,
+      "step": 9095
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.288778376271678e-05,
+      "loss": 5.4375,
+      "step": 9096
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.288615639946054e-05,
+      "loss": 5.4062,
+      "step": 9097
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2884528890318204e-05,
+      "loss": 5.25,
+      "step": 9098
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.288290123530819e-05,
+      "loss": 4.0156,
+      "step": 9099
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.288127343444894e-05,
+      "loss": 5.4531,
+      "step": 9100
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.287964548775885e-05,
+      "loss": 4.3594,
+      "step": 9101
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.287801739525639e-05,
+      "loss": 4.375,
+      "step": 9102
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2876389156959975e-05,
+      "loss": 3.3906,
+      "step": 9103
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.287476077288803e-05,
+      "loss": 2.3125,
+      "step": 9104
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2873132243058994e-05,
+      "loss": 3.4844,
+      "step": 9105
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.287150356749131e-05,
+      "loss": 4.4375,
+      "step": 9106
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2869874746203416e-05,
+      "loss": 2.9531,
+      "step": 9107
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.286824577921375e-05,
+      "loss": 5.6406,
+      "step": 9108
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.286661666654075e-05,
+      "loss": 3.8125,
+      "step": 9109
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.286498740820287e-05,
+      "loss": 4.4688,
+      "step": 9110
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.286335800421855e-05,
+      "loss": 2.6875,
+      "step": 9111
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.286172845460622e-05,
+      "loss": 3.9688,
+      "step": 9112
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.286009875938435e-05,
+      "loss": 3.25,
+      "step": 9113
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.285846891857139e-05,
+      "loss": 4.5312,
+      "step": 9114
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.285683893218577e-05,
+      "loss": 1.9375,
+      "step": 9115
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.2855208800245977e-05,
+      "loss": 2.9062,
+      "step": 9116
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.285357852277043e-05,
+      "loss": 3.3125,
+      "step": 9117
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2851948099777616e-05,
+      "loss": 3.8281,
+      "step": 9118
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.285031753128597e-05,
+      "loss": 5.6719,
+      "step": 9119
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.284868681731397e-05,
+      "loss": 4.7344,
+      "step": 9120
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2847055957880066e-05,
+      "loss": 4.4531,
+      "step": 9121
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.284542495300272e-05,
+      "loss": 3.0,
+      "step": 9122
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.284379380270041e-05,
+      "loss": 3.7812,
+      "step": 9123
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.284216250699158e-05,
+      "loss": 4.3281,
+      "step": 9124
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.284053106589473e-05,
+      "loss": 2.25,
+      "step": 9125
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2838899479428304e-05,
+      "loss": 3.9531,
+      "step": 9126
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.283726774761078e-05,
+      "loss": 4.6094,
+      "step": 9127
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2835635870460635e-05,
+      "loss": 4.5469,
+      "step": 9128
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2834003847996335e-05,
+      "loss": 3.9844,
+      "step": 9129
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.283237168023637e-05,
+      "loss": 3.4219,
+      "step": 9130
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.283073936719921e-05,
+      "loss": 2.75,
+      "step": 9131
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.282910690890333e-05,
+      "loss": 4.1719,
+      "step": 9132
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.282747430536722e-05,
+      "loss": 2.4688,
+      "step": 9133
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.282584155660936e-05,
+      "loss": 3.3125,
+      "step": 9134
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.282420866264824e-05,
+      "loss": 5.2188,
+      "step": 9135
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2822575623502326e-05,
+      "loss": 3.2031,
+      "step": 9136
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.282094243919013e-05,
+      "loss": 2.75,
+      "step": 9137
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.281930910973013e-05,
+      "loss": 3.4375,
+      "step": 9138
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.281767563514082e-05,
+      "loss": 4.0625,
+      "step": 9139
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2816042015440686e-05,
+      "loss": 3.3594,
+      "step": 9140
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2814408250648234e-05,
+      "loss": 2.9375,
+      "step": 9141
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2812774340781955e-05,
+      "loss": 1.6562,
+      "step": 9142
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2811140285860336e-05,
+      "loss": 1.7031,
+      "step": 9143
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.28095060859019e-05,
+      "loss": 2.375,
+      "step": 9144
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2807871740925125e-05,
+      "loss": 1.5312,
+      "step": 9145
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.2806237250948526e-05,
+      "loss": 4.375,
+      "step": 9146
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.280460261599061e-05,
+      "loss": 5.0156,
+      "step": 9147
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.280296783606987e-05,
+      "loss": 2.8594,
+      "step": 9148
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.280133291120482e-05,
+      "loss": 2.6875,
+      "step": 9149
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.279969784141397e-05,
+      "loss": 2.3281,
+      "step": 9150
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.279806262671583e-05,
+      "loss": 2.0312,
+      "step": 9151
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.279642726712892e-05,
+      "loss": 2.5625,
+      "step": 9152
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.279479176267175e-05,
+      "loss": 4.0781,
+      "step": 9153
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.279315611336282e-05,
+      "loss": 4.1719,
+      "step": 9154
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.279152031922067e-05,
+      "loss": 2.75,
+      "step": 9155
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.2789884380263806e-05,
+      "loss": 4.2031,
+      "step": 9156
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.278824829651076e-05,
+      "loss": 5.8125,
+      "step": 9157
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.278661206798004e-05,
+      "loss": 3.125,
+      "step": 9158
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.278497569469018e-05,
+      "loss": 2.4844,
+      "step": 9159
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.27833391766597e-05,
+      "loss": 2.8438,
+      "step": 9160
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.278170251390713e-05,
+      "loss": 3.6875,
+      "step": 9161
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.2780065706451e-05,
+      "loss": 3.4531,
+      "step": 9162
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.2778428754309844e-05,
+      "loss": 3.9062,
+      "step": 9163
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.277679165750218e-05,
+      "loss": 4.7188,
+      "step": 9164
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.2775154416046556e-05,
+      "loss": 2.7812,
+      "step": 9165
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.277351702996151e-05,
+      "loss": 3.7188,
+      "step": 9166
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.277187949926556e-05,
+      "loss": 2.75,
+      "step": 9167
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.2770241823977265e-05,
+      "loss": 2.3125,
+      "step": 9168
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.276860400411515e-05,
+      "loss": 3.6719,
+      "step": 9169
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.276696603969777e-05,
+      "loss": 5.1875,
+      "step": 9170
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.276532793074366e-05,
+      "loss": 3.4062,
+      "step": 9171
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.2763689677271366e-05,
+      "loss": 3.7812,
+      "step": 9172
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.276205127929944e-05,
+      "loss": 3.1719,
+      "step": 9173
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.276041273684643e-05,
+      "loss": 3.8438,
+      "step": 9174
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.2758774049930876e-05,
+      "loss": 3.375,
+      "step": 9175
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.2757135218571344e-05,
+      "loss": 2.3438,
+      "step": 9176
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.275549624278637e-05,
+      "loss": 5.0938,
+      "step": 9177
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.275385712259453e-05,
+      "loss": 2.25,
+      "step": 9178
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.2752217858014365e-05,
+      "loss": 2.1875,
+      "step": 9179
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.275057844906444e-05,
+      "loss": 2.1719,
+      "step": 9180
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.274893889576331e-05,
+      "loss": 2.0,
+      "step": 9181
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.274729919812956e-05,
+      "loss": 3.7188,
+      "step": 9182
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.274565935618171e-05,
+      "loss": 3.2344,
+      "step": 9183
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.274401936993837e-05,
+      "loss": 3.0938,
+      "step": 9184
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.274237923941807e-05,
+      "loss": 5.2812,
+      "step": 9185
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.274073896463939e-05,
+      "loss": 3.5,
+      "step": 9186
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.2739098545620916e-05,
+      "loss": 4.7656,
+      "step": 9187
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.27374579823812e-05,
+      "loss": 3.2812,
+      "step": 9188
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.273581727493882e-05,
+      "loss": 2.5312,
+      "step": 9189
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.2734176423312356e-05,
+      "loss": 2.5938,
+      "step": 9190
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.273253542752038e-05,
+      "loss": 3.2031,
+      "step": 9191
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.2730894287581474e-05,
+      "loss": 2.4375,
+      "step": 9192
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.2729253003514214e-05,
+      "loss": 6.5938,
+      "step": 9193
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.272761157533717e-05,
+      "loss": 6.0469,
+      "step": 9194
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.272597000306895e-05,
+      "loss": 2.2188,
+      "step": 9195
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.272432828672812e-05,
+      "loss": 3.1406,
+      "step": 9196
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.272268642633327e-05,
+      "loss": 2.25,
+      "step": 9197
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.2721044421902985e-05,
+      "loss": 2.0781,
+      "step": 9198
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.271940227345586e-05,
+      "loss": 3.125,
+      "step": 9199
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.271775998101049e-05,
+      "loss": 4.8594,
+      "step": 9200
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.271611754458546e-05,
+      "loss": 2.7656,
+      "step": 9201
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.2714474964199365e-05,
+      "loss": 3.4688,
+      "step": 9202
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.2712832239870803e-05,
+      "loss": 3.2188,
+      "step": 9203
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.271118937161836e-05,
+      "loss": 3.8125,
+      "step": 9204
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.270954635946065e-05,
+      "loss": 5.8438,
+      "step": 9205
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.270790320341628e-05,
+      "loss": 2.2188,
+      "step": 9206
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.270625990350383e-05,
+      "loss": 2.8281,
+      "step": 9207
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.270461645974193e-05,
+      "loss": 2.7188,
+      "step": 9208
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.270297287214916e-05,
+      "loss": 1.8281,
+      "step": 9209
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.270132914074414e-05,
+      "loss": 8.3438,
+      "step": 9210
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.269968526554548e-05,
+      "loss": 2.4062,
+      "step": 9211
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.269804124657178e-05,
+      "loss": 2.5312,
+      "step": 9212
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.269639708384167e-05,
+      "loss": 5.1719,
+      "step": 9213
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.269475277737376e-05,
+      "loss": 1.3125,
+      "step": 9214
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.269310832718664e-05,
+      "loss": 5.7188,
+      "step": 9215
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.269146373329896e-05,
+      "loss": 7.0156,
+      "step": 9216
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.2689818995729326e-05,
+      "loss": 2.7969,
+      "step": 9217
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.2688174114496356e-05,
+      "loss": 3.9688,
+      "step": 9218
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.268652908961867e-05,
+      "loss": 5.5,
+      "step": 9219
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.26848839211149e-05,
+      "loss": 4.3906,
+      "step": 9220
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.268323860900366e-05,
+      "loss": 3.6719,
+      "step": 9221
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.2681593153303595e-05,
+      "loss": 4.8125,
+      "step": 9222
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.2679947554033314e-05,
+      "loss": 3.3438,
+      "step": 9223
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.267830181121145e-05,
+      "loss": 4.3438,
+      "step": 9224
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.267665592485665e-05,
+      "loss": 4.0312,
+      "step": 9225
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.267500989498753e-05,
+      "loss": 1.5312,
+      "step": 9226
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.267336372162273e-05,
+      "loss": 2.75,
+      "step": 9227
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.267171740478089e-05,
+      "loss": 4.9688,
+      "step": 9228
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.2670070944480654e-05,
+      "loss": 1.3125,
+      "step": 9229
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.2668424340740646e-05,
+      "loss": 3.6562,
+      "step": 9230
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.2666777593579514e-05,
+      "loss": 1.9062,
+      "step": 9231
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.2665130703015915e-05,
+      "loss": 3.5469,
+      "step": 9232
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.266348366906847e-05,
+      "loss": 1.9688,
+      "step": 9233
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.2661836491755845e-05,
+      "loss": 2.0625,
+      "step": 9234
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.266018917109668e-05,
+      "loss": 4.6094,
+      "step": 9235
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.2658541707109614e-05,
+      "loss": 5.5625,
+      "step": 9236
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.265689409981332e-05,
+      "loss": 5.0312,
+      "step": 9237
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.265524634922643e-05,
+      "loss": 2.7656,
+      "step": 9238
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.265359845536761e-05,
+      "loss": 2.8125,
+      "step": 9239
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.265195041825551e-05,
+      "loss": 2.3125,
+      "step": 9240
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.265030223790881e-05,
+      "loss": 3.7031,
+      "step": 9241
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.2648653914346137e-05,
+      "loss": 2.8438,
+      "step": 9242
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.264700544758617e-05,
+      "loss": 3.2812,
+      "step": 9243
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.264535683764756e-05,
+      "loss": 3.9844,
+      "step": 9244
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.264370808454898e-05,
+      "loss": 4.2969,
+      "step": 9245
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.26420591883091e-05,
+      "loss": 5.5469,
+      "step": 9246
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.264041014894658e-05,
+      "loss": 3.8438,
+      "step": 9247
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.263876096648009e-05,
+      "loss": 4.1406,
+      "step": 9248
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.26371116409283e-05,
+      "loss": 7.6094,
+      "step": 9249
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.2635462172309885e-05,
+      "loss": 3.2812,
+      "step": 9250
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.2633812560643524e-05,
+      "loss": 3.6562,
+      "step": 9251
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.263216280594788e-05,
+      "loss": 3.1094,
+      "step": 9252
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.263051290824163e-05,
+      "loss": 2.0625,
+      "step": 9253
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.262886286754347e-05,
+      "loss": 4.6875,
+      "step": 9254
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.262721268387206e-05,
+      "loss": 4.8906,
+      "step": 9255
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.262556235724609e-05,
+      "loss": 1.9531,
+      "step": 9256
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.2623911887684255e-05,
+      "loss": 3.3125,
+      "step": 9257
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.262226127520523e-05,
+      "loss": 5.5,
+      "step": 9258
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.262061051982769e-05,
+      "loss": 4.0625,
+      "step": 9259
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.2618959621570336e-05,
+      "loss": 3.5938,
+      "step": 9260
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.2617308580451864e-05,
+      "loss": 5.3125,
+      "step": 9261
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.261565739649096e-05,
+      "loss": 7.0,
+      "step": 9262
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.261400606970631e-05,
+      "loss": 2.9062,
+      "step": 9263
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.261235460011662e-05,
+      "loss": 2.8438,
+      "step": 9264
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.2610702987740585e-05,
+      "loss": 3.9219,
+      "step": 9265
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.260905123259689e-05,
+      "loss": 4.2656,
+      "step": 9266
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.2607399334704255e-05,
+      "loss": 2.75,
+      "step": 9267
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.260574729408137e-05,
+      "loss": 4.4219,
+      "step": 9268
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.260409511074693e-05,
+      "loss": 2.1562,
+      "step": 9269
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.260244278471966e-05,
+      "loss": 3.7812,
+      "step": 9270
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.260079031601825e-05,
+      "loss": 3.9219,
+      "step": 9271
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.2599137704661405e-05,
+      "loss": 2.2656,
+      "step": 9272
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.2597484950667846e-05,
+      "loss": 2.6562,
+      "step": 9273
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.259583205405629e-05,
+      "loss": 3.0625,
+      "step": 9274
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.2594179014845434e-05,
+      "loss": 2.3125,
+      "step": 9275
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.2592525833053995e-05,
+      "loss": 2.0312,
+      "step": 9276
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.259087250870069e-05,
+      "loss": 3.9375,
+      "step": 9277
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.258921904180425e-05,
+      "loss": 2.2812,
+      "step": 9278
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.2587565432383374e-05,
+      "loss": 3.7344,
+      "step": 9279
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.258591168045679e-05,
+      "loss": 1.7344,
+      "step": 9280
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.258425778604323e-05,
+      "loss": 1.6875,
+      "step": 9281
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.2582603749161404e-05,
+      "loss": 3.5938,
+      "step": 9282
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.258094956983005e-05,
+      "loss": 3.0625,
+      "step": 9283
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.2579295248067884e-05,
+      "loss": 2.75,
+      "step": 9284
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.2577640783893635e-05,
+      "loss": 3.5312,
+      "step": 9285
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.2575986177326045e-05,
+      "loss": 4.6562,
+      "step": 9286
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.257433142838383e-05,
+      "loss": 2.9375,
+      "step": 9287
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.2572676537085745e-05,
+      "loss": 4.125,
+      "step": 9288
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.2571021503450506e-05,
+      "loss": 3.2812,
+      "step": 9289
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.256936632749686e-05,
+      "loss": 3.375,
+      "step": 9290
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.256771100924354e-05,
+      "loss": 2.625,
+      "step": 9291
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.256605554870929e-05,
+      "loss": 2.7188,
+      "step": 9292
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.256439994591285e-05,
+      "loss": 3.2812,
+      "step": 9293
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.256274420087296e-05,
+      "loss": 7.25,
+      "step": 9294
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.2561088313608375e-05,
+      "loss": 5.0781,
+      "step": 9295
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.255943228413783e-05,
+      "loss": 3.875,
+      "step": 9296
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.255777611248008e-05,
+      "loss": 4.2188,
+      "step": 9297
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.255611979865387e-05,
+      "loss": 3.0938,
+      "step": 9298
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.255446334267796e-05,
+      "loss": 1.5625,
+      "step": 9299
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.255280674457109e-05,
+      "loss": 7.1094,
+      "step": 9300
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.255115000435203e-05,
+      "loss": 4.3438,
+      "step": 9301
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.254949312203953e-05,
+      "loss": 4.0,
+      "step": 9302
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.2547836097652333e-05,
+      "loss": 3.0312,
+      "step": 9303
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.254617893120922e-05,
+      "loss": 2.2188,
+      "step": 9304
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.2544521622728944e-05,
+      "loss": 3.0938,
+      "step": 9305
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.254286417223026e-05,
+      "loss": 4.1875,
+      "step": 9306
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.254120657973195e-05,
+      "loss": 2.5625,
+      "step": 9307
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.253954884525276e-05,
+      "loss": 2.125,
+      "step": 9308
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.253789096881147e-05,
+      "loss": 4.0312,
+      "step": 9309
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.253623295042685e-05,
+      "loss": 4.0156,
+      "step": 9310
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.253457479011766e-05,
+      "loss": 3.9219,
+      "step": 9311
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.253291648790268e-05,
+      "loss": 4.0625,
+      "step": 9312
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.253125804380069e-05,
+      "loss": 3.4688,
+      "step": 9313
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.252959945783044e-05,
+      "loss": 4.7188,
+      "step": 9314
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.252794073001074e-05,
+      "loss": 3.0938,
+      "step": 9315
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.252628186036035e-05,
+      "loss": 3.0,
+      "step": 9316
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.252462284889805e-05,
+      "loss": 3.9688,
+      "step": 9317
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.252296369564263e-05,
+      "loss": 5.9062,
+      "step": 9318
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.252130440061287e-05,
+      "loss": 3.6875,
+      "step": 9319
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.2519644963827555e-05,
+      "loss": 4.0938,
+      "step": 9320
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.251798538530547e-05,
+      "loss": 2.3125,
+      "step": 9321
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.25163256650654e-05,
+      "loss": 3.2656,
+      "step": 9322
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.251466580312615e-05,
+      "loss": 5.2188,
+      "step": 9323
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.2513005799506503e-05,
+      "loss": 4.7031,
+      "step": 9324
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.2511345654225234e-05,
+      "loss": 3.2344,
+      "step": 9325
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.2509685367301164e-05,
+      "loss": 2.7812,
+      "step": 9326
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.2508024938753077e-05,
+      "loss": 4.4375,
+      "step": 9327
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.2506364368599776e-05,
+      "loss": 4.3906,
+      "step": 9328
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.250470365686006e-05,
+      "loss": 1.625,
+      "step": 9329
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.2503042803552725e-05,
+      "loss": 3.25,
+      "step": 9330
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.2501381808696576e-05,
+      "loss": 4.0625,
+      "step": 9331
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.249972067231042e-05,
+      "loss": 2.5312,
+      "step": 9332
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.249805939441306e-05,
+      "loss": 2.8125,
+      "step": 9333
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.2496397975023306e-05,
+      "loss": 3.625,
+      "step": 9334
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.249473641415996e-05,
+      "loss": 5.6875,
+      "step": 9335
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.2493074711841845e-05,
+      "loss": 3.4844,
+      "step": 9336
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.2491412868087765e-05,
+      "loss": 3.9375,
+      "step": 9337
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.248975088291654e-05,
+      "loss": 3.3594,
+      "step": 9338
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.2488088756346976e-05,
+      "loss": 4.4375,
+      "step": 9339
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.24864264883979e-05,
+      "loss": 2.5938,
+      "step": 9340
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.248476407908813e-05,
+      "loss": 3.2188,
+      "step": 9341
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.2483101528436476e-05,
+      "loss": 2.1406,
+      "step": 9342
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.2481438836461774e-05,
+      "loss": 4.9688,
+      "step": 9343
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.247977600318283e-05,
+      "loss": 3.8438,
+      "step": 9344
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.247811302861849e-05,
+      "loss": 2.3906,
+      "step": 9345
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.2476449912787566e-05,
+      "loss": 2.9375,
+      "step": 9346
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.24747866557089e-05,
+      "loss": 3.875,
+      "step": 9347
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.24731232574013e-05,
+      "loss": 3.5,
+      "step": 9348
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.2471459717883615e-05,
+      "loss": 2.6406,
+      "step": 9349
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.246979603717467e-05,
+      "loss": 3.9375,
+      "step": 9350
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.2468132215293315e-05,
+      "loss": 4.0,
+      "step": 9351
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.246646825225836e-05,
+      "loss": 2.4375,
+      "step": 9352
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.2464804148088666e-05,
+      "loss": 3.5625,
+      "step": 9353
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.2463139902803065e-05,
+      "loss": 2.4219,
+      "step": 9354
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.2461475516420395e-05,
+      "loss": 2.7812,
+      "step": 9355
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.24598109889595e-05,
+      "loss": 1.8906,
+      "step": 9356
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.245814632043923e-05,
+      "loss": 2.7188,
+      "step": 9357
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.245648151087843e-05,
+      "loss": 2.3438,
+      "step": 9358
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.245481656029594e-05,
+      "loss": 1.2969,
+      "step": 9359
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.245315146871061e-05,
+      "loss": 1.7812,
+      "step": 9360
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.24514862361413e-05,
+      "loss": 3.8594,
+      "step": 9361
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.244982086260685e-05,
+      "loss": 2.75,
+      "step": 9362
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.2448155348126126e-05,
+      "loss": 3.0625,
+      "step": 9363
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.2446489692717976e-05,
+      "loss": 3.25,
+      "step": 9364
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.244482389640126e-05,
+      "loss": 2.4844,
+      "step": 9365
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.244315795919484e-05,
+      "loss": 3.25,
+      "step": 9366
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.244149188111756e-05,
+      "loss": 2.3125,
+      "step": 9367
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.2439825662188304e-05,
+      "loss": 4.3438,
+      "step": 9368
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.243815930242592e-05,
+      "loss": 3.4375,
+      "step": 9369
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.243649280184928e-05,
+      "loss": 3.125,
+      "step": 9370
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.243482616047725e-05,
+      "loss": 2.9688,
+      "step": 9371
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.2433159378328703e-05,
+      "loss": 3.1094,
+      "step": 9372
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.24314924554225e-05,
+      "loss": 4.2812,
+      "step": 9373
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.242982539177751e-05,
+      "loss": 4.25,
+      "step": 9374
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.242815818741262e-05,
+      "loss": 4.5938,
+      "step": 9375
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.242649084234669e-05,
+      "loss": 6.5,
+      "step": 9376
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.242482335659861e-05,
+      "loss": 3.9062,
+      "step": 9377
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.242315573018725e-05,
+      "loss": 4.5938,
+      "step": 9378
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.242148796313148e-05,
+      "loss": 3.7031,
+      "step": 9379
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.241982005545021e-05,
+      "loss": 2.5781,
+      "step": 9380
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.241815200716229e-05,
+      "loss": 1.5938,
+      "step": 9381
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.2416483818286625e-05,
+      "loss": 3.4375,
+      "step": 9382
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.241481548884209e-05,
+      "loss": 3.7188,
+      "step": 9383
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.241314701884758e-05,
+      "loss": 1.8125,
+      "step": 9384
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.2411478408321976e-05,
+      "loss": 3.9062,
+      "step": 9385
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.240980965728418e-05,
+      "loss": 4.4062,
+      "step": 9386
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.240814076575307e-05,
+      "loss": 2.4062,
+      "step": 9387
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.240647173374755e-05,
+      "loss": 5.3125,
+      "step": 9388
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.240480256128651e-05,
+      "loss": 3.4688,
+      "step": 9389
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.240313324838885e-05,
+      "loss": 2.4531,
+      "step": 9390
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.240146379507347e-05,
+      "loss": 4.3438,
+      "step": 9391
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.239979420135927e-05,
+      "loss": 4.5781,
+      "step": 9392
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2398124467265146e-05,
+      "loss": 3.5,
+      "step": 9393
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.239645459281e-05,
+      "loss": 2.0312,
+      "step": 9394
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2394784578012755e-05,
+      "loss": 2.5938,
+      "step": 9395
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.23931144228923e-05,
+      "loss": 4.7812,
+      "step": 9396
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.239144412746754e-05,
+      "loss": 3.2812,
+      "step": 9397
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.23897736917574e-05,
+      "loss": 2.5,
+      "step": 9398
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2388103115780775e-05,
+      "loss": 3.7188,
+      "step": 9399
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2386432399556595e-05,
+      "loss": 2.1875,
+      "step": 9400
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2384761543103765e-05,
+      "loss": 2.4531,
+      "step": 9401
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.23830905464412e-05,
+      "loss": 6.2188,
+      "step": 9402
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2381419409587816e-05,
+      "loss": 2.4375,
+      "step": 9403
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2379748132562536e-05,
+      "loss": 2.4844,
+      "step": 9404
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2378076715384276e-05,
+      "loss": 2.625,
+      "step": 9405
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.237640515807198e-05,
+      "loss": 4.6875,
+      "step": 9406
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.237473346064454e-05,
+      "loss": 1.7812,
+      "step": 9407
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.23730616231209e-05,
+      "loss": 4.5312,
+      "step": 9408
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2371389645519984e-05,
+      "loss": 3.6875,
+      "step": 9409
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.236971752786072e-05,
+      "loss": 3.5938,
+      "step": 9410
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2368045270162035e-05,
+      "loss": 2.8125,
+      "step": 9411
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2366372872442864e-05,
+      "loss": 1.8438,
+      "step": 9412
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2364700334722144e-05,
+      "loss": 2.7812,
+      "step": 9413
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.23630276570188e-05,
+      "loss": 2.625,
+      "step": 9414
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.236135483935179e-05,
+      "loss": 5.875,
+      "step": 9415
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.235968188174002e-05,
+      "loss": 2.4688,
+      "step": 9416
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.235800878420246e-05,
+      "loss": 2.7969,
+      "step": 9417
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2356335546758036e-05,
+      "loss": 2.6875,
+      "step": 9418
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2354662169425685e-05,
+      "loss": 3.1875,
+      "step": 9419
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.2352988652224376e-05,
+      "loss": 3.125,
+      "step": 9420
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.235131499517303e-05,
+      "loss": 2.4062,
+      "step": 9421
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2349641198290594e-05,
+      "loss": 3.2188,
+      "step": 9422
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.234796726159604e-05,
+      "loss": 3.1562,
+      "step": 9423
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.234629318510831e-05,
+      "loss": 2.4531,
+      "step": 9424
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2344618968846346e-05,
+      "loss": 4.5312,
+      "step": 9425
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.234294461282911e-05,
+      "loss": 3.3906,
+      "step": 9426
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.234127011707555e-05,
+      "loss": 3.2188,
+      "step": 9427
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2339595481604635e-05,
+      "loss": 4.4688,
+      "step": 9428
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.233792070643532e-05,
+      "loss": 2.5156,
+      "step": 9429
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2336245791586564e-05,
+      "loss": 2.0625,
+      "step": 9430
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.233457073707733e-05,
+      "loss": 3.2812,
+      "step": 9431
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2332895542926574e-05,
+      "loss": 3.125,
+      "step": 9432
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.233122020915327e-05,
+      "loss": 3.3906,
+      "step": 9433
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.232954473577639e-05,
+      "loss": 3.1406,
+      "step": 9434
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2327869122814886e-05,
+      "loss": 1.7031,
+      "step": 9435
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.232619337028774e-05,
+      "loss": 3.2656,
+      "step": 9436
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2324517478213916e-05,
+      "loss": 4.3906,
+      "step": 9437
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2322841446612386e-05,
+      "loss": 3.8438,
+      "step": 9438
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.232116527550214e-05,
+      "loss": 7.3906,
+      "step": 9439
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.231948896490214e-05,
+      "loss": 3.4219,
+      "step": 9440
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2317812514831366e-05,
+      "loss": 2.25,
+      "step": 9441
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.23161359253088e-05,
+      "loss": 4.5781,
+      "step": 9442
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.231445919635342e-05,
+      "loss": 3.0,
+      "step": 9443
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2312782327984206e-05,
+      "loss": 4.0,
+      "step": 9444
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.231110532022015e-05,
+      "loss": 2.9375,
+      "step": 9445
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.230942817308024e-05,
+      "loss": 4.375,
+      "step": 9446
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.2307750886583456e-05,
+      "loss": 4.8281,
+      "step": 9447
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.230607346074878e-05,
+      "loss": 3.2656,
+      "step": 9448
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.230439589559521e-05,
+      "loss": 3.9062,
+      "step": 9449
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.230271819114175e-05,
+      "loss": 3.8438,
+      "step": 9450
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.230104034740737e-05,
+      "loss": 3.6094,
+      "step": 9451
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.2299362364411084e-05,
+      "loss": 1.5,
+      "step": 9452
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.229768424217187e-05,
+      "loss": 3.0469,
+      "step": 9453
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.229600598070875e-05,
+      "loss": 2.9688,
+      "step": 9454
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.22943275800407e-05,
+      "loss": 4.0156,
+      "step": 9455
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.229264904018675e-05,
+      "loss": 6.4219,
+      "step": 9456
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.2290970361165874e-05,
+      "loss": 4.0,
+      "step": 9457
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.228929154299709e-05,
+      "loss": 2.7344,
+      "step": 9458
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.228761258569941e-05,
+      "loss": 3.3281,
+      "step": 9459
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.228593348929183e-05,
+      "loss": 3.125,
+      "step": 9460
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.2284254253793364e-05,
+      "loss": 2.6562,
+      "step": 9461
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.2282574879223015e-05,
+      "loss": 3.5312,
+      "step": 9462
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.228089536559981e-05,
+      "loss": 3.4688,
+      "step": 9463
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.2279215712942755e-05,
+      "loss": 1.6875,
+      "step": 9464
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.227753592127087e-05,
+      "loss": 4.8125,
+      "step": 9465
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.2275855990603166e-05,
+      "loss": 2.3438,
+      "step": 9466
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.227417592095867e-05,
+      "loss": 2.0,
+      "step": 9467
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.227249571235639e-05,
+      "loss": 3.9219,
+      "step": 9468
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.227081536481535e-05,
+      "loss": 4.0781,
+      "step": 9469
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.226913487835459e-05,
+      "loss": 3.0938,
+      "step": 9470
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.226745425299312e-05,
+      "loss": 5.4375,
+      "step": 9471
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.226577348874997e-05,
+      "loss": 3.2656,
+      "step": 9472
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.226409258564417e-05,
+      "loss": 3.2812,
+      "step": 9473
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.226241154369474e-05,
+      "loss": 2.7812,
+      "step": 9474
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.226073036292072e-05,
+      "loss": 2.8125,
+      "step": 9475
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.2259049043341156e-05,
+      "loss": 2.6562,
+      "step": 9476
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.225736758497505e-05,
+      "loss": 4.6094,
+      "step": 9477
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.225568598784147e-05,
+      "loss": 5.6562,
+      "step": 9478
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.225400425195944e-05,
+      "loss": 4.7031,
+      "step": 9479
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.2252322377348e-05,
+      "loss": 3.6562,
+      "step": 9480
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.225064036402618e-05,
+      "loss": 4.4375,
+      "step": 9481
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.224895821201304e-05,
+      "loss": 2.75,
+      "step": 9482
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.2247275921327613e-05,
+      "loss": 2.6875,
+      "step": 9483
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.224559349198895e-05,
+      "loss": 4.3438,
+      "step": 9484
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.22439109240161e-05,
+      "loss": 2.1875,
+      "step": 9485
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.2242228217428104e-05,
+      "loss": 2.9375,
+      "step": 9486
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.224054537224401e-05,
+      "loss": 4.4531,
+      "step": 9487
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.223886238848288e-05,
+      "loss": 6.6719,
+      "step": 9488
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.223717926616377e-05,
+      "loss": 2.9531,
+      "step": 9489
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.223549600530572e-05,
+      "loss": 3.25,
+      "step": 9490
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.223381260592779e-05,
+      "loss": 2.3125,
+      "step": 9491
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.2232129068049053e-05,
+      "loss": 1.7656,
+      "step": 9492
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.223044539168855e-05,
+      "loss": 2.5,
+      "step": 9493
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.222876157686536e-05,
+      "loss": 4.0781,
+      "step": 9494
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.222707762359852e-05,
+      "loss": 3.2812,
+      "step": 9495
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.222539353190712e-05,
+      "loss": 2.125,
+      "step": 9496
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.222370930181022e-05,
+      "loss": 2.5312,
+      "step": 9497
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.222202493332687e-05,
+      "loss": 3.5156,
+      "step": 9498
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.222034042647616e-05,
+      "loss": 2.3906,
+      "step": 9499
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.221865578127715e-05,
+      "loss": 3.6406,
+      "step": 9500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.221697099774891e-05,
+      "loss": 4.4375,
+      "step": 9501
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.2215286075910535e-05,
+      "loss": 4.3594,
+      "step": 9502
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.2213601015781066e-05,
+      "loss": 4.4062,
+      "step": 9503
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.221191581737961e-05,
+      "loss": 5.6719,
+      "step": 9504
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.221023048072523e-05,
+      "loss": 9.6719,
+      "step": 9505
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.2208545005837004e-05,
+      "loss": 3.5312,
+      "step": 9506
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.220685939273402e-05,
+      "loss": 4.8438,
+      "step": 9507
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.220517364143536e-05,
+      "loss": 4.1719,
+      "step": 9508
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.2203487751960115e-05,
+      "loss": 3.9062,
+      "step": 9509
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.220180172432735e-05,
+      "loss": 4.75,
+      "step": 9510
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.220011555855618e-05,
+      "loss": 3.9219,
+      "step": 9511
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.219842925466567e-05,
+      "loss": 4.4688,
+      "step": 9512
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.219674281267494e-05,
+      "loss": 3.3906,
+      "step": 9513
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.219505623260305e-05,
+      "loss": 3.9688,
+      "step": 9514
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2193369514469114e-05,
+      "loss": 7.0312,
+      "step": 9515
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.219168265829222e-05,
+      "loss": 3.0781,
+      "step": 9516
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2189995664091464e-05,
+      "loss": 3.1562,
+      "step": 9517
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2188308531885964e-05,
+      "loss": 3.0781,
+      "step": 9518
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2186621261694785e-05,
+      "loss": 3.4375,
+      "step": 9519
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2184933853537064e-05,
+      "loss": 3.4688,
+      "step": 9520
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2183246307431875e-05,
+      "loss": 3.5938,
+      "step": 9521
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.218155862339835e-05,
+      "loss": 2.6875,
+      "step": 9522
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.217987080145557e-05,
+      "loss": 2.8438,
+      "step": 9523
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2178182841622664e-05,
+      "loss": 4.2812,
+      "step": 9524
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.217649474391873e-05,
+      "loss": 7.5625,
+      "step": 9525
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.217480650836288e-05,
+      "loss": 3.0,
+      "step": 9526
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2173118134974234e-05,
+      "loss": 4.3438,
+      "step": 9527
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2171429623771896e-05,
+      "loss": 1.9062,
+      "step": 9528
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.216974097477499e-05,
+      "loss": 5.1562,
+      "step": 9529
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.216805218800263e-05,
+      "loss": 2.4688,
+      "step": 9530
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.216636326347394e-05,
+      "loss": 2.3125,
+      "step": 9531
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.216467420120803e-05,
+      "loss": 2.2188,
+      "step": 9532
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.216298500122403e-05,
+      "loss": 3.6875,
+      "step": 9533
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.216129566354107e-05,
+      "loss": 2.7188,
+      "step": 9534
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2159606188178255e-05,
+      "loss": 2.625,
+      "step": 9535
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.215791657515473e-05,
+      "loss": 4.5938,
+      "step": 9536
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.215622682448961e-05,
+      "loss": 4.2812,
+      "step": 9537
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.215453693620204e-05,
+      "loss": 4.7656,
+      "step": 9538
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.215284691031115e-05,
+      "loss": 3.1562,
+      "step": 9539
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2151156746836056e-05,
+      "loss": 4.0938,
+      "step": 9540
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.214946644579591e-05,
+      "loss": 2.25,
+      "step": 9541
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2147776007209834e-05,
+      "loss": 3.4688,
+      "step": 9542
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.214608543109698e-05,
+      "loss": 3.25,
+      "step": 9543
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.214439471747648e-05,
+      "loss": 2.0,
+      "step": 9544
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.214270386636747e-05,
+      "loss": 3.4062,
+      "step": 9545
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.214101287778911e-05,
+      "loss": 3.4375,
+      "step": 9546
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2139321751760514e-05,
+      "loss": 2.5,
+      "step": 9547
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2137630488300864e-05,
+      "loss": 2.0625,
+      "step": 9548
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2135939087429276e-05,
+      "loss": 3.8438,
+      "step": 9549
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.213424754916491e-05,
+      "loss": 4.0938,
+      "step": 9550
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.213255587352692e-05,
+      "loss": 4.2188,
+      "step": 9551
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.213086406053446e-05,
+      "loss": 2.75,
+      "step": 9552
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.212917211020667e-05,
+      "loss": 4.5781,
+      "step": 9553
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.212748002256272e-05,
+      "loss": 3.2812,
+      "step": 9554
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2125787797621756e-05,
+      "loss": 4.75,
+      "step": 9555
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2124095435402934e-05,
+      "loss": 4.0625,
+      "step": 9556
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.212240293592542e-05,
+      "loss": 2.0625,
+      "step": 9557
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.212071029920838e-05,
+      "loss": 3.5938,
+      "step": 9558
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2119017525270964e-05,
+      "loss": 3.3281,
+      "step": 9559
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2117324614132345e-05,
+      "loss": 2.8438,
+      "step": 9560
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2115631565811686e-05,
+      "loss": 2.1406,
+      "step": 9561
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2113938380328156e-05,
+      "loss": 2.625,
+      "step": 9562
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2112245057700925e-05,
+      "loss": 4.0625,
+      "step": 9563
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2110551597949156e-05,
+      "loss": 1.4688,
+      "step": 9564
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.210885800109203e-05,
+      "loss": 4.1094,
+      "step": 9565
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.210716426714871e-05,
+      "loss": 2.3438,
+      "step": 9566
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2105470396138374e-05,
+      "loss": 3.3125,
+      "step": 9567
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.21037763880802e-05,
+      "loss": 3.4219,
+      "step": 9568
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.210208224299338e-05,
+      "loss": 2.75,
+      "step": 9569
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.210038796089707e-05,
+      "loss": 3.3438,
+      "step": 9570
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.209869354181047e-05,
+      "loss": 5.3125,
+      "step": 9571
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2096998985752755e-05,
+      "loss": 2.2969,
+      "step": 9572
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2095304292743106e-05,
+      "loss": 3.2031,
+      "step": 9573
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.2093609462800715e-05,
+      "loss": 2.5,
+      "step": 9574
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.2091914495944765e-05,
+      "loss": 2.9531,
+      "step": 9575
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.209021939219445e-05,
+      "loss": 4.7812,
+      "step": 9576
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.2088524151568945e-05,
+      "loss": 4.125,
+      "step": 9577
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.208682877408746e-05,
+      "loss": 3.7812,
+      "step": 9578
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.2085133259769195e-05,
+      "loss": 4.0,
+      "step": 9579
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.208343760863333e-05,
+      "loss": 3.7031,
+      "step": 9580
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.208174182069906e-05,
+      "loss": 2.2188,
+      "step": 9581
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.2080045895985584e-05,
+      "loss": 2.0469,
+      "step": 9582
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.207834983451211e-05,
+      "loss": 2.875,
+      "step": 9583
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.2076653636297834e-05,
+      "loss": 3.1562,
+      "step": 9584
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.207495730136196e-05,
+      "loss": 3.3516,
+      "step": 9585
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.20732608297237e-05,
+      "loss": 6.25,
+      "step": 9586
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.207156422140225e-05,
+      "loss": 2.9375,
+      "step": 9587
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.206986747641681e-05,
+      "loss": 11.0,
+      "step": 9588
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.206817059478661e-05,
+      "loss": 3.7188,
+      "step": 9589
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.206647357653084e-05,
+      "loss": 3.6562,
+      "step": 9590
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.2064776421668735e-05,
+      "loss": 2.9688,
+      "step": 9591
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.206307913021949e-05,
+      "loss": 2.8906,
+      "step": 9592
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.206138170220233e-05,
+      "loss": 5.4375,
+      "step": 9593
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.205968413763646e-05,
+      "loss": 3.9062,
+      "step": 9594
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.2057986436541115e-05,
+      "loss": 2.2812,
+      "step": 9595
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.20562885989355e-05,
+      "loss": 2.25,
+      "step": 9596
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.205459062483885e-05,
+      "loss": 3.2188,
+      "step": 9597
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.205289251427037e-05,
+      "loss": 4.8438,
+      "step": 9598
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.20511942672493e-05,
+      "loss": 3.2812,
+      "step": 9599
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.204949588379487e-05,
+      "loss": 3.3438,
+      "step": 9600
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.204779736392629e-05,
+      "loss": 2.375,
+      "step": 9601
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.20460987076628e-05,
+      "loss": 1.6562,
+      "step": 9602
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.204439991502363e-05,
+      "loss": 3.5312,
+      "step": 9603
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.204270098602802e-05,
+      "loss": 7.0,
+      "step": 9604
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.2041001920695184e-05,
+      "loss": 3.8125,
+      "step": 9605
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.2039302719044374e-05,
+      "loss": 4.2188,
+      "step": 9606
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.2037603381094815e-05,
+      "loss": 3.375,
+      "step": 9607
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.203590390686576e-05,
+      "loss": 2.7812,
+      "step": 9608
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.203420429637643e-05,
+      "loss": 3.3125,
+      "step": 9609
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.203250454964609e-05,
+      "loss": 4.6875,
+      "step": 9610
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.2030804666693964e-05,
+      "loss": 2.6562,
+      "step": 9611
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.2029104647539306e-05,
+      "loss": 4.1562,
+      "step": 9612
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.202740449220136e-05,
+      "loss": 3.4375,
+      "step": 9613
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.202570420069937e-05,
+      "loss": 4.5156,
+      "step": 9614
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.2024003773052594e-05,
+      "loss": 2.2812,
+      "step": 9615
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.2022303209280274e-05,
+      "loss": 2.7188,
+      "step": 9616
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.202060250940166e-05,
+      "loss": 3.7188,
+      "step": 9617
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.201890167343602e-05,
+      "loss": 2.7812,
+      "step": 9618
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.2017200701402595e-05,
+      "loss": 4.4375,
+      "step": 9619
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.2015499593320656e-05,
+      "loss": 3.0625,
+      "step": 9620
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.201379834920944e-05,
+      "loss": 4.2656,
+      "step": 9621
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.2012096969088234e-05,
+      "loss": 1.9844,
+      "step": 9622
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.2010395452976273e-05,
+      "loss": 2.0938,
+      "step": 9623
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.2008693800892845e-05,
+      "loss": 3.2812,
+      "step": 9624
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.20069920128572e-05,
+      "loss": 3.9375,
+      "step": 9625
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.20052900888886e-05,
+      "loss": 2.2188,
+      "step": 9626
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.200358802900633e-05,
+      "loss": 4.3594,
+      "step": 9627
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.200188583322964e-05,
+      "loss": 5.0781,
+      "step": 9628
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.200018350157781e-05,
+      "loss": 2.7188,
+      "step": 9629
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.199848103407012e-05,
+      "loss": 3.375,
+      "step": 9630
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.199677843072583e-05,
+      "loss": 2.4375,
+      "step": 9631
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.1995075691564224e-05,
+      "loss": 2.5938,
+      "step": 9632
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.199337281660457e-05,
+      "loss": 2.0625,
+      "step": 9633
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.199166980586616e-05,
+      "loss": 3.5469,
+      "step": 9634
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.198996665936827e-05,
+      "loss": 3.75,
+      "step": 9635
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.198826337713016e-05,
+      "loss": 1.875,
+      "step": 9636
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.1986559959171155e-05,
+      "loss": 4.2031,
+      "step": 9637
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.1984856405510504e-05,
+      "loss": 7.5781,
+      "step": 9638
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.1983152716167507e-05,
+      "loss": 1.2656,
+      "step": 9639
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.198144889116145e-05,
+      "loss": 3.0625,
+      "step": 9640
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.1979744930511624e-05,
+      "loss": 4.375,
+      "step": 9641
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.1978040834237316e-05,
+      "loss": 2.9688,
+      "step": 9642
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.1976336602357824e-05,
+      "loss": 5.5938,
+      "step": 9643
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.197463223489243e-05,
+      "loss": 2.375,
+      "step": 9644
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.1972927731860455e-05,
+      "loss": 4.0312,
+      "step": 9645
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.197122309328116e-05,
+      "loss": 4.8438,
+      "step": 9646
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.196951831917387e-05,
+      "loss": 3.3438,
+      "step": 9647
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.196781340955788e-05,
+      "loss": 2.8438,
+      "step": 9648
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.196610836445248e-05,
+      "loss": 3.5625,
+      "step": 9649
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.196440318387699e-05,
+      "loss": 2.9219,
+      "step": 9650
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.19626978678507e-05,
+      "loss": 4.4062,
+      "step": 9651
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.196099241639293e-05,
+      "loss": 2.8281,
+      "step": 9652
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.195928682952298e-05,
+      "loss": 2.9844,
+      "step": 9653
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.195758110726016e-05,
+      "loss": 6.1094,
+      "step": 9654
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.195587524962377e-05,
+      "loss": 5.4844,
+      "step": 9655
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.195416925663314e-05,
+      "loss": 4.9219,
+      "step": 9656
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.195246312830757e-05,
+      "loss": 4.1719,
+      "step": 9657
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.195075686466639e-05,
+      "loss": 0.9062,
+      "step": 9658
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.194905046572891e-05,
+      "loss": 3.9375,
+      "step": 9659
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.194734393151444e-05,
+      "loss": 3.4219,
+      "step": 9660
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.194563726204231e-05,
+      "loss": 4.7812,
+      "step": 9661
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.194393045733184e-05,
+      "loss": 2.2188,
+      "step": 9662
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.194222351740235e-05,
+      "loss": 4.5312,
+      "step": 9663
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.194051644227315e-05,
+      "loss": 3.5938,
+      "step": 9664
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1938809231963596e-05,
+      "loss": 3.5312,
+      "step": 9665
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1937101886493006e-05,
+      "loss": 2.5938,
+      "step": 9666
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1935394405880695e-05,
+      "loss": 2.7188,
+      "step": 9667
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.193368679014601e-05,
+      "loss": 4.8438,
+      "step": 9668
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.193197903930827e-05,
+      "loss": 4.4219,
+      "step": 9669
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.193027115338682e-05,
+      "loss": 4.4375,
+      "step": 9670
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.192856313240099e-05,
+      "loss": 4.2656,
+      "step": 9671
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1926854976370116e-05,
+      "loss": 3.7656,
+      "step": 9672
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.192514668531353e-05,
+      "loss": 3.6094,
+      "step": 9673
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.192343825925059e-05,
+      "loss": 3.3594,
+      "step": 9674
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.192172969820062e-05,
+      "loss": 3.4062,
+      "step": 9675
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.192002100218298e-05,
+      "loss": 2.0625,
+      "step": 9676
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.191831217121699e-05,
+      "loss": 3.0469,
+      "step": 9677
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.191660320532201e-05,
+      "loss": 4.0625,
+      "step": 9678
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.19148941045174e-05,
+      "loss": 5.8125,
+      "step": 9679
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.191318486882249e-05,
+      "loss": 4.1875,
+      "step": 9680
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1911475498256634e-05,
+      "loss": 2.625,
+      "step": 9681
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.190976599283919e-05,
+      "loss": 2.8438,
+      "step": 9682
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.19080563525895e-05,
+      "loss": 3.4062,
+      "step": 9683
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1906346577526934e-05,
+      "loss": 2.2188,
+      "step": 9684
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.190463666767084e-05,
+      "loss": 1.8438,
+      "step": 9685
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.190292662304059e-05,
+      "loss": 3.9062,
+      "step": 9686
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.190121644365551e-05,
+      "loss": 2.3438,
+      "step": 9687
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.189950612953499e-05,
+      "loss": 4.375,
+      "step": 9688
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1897795680698395e-05,
+      "loss": 3.2812,
+      "step": 9689
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1896085097165074e-05,
+      "loss": 4.5469,
+      "step": 9690
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1894374378954395e-05,
+      "loss": 3.1719,
+      "step": 9691
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.189266352608574e-05,
+      "loss": 3.6094,
+      "step": 9692
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.189095253857845e-05,
+      "loss": 2.5312,
+      "step": 9693
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.1889241416451926e-05,
+      "loss": 2.6719,
+      "step": 9694
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.188753015972553e-05,
+      "loss": 1.7969,
+      "step": 9695
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.1885818768418615e-05,
+      "loss": 1.5,
+      "step": 9696
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.188410724255058e-05,
+      "loss": 3.8281,
+      "step": 9697
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.188239558214079e-05,
+      "loss": 2.875,
+      "step": 9698
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.188068378720863e-05,
+      "loss": 2.3125,
+      "step": 9699
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.187897185777348e-05,
+      "loss": 1.9375,
+      "step": 9700
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.1877259793854716e-05,
+      "loss": 4.5312,
+      "step": 9701
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.187554759547172e-05,
+      "loss": 5.3594,
+      "step": 9702
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.187383526264387e-05,
+      "loss": 2.7969,
+      "step": 9703
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.187212279539058e-05,
+      "loss": 6.2812,
+      "step": 9704
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.18704101937312e-05,
+      "loss": 3.625,
+      "step": 9705
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.186869745768513e-05,
+      "loss": 2.7969,
+      "step": 9706
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.186698458727178e-05,
+      "loss": 3.0,
+      "step": 9707
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.186527158251052e-05,
+      "loss": 3.9219,
+      "step": 9708
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.186355844342075e-05,
+      "loss": 3.1875,
+      "step": 9709
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.1861845170021864e-05,
+      "loss": 4.625,
+      "step": 9710
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.1860131762333266e-05,
+      "loss": 3.5156,
+      "step": 9711
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.185841822037434e-05,
+      "loss": 4.9375,
+      "step": 9712
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.185670454416449e-05,
+      "loss": 4.2031,
+      "step": 9713
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.185499073372312e-05,
+      "loss": 3.2188,
+      "step": 9714
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.185327678906964e-05,
+      "loss": 1.5,
+      "step": 9715
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.185156271022343e-05,
+      "loss": 4.4531,
+      "step": 9716
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.184984849720392e-05,
+      "loss": 2.6562,
+      "step": 9717
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.1848134150030505e-05,
+      "loss": 2.5,
+      "step": 9718
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.18464196687226e-05,
+      "loss": 1.9062,
+      "step": 9719
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.1844705053299606e-05,
+      "loss": 2.1875,
+      "step": 9720
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.184299030378093e-05,
+      "loss": 4.0625,
+      "step": 9721
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.184127542018601e-05,
+      "loss": 6.375,
+      "step": 9722
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.183956040253423e-05,
+      "loss": 1.7188,
+      "step": 9723
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.183784525084503e-05,
+      "loss": 3.8125,
+      "step": 9724
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.183612996513782e-05,
+      "loss": 2.2812,
+      "step": 9725
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.1834414545432006e-05,
+      "loss": 3.125,
+      "step": 9726
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.183269899174703e-05,
+      "loss": 4.9062,
+      "step": 9727
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.183098330410229e-05,
+      "loss": 1.7812,
+      "step": 9728
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.182926748251723e-05,
+      "loss": 3.8438,
+      "step": 9729
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.1827551527011275e-05,
+      "loss": 4.0938,
+      "step": 9730
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.1825835437603836e-05,
+      "loss": 4.8125,
+      "step": 9731
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.182411921431435e-05,
+      "loss": 1.4062,
+      "step": 9732
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.182240285716224e-05,
+      "loss": 4.4375,
+      "step": 9733
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.1820686366166956e-05,
+      "loss": 2.2188,
+      "step": 9734
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.181896974134792e-05,
+      "loss": 2.4531,
+      "step": 9735
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.181725298272455e-05,
+      "loss": 3.4844,
+      "step": 9736
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.181553609031631e-05,
+      "loss": 3.5625,
+      "step": 9737
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.1813819064142616e-05,
+      "loss": 2.25,
+      "step": 9738
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.181210190422291e-05,
+      "loss": 1.1562,
+      "step": 9739
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.181038461057663e-05,
+      "loss": 5.7656,
+      "step": 9740
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.180866718322324e-05,
+      "loss": 2.5156,
+      "step": 9741
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.180694962218215e-05,
+      "loss": 3.5156,
+      "step": 9742
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.1805231927472844e-05,
+      "loss": 3.3438,
+      "step": 9743
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.1803514099114726e-05,
+      "loss": 2.2344,
+      "step": 9744
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.1801796137127275e-05,
+      "loss": 2.2969,
+      "step": 9745
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.1800078041529924e-05,
+      "loss": 3.1719,
+      "step": 9746
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.179835981234213e-05,
+      "loss": 3.0938,
+      "step": 9747
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.179664144958334e-05,
+      "loss": 4.0312,
+      "step": 9748
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.179492295327302e-05,
+      "loss": 1.9062,
+      "step": 9749
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.179320432343061e-05,
+      "loss": 2.6562,
+      "step": 9750
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.179148556007558e-05,
+      "loss": 4.3438,
+      "step": 9751
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.178976666322738e-05,
+      "loss": 3.2812,
+      "step": 9752
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.1788047632905474e-05,
+      "loss": 3.4219,
+      "step": 9753
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.178632846912932e-05,
+      "loss": 3.2812,
+      "step": 9754
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.1784609171918386e-05,
+      "loss": 3.5938,
+      "step": 9755
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.1782889741292126e-05,
+      "loss": 3.125,
+      "step": 9756
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.178117017727002e-05,
+      "loss": 2.5312,
+      "step": 9757
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.177945047987153e-05,
+      "loss": 4.3125,
+      "step": 9758
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.177773064911612e-05,
+      "loss": 1.8438,
+      "step": 9759
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.177601068502326e-05,
+      "loss": 3.3281,
+      "step": 9760
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.1774290587612435e-05,
+      "loss": 3.9375,
+      "step": 9761
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.1772570356903105e-05,
+      "loss": 2.9375,
+      "step": 9762
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.1770849992914747e-05,
+      "loss": 2.0156,
+      "step": 9763
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.176912949566684e-05,
+      "loss": 2.5625,
+      "step": 9764
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.1767408865178865e-05,
+      "loss": 3.7344,
+      "step": 9765
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.17656881014703e-05,
+      "loss": 2.875,
+      "step": 9766
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.176396720456062e-05,
+      "loss": 1.625,
+      "step": 9767
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.176224617446932e-05,
+      "loss": 6.3594,
+      "step": 9768
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.1760525011215865e-05,
+      "loss": 1.375,
+      "step": 9769
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.175880371481976e-05,
+      "loss": 3.75,
+      "step": 9770
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.1757082285300475e-05,
+      "loss": 2.8281,
+      "step": 9771
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.1755360722677514e-05,
+      "loss": 2.0,
+      "step": 9772
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.175363902697036e-05,
+      "loss": 1.9531,
+      "step": 9773
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.17519171981985e-05,
+      "loss": 5.1562,
+      "step": 9774
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.1750195236381436e-05,
+      "loss": 2.8438,
+      "step": 9775
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.1748473141538666e-05,
+      "loss": 3.9688,
+      "step": 9776
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.1746750913689665e-05,
+      "loss": 3.4062,
+      "step": 9777
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.174502855285395e-05,
+      "loss": 2.8438,
+      "step": 9778
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.174330605905102e-05,
+      "loss": 3.9219,
+      "step": 9779
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.174158343230036e-05,
+      "loss": 5.1719,
+      "step": 9780
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.1739860672621485e-05,
+      "loss": 4.0625,
+      "step": 9781
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.17381377800339e-05,
+      "loss": 2.25,
+      "step": 9782
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.17364147545571e-05,
+      "loss": 2.8125,
+      "step": 9783
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.17346915962106e-05,
+      "loss": 3.6406,
+      "step": 9784
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.173296830501391e-05,
+      "loss": 2.3906,
+      "step": 9785
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.1731244880986526e-05,
+      "loss": 2.5625,
+      "step": 9786
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.1729521324147975e-05,
+      "loss": 2.5,
+      "step": 9787
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.172779763451776e-05,
+      "loss": 3.3125,
+      "step": 9788
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.1726073812115404e-05,
+      "loss": 3.1406,
+      "step": 9789
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.172434985696041e-05,
+      "loss": 4.2812,
+      "step": 9790
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.172262576907231e-05,
+      "loss": 2.6875,
+      "step": 9791
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.1720901548470604e-05,
+      "loss": 3.0156,
+      "step": 9792
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.1719177195174826e-05,
+      "loss": 4.1875,
+      "step": 9793
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.1717452709204496e-05,
+      "loss": 3.1094,
+      "step": 9794
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.171572809057914e-05,
+      "loss": 3.6875,
+      "step": 9795
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.171400333931828e-05,
+      "loss": 2.7812,
+      "step": 9796
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.171227845544143e-05,
+      "loss": 3.7188,
+      "step": 9797
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.171055343896813e-05,
+      "loss": 2.125,
+      "step": 9798
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.170882828991791e-05,
+      "loss": 4.875,
+      "step": 9799
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.1707103008310296e-05,
+      "loss": 2.8438,
+      "step": 9800
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.170537759416482e-05,
+      "loss": 3.4531,
+      "step": 9801
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.1703652047501026e-05,
+      "loss": 2.125,
+      "step": 9802
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.170192636833844e-05,
+      "loss": 3.8281,
+      "step": 9803
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.1700200556696594e-05,
+      "loss": 3.6562,
+      "step": 9804
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.169847461259503e-05,
+      "loss": 3.1562,
+      "step": 9805
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.1696748536053294e-05,
+      "loss": 3.0938,
+      "step": 9806
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.169502232709092e-05,
+      "loss": 2.2812,
+      "step": 9807
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.169329598572746e-05,
+      "loss": 6.2188,
+      "step": 9808
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.1691569511982446e-05,
+      "loss": 4.875,
+      "step": 9809
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.168984290587543e-05,
+      "loss": 3.25,
+      "step": 9810
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.168811616742596e-05,
+      "loss": 4.3125,
+      "step": 9811
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.1686389296653576e-05,
+      "loss": 3.5938,
+      "step": 9812
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.168466229357784e-05,
+      "loss": 3.6875,
+      "step": 9813
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.16829351582183e-05,
+      "loss": 1.9219,
+      "step": 9814
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.16812078905945e-05,
+      "loss": 1.6875,
+      "step": 9815
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.1679480490726016e-05,
+      "loss": 3.8438,
+      "step": 9816
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.1677752958632376e-05,
+      "loss": 3.9062,
+      "step": 9817
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.167602529433316e-05,
+      "loss": 3.375,
+      "step": 9818
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.1674297497847926e-05,
+      "loss": 4.0625,
+      "step": 9819
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.167256956919622e-05,
+      "loss": 2.5781,
+      "step": 9820
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.167084150839761e-05,
+      "loss": 3.8438,
+      "step": 9821
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.166911331547167e-05,
+      "loss": 3.1875,
+      "step": 9822
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.1667384990437954e-05,
+      "loss": 1.6406,
+      "step": 9823
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.166565653331603e-05,
+      "loss": 2.1719,
+      "step": 9824
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.166392794412547e-05,
+      "loss": 3.7734,
+      "step": 9825
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.1662199222885845e-05,
+      "loss": 1.4688,
+      "step": 9826
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.166047036961671e-05,
+      "loss": 3.3438,
+      "step": 9827
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.1658741384337664e-05,
+      "loss": 3.4219,
+      "step": 9828
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.165701226706825e-05,
+      "loss": 4.8594,
+      "step": 9829
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.165528301782808e-05,
+      "loss": 3.0547,
+      "step": 9830
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.16535536366367e-05,
+      "loss": 2.5938,
+      "step": 9831
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.165182412351371e-05,
+      "loss": 3.25,
+      "step": 9832
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.165009447847867e-05,
+      "loss": 1.7031,
+      "step": 9833
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.164836470155118e-05,
+      "loss": 3.2188,
+      "step": 9834
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.164663479275081e-05,
+      "loss": 2.3438,
+      "step": 9835
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.164490475209715e-05,
+      "loss": 3.875,
+      "step": 9836
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.164317457960979e-05,
+      "loss": 2.2188,
+      "step": 9837
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.164144427530831e-05,
+      "loss": 1.75,
+      "step": 9838
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.163971383921231e-05,
+      "loss": 4.0938,
+      "step": 9839
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.163798327134136e-05,
+      "loss": 3.0625,
+      "step": 9840
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.163625257171507e-05,
+      "loss": 2.9062,
+      "step": 9841
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.163452174035304e-05,
+      "loss": 3.1094,
+      "step": 9842
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.163279077727484e-05,
+      "loss": 4.5781,
+      "step": 9843
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.163105968250008e-05,
+      "loss": 3.375,
+      "step": 9844
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.162932845604837e-05,
+      "loss": 3.9688,
+      "step": 9845
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.162759709793928e-05,
+      "loss": 2.75,
+      "step": 9846
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.162586560819244e-05,
+      "loss": 2.5938,
+      "step": 9847
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.162413398682744e-05,
+      "loss": 2.5781,
+      "step": 9848
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.1622402233863875e-05,
+      "loss": 4.0,
+      "step": 9849
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.162067034932137e-05,
+      "loss": 2.9375,
+      "step": 9850
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.161893833321952e-05,
+      "loss": 3.0625,
+      "step": 9851
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.1617206185577935e-05,
+      "loss": 4.9688,
+      "step": 9852
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.161547390641623e-05,
+      "loss": 9.375,
+      "step": 9853
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.161374149575401e-05,
+      "loss": 2.375,
+      "step": 9854
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.1612008953610885e-05,
+      "loss": 3.3125,
+      "step": 9855
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.161027628000647e-05,
+      "loss": 2.0938,
+      "step": 9856
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.1608543474960395e-05,
+      "loss": 2.9688,
+      "step": 9857
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.160681053849226e-05,
+      "loss": 3.9062,
+      "step": 9858
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.1605077470621694e-05,
+      "loss": 5.9688,
+      "step": 9859
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.160334427136832e-05,
+      "loss": 2.3594,
+      "step": 9860
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.160161094075175e-05,
+      "loss": 3.7812,
+      "step": 9861
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.159987747879161e-05,
+      "loss": 2.7031,
+      "step": 9862
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.1598143885507524e-05,
+      "loss": 3.6562,
+      "step": 9863
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.1596410160919125e-05,
+      "loss": 3.0,
+      "step": 9864
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.159467630504604e-05,
+      "loss": 4.0156,
+      "step": 9865
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.1592942317907886e-05,
+      "loss": 3.1562,
+      "step": 9866
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.159120819952431e-05,
+      "loss": 3.8438,
+      "step": 9867
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.158947394991493e-05,
+      "loss": 2.4531,
+      "step": 9868
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.158773956909939e-05,
+      "loss": 2.7656,
+      "step": 9869
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.158600505709732e-05,
+      "loss": 2.875,
+      "step": 9870
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.158427041392836e-05,
+      "loss": 3.5312,
+      "step": 9871
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.158253563961214e-05,
+      "loss": 3.3281,
+      "step": 9872
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.158080073416831e-05,
+      "loss": 4.4844,
+      "step": 9873
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.15790656976165e-05,
+      "loss": 4.25,
+      "step": 9874
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.1577330529976374e-05,
+      "loss": 3.9219,
+      "step": 9875
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.1575595231267556e-05,
+      "loss": 3.4062,
+      "step": 9876
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.157385980150969e-05,
+      "loss": 3.1719,
+      "step": 9877
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.1572124240722435e-05,
+      "loss": 4.9688,
+      "step": 9878
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.157038854892544e-05,
+      "loss": 1.9688,
+      "step": 9879
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.1568652726138344e-05,
+      "loss": 4.2344,
+      "step": 9880
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.1566916772380804e-05,
+      "loss": 2.6719,
+      "step": 9881
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.156518068767248e-05,
+      "loss": 2.4062,
+      "step": 9882
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.1563444472033014e-05,
+      "loss": 2.0938,
+      "step": 9883
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.156170812548207e-05,
+      "loss": 2.0156,
+      "step": 9884
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.15599716480393e-05,
+      "loss": 3.5312,
+      "step": 9885
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.155823503972437e-05,
+      "loss": 5.8906,
+      "step": 9886
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.155649830055693e-05,
+      "loss": 3.0,
+      "step": 9887
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.155476143055666e-05,
+      "loss": 4.0,
+      "step": 9888
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.15530244297432e-05,
+      "loss": 3.1094,
+      "step": 9889
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.155128729813623e-05,
+      "loss": 4.5156,
+      "step": 9890
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.154955003575542e-05,
+      "loss": 3.0625,
+      "step": 9891
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.154781264262043e-05,
+      "loss": 4.2656,
+      "step": 9892
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.154607511875092e-05,
+      "loss": 3.5312,
+      "step": 9893
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.154433746416658e-05,
+      "loss": 2.8906,
+      "step": 9894
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.154259967888707e-05,
+      "loss": 2.1562,
+      "step": 9895
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.1540861762932064e-05,
+      "loss": 3.6875,
+      "step": 9896
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.153912371632124e-05,
+      "loss": 2.6719,
+      "step": 9897
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.1537385539074285e-05,
+      "loss": 3.4062,
+      "step": 9898
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.1535647231210855e-05,
+      "loss": 1.5938,
+      "step": 9899
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.1533908792750646e-05,
+      "loss": 2.4062,
+      "step": 9900
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.1532170223713335e-05,
+      "loss": 3.2188,
+      "step": 9901
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.153043152411861e-05,
+      "loss": 2.375,
+      "step": 9902
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.1528692693986135e-05,
+      "loss": 5.9219,
+      "step": 9903
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.1526953733335616e-05,
+      "loss": 4.2188,
+      "step": 9904
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.1525214642186736e-05,
+      "loss": 3.5,
+      "step": 9905
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.1523475420559186e-05,
+      "loss": 3.8438,
+      "step": 9906
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.1521736068472644e-05,
+      "loss": 2.9375,
+      "step": 9907
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.151999658594681e-05,
+      "loss": 2.7656,
+      "step": 9908
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.1518256973001376e-05,
+      "loss": 4.9688,
+      "step": 9909
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.1516517229656045e-05,
+      "loss": 2.2031,
+      "step": 9910
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.151477735593049e-05,
+      "loss": 3.9844,
+      "step": 9911
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.151303735184443e-05,
+      "loss": 2.5156,
+      "step": 9912
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.151129721741756e-05,
+      "loss": 2.6875,
+      "step": 9913
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.150955695266957e-05,
+      "loss": 6.5625,
+      "step": 9914
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.150781655762017e-05,
+      "loss": 6.4531,
+      "step": 9915
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.150607603228906e-05,
+      "loss": 3.6875,
+      "step": 9916
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.1504335376695946e-05,
+      "loss": 3.1484,
+      "step": 9917
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.150259459086054e-05,
+      "loss": 4.1562,
+      "step": 9918
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.150085367480254e-05,
+      "loss": 4.875,
+      "step": 9919
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.149911262854166e-05,
+      "loss": 3.9688,
+      "step": 9920
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.149737145209762e-05,
+      "loss": 3.3281,
+      "step": 9921
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.1495630145490105e-05,
+      "loss": 2.8594,
+      "step": 9922
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.1493888708738854e-05,
+      "loss": 3.4844,
+      "step": 9923
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.149214714186358e-05,
+      "loss": 7.4062,
+      "step": 9924
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.149040544488399e-05,
+      "loss": 2.4688,
+      "step": 9925
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.14886636178198e-05,
+      "loss": 4.9062,
+      "step": 9926
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.1486921660690736e-05,
+      "loss": 2.5312,
+      "step": 9927
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.1485179573516526e-05,
+      "loss": 3.4688,
+      "step": 9928
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.1483437356316876e-05,
+      "loss": 2.875,
+      "step": 9929
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.148169500911152e-05,
+      "loss": 4.2031,
+      "step": 9930
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.147995253192018e-05,
+      "loss": 5.2812,
+      "step": 9931
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.147820992476259e-05,
+      "loss": 2.4688,
+      "step": 9932
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.1476467187658475e-05,
+      "loss": 3.5312,
+      "step": 9933
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.1474724320627555e-05,
+      "loss": 3.2812,
+      "step": 9934
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.147298132368957e-05,
+      "loss": 2.1562,
+      "step": 9935
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.1471238196864255e-05,
+      "loss": 5.3594,
+      "step": 9936
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.1469494940171346e-05,
+      "loss": 3.6406,
+      "step": 9937
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.1467751553630556e-05,
+      "loss": 2.0312,
+      "step": 9938
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.146600803726165e-05,
+      "loss": 5.3594,
+      "step": 9939
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.146426439108436e-05,
+      "loss": 4.0312,
+      "step": 9940
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.1462520615118423e-05,
+      "loss": 3.5,
+      "step": 9941
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.146077670938358e-05,
+      "loss": 3.6875,
+      "step": 9942
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.145903267389956e-05,
+      "loss": 2.875,
+      "step": 9943
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.145728850868614e-05,
+      "loss": 5.3906,
+      "step": 9944
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.145554421376304e-05,
+      "loss": 3.875,
+      "step": 9945
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.1453799789150006e-05,
+      "loss": 3.3281,
+      "step": 9946
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.14520552348668e-05,
+      "loss": 3.4062,
+      "step": 9947
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.1450310550933173e-05,
+      "loss": 2.6875,
+      "step": 9948
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.144856573736887e-05,
+      "loss": 3.2188,
+      "step": 9949
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.144682079419365e-05,
+      "loss": 1.9062,
+      "step": 9950
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.144507572142725e-05,
+      "loss": 3.3594,
+      "step": 9951
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.144333051908946e-05,
+      "loss": 3.0625,
+      "step": 9952
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.14415851872e-05,
+      "loss": 3.3438,
+      "step": 9953
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.143983972577865e-05,
+      "loss": 4.2344,
+      "step": 9954
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.143809413484518e-05,
+      "loss": 6.4375,
+      "step": 9955
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.143634841441933e-05,
+      "loss": 2.5312,
+      "step": 9956
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.1434602564520874e-05,
+      "loss": 2.3281,
+      "step": 9957
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.1432856585169576e-05,
+      "loss": 6.4688,
+      "step": 9958
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.1431110476385206e-05,
+      "loss": 5.0,
+      "step": 9959
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.1429364238187524e-05,
+      "loss": 2.0,
+      "step": 9960
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.142761787059631e-05,
+      "loss": 2.5312,
+      "step": 9961
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.142587137363132e-05,
+      "loss": 3.3594,
+      "step": 9962
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.142412474731234e-05,
+      "loss": 2.6875,
+      "step": 9963
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.1422377991659144e-05,
+      "loss": 2.9531,
+      "step": 9964
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.14206311066915e-05,
+      "loss": 2.8438,
+      "step": 9965
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.1418884092429185e-05,
+      "loss": 4.5,
+      "step": 9966
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.141713694889198e-05,
+      "loss": 6.2031,
+      "step": 9967
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.1415389676099666e-05,
+      "loss": 3.0625,
+      "step": 9968
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.141364227407202e-05,
+      "loss": 9.1875,
+      "step": 9969
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.1411894742828826e-05,
+      "loss": 3.3125,
+      "step": 9970
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.141014708238987e-05,
+      "loss": 2.7188,
+      "step": 9971
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.140839929277494e-05,
+      "loss": 2.9375,
+      "step": 9972
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.140665137400381e-05,
+      "loss": 2.6562,
+      "step": 9973
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.140490332609629e-05,
+      "loss": 2.8125,
+      "step": 9974
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.140315514907215e-05,
+      "loss": 2.4688,
+      "step": 9975
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.1401406842951194e-05,
+      "loss": 3.125,
+      "step": 9976
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.1399658407753206e-05,
+      "loss": 4.9531,
+      "step": 9977
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.139790984349799e-05,
+      "loss": 4.25,
+      "step": 9978
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.1396161150205324e-05,
+      "loss": 4.0312,
+      "step": 9979
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.1394412327895024e-05,
+      "loss": 3.8125,
+      "step": 9980
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.1392663376586885e-05,
+      "loss": 3.9062,
+      "step": 9981
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.13909142963007e-05,
+      "loss": 2.7812,
+      "step": 9982
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.138916508705628e-05,
+      "loss": 3.3438,
+      "step": 9983
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.138741574887342e-05,
+      "loss": 1.875,
+      "step": 9984
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.138566628177192e-05,
+      "loss": 2.4062,
+      "step": 9985
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.13839166857716e-05,
+      "loss": 2.5938,
+      "step": 9986
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.138216696089227e-05,
+      "loss": 4.75,
+      "step": 9987
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.1380417107153715e-05,
+      "loss": 4.5312,
+      "step": 9988
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.1378667124575766e-05,
+      "loss": 3.5156,
+      "step": 9989
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.137691701317823e-05,
+      "loss": 2.7031,
+      "step": 9990
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.137516677298091e-05,
+      "loss": 4.4062,
+      "step": 9991
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.137341640400364e-05,
+      "loss": 3.375,
+      "step": 9992
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.137166590626622e-05,
+      "loss": 2.7656,
+      "step": 9993
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.1369915279788474e-05,
+      "loss": 4.7188,
+      "step": 9994
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.136816452459022e-05,
+      "loss": 3.7969,
+      "step": 9995
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.136641364069128e-05,
+      "loss": 4.0938,
+      "step": 9996
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.136466262811148e-05,
+      "loss": 1.6875,
+      "step": 9997
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.1362911486870636e-05,
+      "loss": 2.3906,
+      "step": 9998
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.1361160216988575e-05,
+      "loss": 2.8125,
+      "step": 9999
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.135940881848512e-05,
+      "loss": 5.0625,
+      "step": 10000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.1357657291380115e-05,
+      "loss": 1.6875,
+      "step": 10001
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.135590563569337e-05,
+      "loss": 4.5625,
+      "step": 10002
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.135415385144472e-05,
+      "loss": 7.1406,
+      "step": 10003
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.1352401938654e-05,
+      "loss": 2.6406,
+      "step": 10004
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.1350649897341044e-05,
+      "loss": 3.4062,
+      "step": 10005
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.134889772752569e-05,
+      "loss": 3.4688,
+      "step": 10006
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.134714542922777e-05,
+      "loss": 4.4062,
+      "step": 10007
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.134539300246712e-05,
+      "loss": 3.1406,
+      "step": 10008
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.1343640447263597e-05,
+      "loss": 2.3125,
+      "step": 10009
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.1341887763637014e-05,
+      "loss": 3.9375,
+      "step": 10010
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.134013495160723e-05,
+      "loss": 2.0625,
+      "step": 10011
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.133838201119409e-05,
+      "loss": 3.625,
+      "step": 10012
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.133662894241743e-05,
+      "loss": 2.5,
+      "step": 10013
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.133487574529711e-05,
+      "loss": 3.125,
+      "step": 10014
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.133312241985297e-05,
+      "loss": 3.9688,
+      "step": 10015
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.133136896610485e-05,
+      "loss": 3.8438,
+      "step": 10016
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.132961538407262e-05,
+      "loss": 2.625,
+      "step": 10017
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.132786167377612e-05,
+      "loss": 4.75,
+      "step": 10018
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.1326107835235205e-05,
+      "loss": 3.4375,
+      "step": 10019
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.132435386846974e-05,
+      "loss": 5.375,
+      "step": 10020
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.132259977349957e-05,
+      "loss": 2.0938,
+      "step": 10021
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.1320845550344556e-05,
+      "loss": 5.7188,
+      "step": 10022
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.131909119902457e-05,
+      "loss": 2.2812,
+      "step": 10023
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.131733671955947e-05,
+      "loss": 5.0938,
+      "step": 10024
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.131558211196909e-05,
+      "loss": 2.0625,
+      "step": 10025
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.131382737627333e-05,
+      "loss": 1.9062,
+      "step": 10026
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.131207251249204e-05,
+      "loss": 2.875,
+      "step": 10027
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.131031752064509e-05,
+      "loss": 3.2188,
+      "step": 10028
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.1308562400752344e-05,
+      "loss": 2.4062,
+      "step": 10029
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.130680715283368e-05,
+      "loss": 3.1719,
+      "step": 10030
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.1305051776908965e-05,
+      "loss": 4.6719,
+      "step": 10031
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.130329627299807e-05,
+      "loss": 3.3281,
+      "step": 10032
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.130154064112088e-05,
+      "loss": 2.4062,
+      "step": 10033
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.129978488129725e-05,
+      "loss": 2.5156,
+      "step": 10034
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.129802899354707e-05,
+      "loss": 3.5312,
+      "step": 10035
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.1296272977890224e-05,
+      "loss": 2.625,
+      "step": 10036
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.129451683434658e-05,
+      "loss": 2.6875,
+      "step": 10037
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.129276056293604e-05,
+      "loss": 3.9062,
+      "step": 10038
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.129100416367846e-05,
+      "loss": 1.5312,
+      "step": 10039
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.128924763659374e-05,
+      "loss": 4.4062,
+      "step": 10040
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.128749098170176e-05,
+      "loss": 5.0312,
+      "step": 10041
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.1285734199022416e-05,
+      "loss": 2.8594,
+      "step": 10042
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.128397728857559e-05,
+      "loss": 3.1562,
+      "step": 10043
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.128222025038117e-05,
+      "loss": 2.4062,
+      "step": 10044
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.1280463084459056e-05,
+      "loss": 1.6562,
+      "step": 10045
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.127870579082913e-05,
+      "loss": 2.125,
+      "step": 10046
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.12769483695113e-05,
+      "loss": 4.875,
+      "step": 10047
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.127519082052545e-05,
+      "loss": 2.9062,
+      "step": 10048
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.1273433143891485e-05,
+      "loss": 5.0,
+      "step": 10049
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.1271675339629304e-05,
+      "loss": 4.8906,
+      "step": 10050
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.126991740775879e-05,
+      "loss": 4.9375,
+      "step": 10051
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.1268159348299874e-05,
+      "loss": 3.3281,
+      "step": 10052
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.126640116127244e-05,
+      "loss": 2.5312,
+      "step": 10053
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.1264642846696404e-05,
+      "loss": 3.125,
+      "step": 10054
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.126288440459166e-05,
+      "loss": 4.3438,
+      "step": 10055
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.126112583497812e-05,
+      "loss": 3.1875,
+      "step": 10056
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.12593671378757e-05,
+      "loss": 2.6875,
+      "step": 10057
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.12576083133043e-05,
+      "loss": 4.0625,
+      "step": 10058
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.125584936128383e-05,
+      "loss": 2.375,
+      "step": 10059
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.1254090281834217e-05,
+      "loss": 2.8125,
+      "step": 10060
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.125233107497537e-05,
+      "loss": 4.8438,
+      "step": 10061
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.12505717407272e-05,
+      "loss": 2.8125,
+      "step": 10062
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.124881227910963e-05,
+      "loss": 4.2891,
+      "step": 10063
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.124705269014259e-05,
+      "loss": 4.3125,
+      "step": 10064
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.124529297384597e-05,
+      "loss": 2.4375,
+      "step": 10065
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.124353313023972e-05,
+      "loss": 2.125,
+      "step": 10066
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.124177315934375e-05,
+      "loss": 2.0312,
+      "step": 10067
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.124001306117799e-05,
+      "loss": 4.9062,
+      "step": 10068
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.1238252835762365e-05,
+      "loss": 3.4219,
+      "step": 10069
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.1236492483116805e-05,
+      "loss": 2.1562,
+      "step": 10070
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.1234732003261235e-05,
+      "loss": 3.375,
+      "step": 10071
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.123297139621559e-05,
+      "loss": 2.5938,
+      "step": 10072
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.12312106619998e-05,
+      "loss": 3.1094,
+      "step": 10073
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.1229449800633794e-05,
+      "loss": 2.9062,
+      "step": 10074
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.1227688812137514e-05,
+      "loss": 2.0469,
+      "step": 10075
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.122592769653089e-05,
+      "loss": 2.9219,
+      "step": 10076
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.122416645383387e-05,
+      "loss": 2.25,
+      "step": 10077
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.122240508406639e-05,
+      "loss": 4.375,
+      "step": 10078
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.122064358724837e-05,
+      "loss": 4.4844,
+      "step": 10079
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.1218881963399785e-05,
+      "loss": 3.2812,
+      "step": 10080
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.121712021254056e-05,
+      "loss": 2.3125,
+      "step": 10081
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.121535833469065e-05,
+      "loss": 3.4219,
+      "step": 10082
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.121359632986998e-05,
+      "loss": 3.4375,
+      "step": 10083
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.121183419809852e-05,
+      "loss": 2.8906,
+      "step": 10084
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.1210071939396214e-05,
+      "loss": 3.0469,
+      "step": 10085
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.120830955378301e-05,
+      "loss": 3.6562,
+      "step": 10086
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.120654704127886e-05,
+      "loss": 3.1094,
+      "step": 10087
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.1204784401903716e-05,
+      "loss": 2.9062,
+      "step": 10088
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.120302163567755e-05,
+      "loss": 2.0469,
+      "step": 10089
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.120125874262029e-05,
+      "loss": 1.6719,
+      "step": 10090
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.1199495722751914e-05,
+      "loss": 2.2188,
+      "step": 10091
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.119773257609237e-05,
+      "loss": 4.0312,
+      "step": 10092
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.119596930266163e-05,
+      "loss": 2.3438,
+      "step": 10093
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.119420590247965e-05,
+      "loss": 5.4688,
+      "step": 10094
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.11924423755664e-05,
+      "loss": 2.7344,
+      "step": 10095
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.119067872194183e-05,
+      "loss": 2.2188,
+      "step": 10096
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.118891494162593e-05,
+      "loss": 2.625,
+      "step": 10097
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.1187151034638644e-05,
+      "loss": 2.375,
+      "step": 10098
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.118538700099995e-05,
+      "loss": 2.2969,
+      "step": 10099
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.118362284072983e-05,
+      "loss": 4.1562,
+      "step": 10100
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.118185855384824e-05,
+      "loss": 3.25,
+      "step": 10101
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.118009414037516e-05,
+      "loss": 3.7344,
+      "step": 10102
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.117832960033057e-05,
+      "loss": 4.9375,
+      "step": 10103
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.117656493373444e-05,
+      "loss": 2.5312,
+      "step": 10104
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.1174800140606756e-05,
+      "loss": 3.0,
+      "step": 10105
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.117303522096749e-05,
+      "loss": 3.3438,
+      "step": 10106
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.117127017483662e-05,
+      "loss": 3.625,
+      "step": 10107
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.116950500223414e-05,
+      "loss": 2.7031,
+      "step": 10108
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.1167739703180024e-05,
+      "loss": 2.6562,
+      "step": 10109
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.116597427769426e-05,
+      "loss": 3.3281,
+      "step": 10110
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.116420872579684e-05,
+      "loss": 2.5312,
+      "step": 10111
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.116244304750774e-05,
+      "loss": 3.7031,
+      "step": 10112
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.116067724284696e-05,
+      "loss": 2.7344,
+      "step": 10113
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.115891131183449e-05,
+      "loss": 4.1406,
+      "step": 10114
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.115714525449032e-05,
+      "loss": 2.25,
+      "step": 10115
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.115537907083444e-05,
+      "loss": 3.6562,
+      "step": 10116
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.1153612760886856e-05,
+      "loss": 3.75,
+      "step": 10117
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.115184632466755e-05,
+      "loss": 2.5,
+      "step": 10118
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.1150079762196534e-05,
+      "loss": 1.6875,
+      "step": 10119
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.11483130734938e-05,
+      "loss": 2.125,
+      "step": 10120
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.114654625857934e-05,
+      "loss": 5.375,
+      "step": 10121
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.1144779317473176e-05,
+      "loss": 2.1562,
+      "step": 10122
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.114301225019531e-05,
+      "loss": 3.0625,
+      "step": 10123
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.1141245056765725e-05,
+      "loss": 3.2812,
+      "step": 10124
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.1139477737204446e-05,
+      "loss": 3.5,
+      "step": 10125
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.113771029153149e-05,
+      "loss": 3.0469,
+      "step": 10126
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.113594271976684e-05,
+      "loss": 4.0625,
+      "step": 10127
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.1134175021930526e-05,
+      "loss": 3.6562,
+      "step": 10128
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.113240719804255e-05,
+      "loss": 4.8594,
+      "step": 10129
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.113063924812294e-05,
+      "loss": 3.5625,
+      "step": 10130
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.11288711721917e-05,
+      "loss": 3.8594,
+      "step": 10131
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.112710297026884e-05,
+      "loss": 2.5,
+      "step": 10132
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.1125334642374396e-05,
+      "loss": 2.2969,
+      "step": 10133
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.112356618852838e-05,
+      "loss": 2.5156,
+      "step": 10134
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.1121797608750805e-05,
+      "loss": 3.2188,
+      "step": 10135
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.112002890306171e-05,
+      "loss": 3.1562,
+      "step": 10136
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.11182600714811e-05,
+      "loss": 4.6719,
+      "step": 10137
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.1116491114029007e-05,
+      "loss": 3.2344,
+      "step": 10138
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.1114722030725465e-05,
+      "loss": 3.2812,
+      "step": 10139
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.11129528215905e-05,
+      "loss": 2.4219,
+      "step": 10140
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.1111183486644126e-05,
+      "loss": 3.6875,
+      "step": 10141
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.110941402590639e-05,
+      "loss": 2.375,
+      "step": 10142
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.1107644439397325e-05,
+      "loss": 3.5938,
+      "step": 10143
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.110587472713696e-05,
+      "loss": 3.75,
+      "step": 10144
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.110410488914532e-05,
+      "loss": 1.4688,
+      "step": 10145
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.1102334925442464e-05,
+      "loss": 4.0078,
+      "step": 10146
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.1100564836048414e-05,
+      "loss": 4.5,
+      "step": 10147
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.109879462098321e-05,
+      "loss": 3.2969,
+      "step": 10148
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.10970242802669e-05,
+      "loss": 10.25,
+      "step": 10149
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.109525381391952e-05,
+      "loss": 3.7031,
+      "step": 10150
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.109348322196112e-05,
+      "loss": 2.1562,
+      "step": 10151
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.109171250441174e-05,
+      "loss": 4.3438,
+      "step": 10152
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.108994166129142e-05,
+      "loss": 3.5469,
+      "step": 10153
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.1088170692620226e-05,
+      "loss": 3.5938,
+      "step": 10154
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.108639959841819e-05,
+      "loss": 3.25,
+      "step": 10155
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.1084628378705375e-05,
+      "loss": 2.5,
+      "step": 10156
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.108285703350183e-05,
+      "loss": 3.9531,
+      "step": 10157
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.1081085562827605e-05,
+      "loss": 2.5,
+      "step": 10158
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.1079313966702746e-05,
+      "loss": 4.2812,
+      "step": 10159
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.107754224514733e-05,
+      "loss": 2.375,
+      "step": 10160
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.107577039818141e-05,
+      "loss": 2.3438,
+      "step": 10161
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.1073998425825034e-05,
+      "loss": 3.0312,
+      "step": 10162
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.1072226328098275e-05,
+      "loss": 2.0,
+      "step": 10163
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.107045410502118e-05,
+      "loss": 3.25,
+      "step": 10164
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.106868175661383e-05,
+      "loss": 2.7812,
+      "step": 10165
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.106690928289628e-05,
+      "loss": 3.3438,
+      "step": 10166
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.10651366838886e-05,
+      "loss": 4.6719,
+      "step": 10167
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.106336395961086e-05,
+      "loss": 3.125,
+      "step": 10168
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.106159111008312e-05,
+      "loss": 3.4844,
+      "step": 10169
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.105981813532545e-05,
+      "loss": 3.0469,
+      "step": 10170
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.105804503535794e-05,
+      "loss": 4.7812,
+      "step": 10171
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.105627181020066e-05,
+      "loss": 3.5625,
+      "step": 10172
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.105449845987366e-05,
+      "loss": 3.0625,
+      "step": 10173
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.105272498439704e-05,
+      "loss": 2.1875,
+      "step": 10174
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.1050951383790876e-05,
+      "loss": 3.2812,
+      "step": 10175
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.1049177658075236e-05,
+      "loss": 3.9062,
+      "step": 10176
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.104740380727021e-05,
+      "loss": 2.875,
+      "step": 10177
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.104562983139587e-05,
+      "loss": 3.0938,
+      "step": 10178
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.1043855730472316e-05,
+      "loss": 2.6406,
+      "step": 10179
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.1042081504519624e-05,
+      "loss": 6.3828,
+      "step": 10180
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.104030715355788e-05,
+      "loss": 3.5781,
+      "step": 10181
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.103853267760716e-05,
+      "loss": 2.6094,
+      "step": 10182
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.103675807668758e-05,
+      "loss": 3.8438,
+      "step": 10183
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.1034983350819204e-05,
+      "loss": 5.4844,
+      "step": 10184
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.103320850002214e-05,
+      "loss": 4.7812,
+      "step": 10185
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.103143352431647e-05,
+      "loss": 3.9375,
+      "step": 10186
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.10296584237223e-05,
+      "loss": 4.5625,
+      "step": 10187
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.1027883198259725e-05,
+      "loss": 3.375,
+      "step": 10188
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.102610784794884e-05,
+      "loss": 2.0625,
+      "step": 10189
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.1024332372809735e-05,
+      "loss": 5.0625,
+      "step": 10190
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.102255677286252e-05,
+      "loss": 2.375,
+      "step": 10191
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.102078104812729e-05,
+      "loss": 2.25,
+      "step": 10192
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.101900519862416e-05,
+      "loss": 3.9219,
+      "step": 10193
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.101722922437323e-05,
+      "loss": 2.7812,
+      "step": 10194
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.10154531253946e-05,
+      "loss": 2.9844,
+      "step": 10195
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.101367690170839e-05,
+      "loss": 2.4688,
+      "step": 10196
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.1011900553334695e-05,
+      "loss": 3.9062,
+      "step": 10197
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.101012408029362e-05,
+      "loss": 5.1719,
+      "step": 10198
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.1008347482605303e-05,
+      "loss": 2.625,
+      "step": 10199
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.1006570760289835e-05,
+      "loss": 3.1406,
+      "step": 10200
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.100479391336734e-05,
+      "loss": 3.2812,
+      "step": 10201
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.1003016941857925e-05,
+      "loss": 3.5938,
+      "step": 10202
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.100123984578172e-05,
+      "loss": 3.7812,
+      "step": 10203
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.099946262515884e-05,
+      "loss": 3.0156,
+      "step": 10204
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.0997685280009396e-05,
+      "loss": 2.2031,
+      "step": 10205
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.099590781035351e-05,
+      "loss": 2.8125,
+      "step": 10206
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.099413021621133e-05,
+      "loss": 3.625,
+      "step": 10207
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.099235249760295e-05,
+      "loss": 2.5312,
+      "step": 10208
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.09905746545485e-05,
+      "loss": 4.3438,
+      "step": 10209
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.098879668706812e-05,
+      "loss": 3.2812,
+      "step": 10210
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.098701859518194e-05,
+      "loss": 3.6875,
+      "step": 10211
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.0985240378910084e-05,
+      "loss": 3.1875,
+      "step": 10212
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.098346203827267e-05,
+      "loss": 5.0,
+      "step": 10213
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.0981683573289844e-05,
+      "loss": 3.375,
+      "step": 10214
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.097990498398175e-05,
+      "loss": 2.2188,
+      "step": 10215
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.097812627036851e-05,
+      "loss": 2.4375,
+      "step": 10216
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.097634743247026e-05,
+      "loss": 1.375,
+      "step": 10217
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.0974568470307155e-05,
+      "loss": 3.4375,
+      "step": 10218
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.097278938389931e-05,
+      "loss": 3.2344,
+      "step": 10219
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.0971010173266884e-05,
+      "loss": 2.9688,
+      "step": 10220
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.096923083843002e-05,
+      "loss": 3.1875,
+      "step": 10221
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.096745137940885e-05,
+      "loss": 4.7812,
+      "step": 10222
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.096567179622353e-05,
+      "loss": 4.9688,
+      "step": 10223
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.09638920888942e-05,
+      "loss": 2.6562,
+      "step": 10224
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.096211225744101e-05,
+      "loss": 2.0625,
+      "step": 10225
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.096033230188412e-05,
+      "loss": 3.1094,
+      "step": 10226
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.0958552222243665e-05,
+      "loss": 2.7812,
+      "step": 10227
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.095677201853981e-05,
+      "loss": 3.0938,
+      "step": 10228
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.09549916907927e-05,
+      "loss": 1.4375,
+      "step": 10229
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.0953211239022497e-05,
+      "loss": 3.7812,
+      "step": 10230
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.0951430663249354e-05,
+      "loss": 1.0938,
+      "step": 10231
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.094964996349343e-05,
+      "loss": 3.4219,
+      "step": 10232
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.0947869139774894e-05,
+      "loss": 4.2656,
+      "step": 10233
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.094608819211389e-05,
+      "loss": 2.6875,
+      "step": 10234
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.0944307120530586e-05,
+      "loss": 3.3125,
+      "step": 10235
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.094252592504515e-05,
+      "loss": 3.4062,
+      "step": 10236
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.094074460567775e-05,
+      "loss": 3.3281,
+      "step": 10237
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.093896316244855e-05,
+      "loss": 4.5156,
+      "step": 10238
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.09371815953777e-05,
+      "loss": 2.0938,
+      "step": 10239
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.0935399904485396e-05,
+      "loss": 2.8438,
+      "step": 10240
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.0933618089791795e-05,
+      "loss": 3.5625,
+      "step": 10241
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.0931836151317075e-05,
+      "loss": 1.8438,
+      "step": 10242
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.093005408908141e-05,
+      "loss": 2.7656,
+      "step": 10243
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.092827190310497e-05,
+      "loss": 3.3125,
+      "step": 10244
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.092648959340793e-05,
+      "loss": 2.625,
+      "step": 10245
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.092470716001047e-05,
+      "loss": 3.2656,
+      "step": 10246
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.092292460293278e-05,
+      "loss": 3.2812,
+      "step": 10247
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.0921141922195015e-05,
+      "loss": 4.3125,
+      "step": 10248
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.0919359117817386e-05,
+      "loss": 3.375,
+      "step": 10249
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.091757618982006e-05,
+      "loss": 3.1406,
+      "step": 10250
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.091579313822322e-05,
+      "loss": 4.0938,
+      "step": 10251
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.091400996304706e-05,
+      "loss": 2.7656,
+      "step": 10252
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.091222666431177e-05,
+      "loss": 2.5625,
+      "step": 10253
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.091044324203753e-05,
+      "loss": 6.2188,
+      "step": 10254
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.090865969624453e-05,
+      "loss": 4.2188,
+      "step": 10255
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.0906876026952976e-05,
+      "loss": 2.6406,
+      "step": 10256
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.0905092234183036e-05,
+      "loss": 3.0938,
+      "step": 10257
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.090330831795493e-05,
+      "loss": 2.0938,
+      "step": 10258
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.090152427828884e-05,
+      "loss": 3.25,
+      "step": 10259
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.089974011520497e-05,
+      "loss": 3.6719,
+      "step": 10260
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.0897955828723515e-05,
+      "loss": 2.0625,
+      "step": 10261
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.0896171418864675e-05,
+      "loss": 3.7969,
+      "step": 10262
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.089438688564866e-05,
+      "loss": 3.4688,
+      "step": 10263
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.089260222909566e-05,
+      "loss": 2.1562,
+      "step": 10264
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.089081744922588e-05,
+      "loss": 3.6562,
+      "step": 10265
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.088903254605954e-05,
+      "loss": 5.5781,
+      "step": 10266
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.088724751961683e-05,
+      "loss": 4.0312,
+      "step": 10267
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.088546236991797e-05,
+      "loss": 4.2188,
+      "step": 10268
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.088367709698316e-05,
+      "loss": 2.8516,
+      "step": 10269
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.088189170083263e-05,
+      "loss": 3.375,
+      "step": 10270
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.0880106181486566e-05,
+      "loss": 3.125,
+      "step": 10271
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.087832053896521e-05,
+      "loss": 3.1719,
+      "step": 10272
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.0876534773288756e-05,
+      "loss": 4.125,
+      "step": 10273
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.087474888447743e-05,
+      "loss": 4.7812,
+      "step": 10274
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.087296287255145e-05,
+      "loss": 5.1562,
+      "step": 10275
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.087117673753103e-05,
+      "loss": 4.7812,
+      "step": 10276
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.08693904794364e-05,
+      "loss": 4.4062,
+      "step": 10277
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.086760409828779e-05,
+      "loss": 4.125,
+      "step": 10278
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.0865817594105404e-05,
+      "loss": 4.6406,
+      "step": 10279
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.086403096690947e-05,
+      "loss": 2.9531,
+      "step": 10280
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.086224421672023e-05,
+      "loss": 4.5625,
+      "step": 10281
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.0860457343557894e-05,
+      "loss": 2.1875,
+      "step": 10282
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.0858670347442704e-05,
+      "loss": 2.4688,
+      "step": 10283
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.0856883228394885e-05,
+      "loss": 3.8125,
+      "step": 10284
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.085509598643467e-05,
+      "loss": 3.8125,
+      "step": 10285
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.08533086215823e-05,
+      "loss": 5.7812,
+      "step": 10286
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.0851521133858004e-05,
+      "loss": 5.7188,
+      "step": 10287
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.0849733523282014e-05,
+      "loss": 2.25,
+      "step": 10288
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.084794578987457e-05,
+      "loss": 2.9062,
+      "step": 10289
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.0846157933655915e-05,
+      "loss": 2.375,
+      "step": 10290
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.084436995464628e-05,
+      "loss": 3.7188,
+      "step": 10291
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.0842581852865924e-05,
+      "loss": 2.7188,
+      "step": 10292
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.0840793628335083e-05,
+      "loss": 2.75,
+      "step": 10293
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.0839005281074e-05,
+      "loss": 4.1875,
+      "step": 10294
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.0837216811102915e-05,
+      "loss": 2.75,
+      "step": 10295
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.083542821844208e-05,
+      "loss": 5.3125,
+      "step": 10296
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.083363950311175e-05,
+      "loss": 4.375,
+      "step": 10297
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.083185066513217e-05,
+      "loss": 2.8125,
+      "step": 10298
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.083006170452359e-05,
+      "loss": 3.1875,
+      "step": 10299
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.082827262130627e-05,
+      "loss": 2.8438,
+      "step": 10300
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.082648341550046e-05,
+      "loss": 2.5469,
+      "step": 10301
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.082469408712641e-05,
+      "loss": 3.1094,
+      "step": 10302
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.08229046362044e-05,
+      "loss": 2.9688,
+      "step": 10303
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.0821115062754655e-05,
+      "loss": 5.6719,
+      "step": 10304
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.081932536679745e-05,
+      "loss": 3.7656,
+      "step": 10305
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.0817535548353053e-05,
+      "loss": 4.5938,
+      "step": 10306
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.081574560744172e-05,
+      "loss": 2.8125,
+      "step": 10307
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.081395554408372e-05,
+      "loss": 3.4844,
+      "step": 10308
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.081216535829932e-05,
+      "loss": 6.9062,
+      "step": 10309
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.081037505010878e-05,
+      "loss": 1.9688,
+      "step": 10310
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.080858461953237e-05,
+      "loss": 4.125,
+      "step": 10311
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.080679406659036e-05,
+      "loss": 4.4688,
+      "step": 10312
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.080500339130303e-05,
+      "loss": 4.4688,
+      "step": 10313
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.080321259369063e-05,
+      "loss": 3.1875,
+      "step": 10314
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.080142167377345e-05,
+      "loss": 4.125,
+      "step": 10315
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.079963063157177e-05,
+      "loss": 3.2656,
+      "step": 10316
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.0797839467105865e-05,
+      "loss": 2.2656,
+      "step": 10317
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.0796048180396e-05,
+      "loss": 3.8594,
+      "step": 10318
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.079425677146247e-05,
+      "loss": 4.625,
+      "step": 10319
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.079246524032554e-05,
+      "loss": 3.7812,
+      "step": 10320
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.0790673587005504e-05,
+      "loss": 2.2188,
+      "step": 10321
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.078888181152264e-05,
+      "loss": 3.0781,
+      "step": 10322
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.078708991389724e-05,
+      "loss": 3.6562,
+      "step": 10323
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.078529789414958e-05,
+      "loss": 1.8438,
+      "step": 10324
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.078350575229996e-05,
+      "loss": 4.4062,
+      "step": 10325
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.078171348836866e-05,
+      "loss": 2.8125,
+      "step": 10326
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.077992110237597e-05,
+      "loss": 2.4375,
+      "step": 10327
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.077812859434219e-05,
+      "loss": 4.4062,
+      "step": 10328
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.0776335964287604e-05,
+      "loss": 2.6562,
+      "step": 10329
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.0774543212232513e-05,
+      "loss": 3.3125,
+      "step": 10330
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.077275033819721e-05,
+      "loss": 2.125,
+      "step": 10331
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.077095734220199e-05,
+      "loss": 2.9531,
+      "step": 10332
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.076916422426716e-05,
+      "loss": 5.1875,
+      "step": 10333
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.0767370984413015e-05,
+      "loss": 3.0625,
+      "step": 10334
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.076557762265985e-05,
+      "loss": 2.7188,
+      "step": 10335
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0763784139027985e-05,
+      "loss": 5.0625,
+      "step": 10336
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.07619905335377e-05,
+      "loss": 4.8281,
+      "step": 10337
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.076019680620932e-05,
+      "loss": 2.0625,
+      "step": 10338
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.075840295706315e-05,
+      "loss": 2.1719,
+      "step": 10339
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.075660898611949e-05,
+      "loss": 2.6406,
+      "step": 10340
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0754814893398654e-05,
+      "loss": 2.4062,
+      "step": 10341
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0753020678920966e-05,
+      "loss": 5.1875,
+      "step": 10342
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0751226342706714e-05,
+      "loss": 2.6875,
+      "step": 10343
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.074943188477623e-05,
+      "loss": 4.1875,
+      "step": 10344
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.074763730514982e-05,
+      "loss": 5.0312,
+      "step": 10345
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.074584260384781e-05,
+      "loss": 4.0625,
+      "step": 10346
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.074404778089051e-05,
+      "loss": 4.1094,
+      "step": 10347
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.074225283629824e-05,
+      "loss": 2.75,
+      "step": 10348
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0740457770091324e-05,
+      "loss": 6.8047,
+      "step": 10349
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.073866258229008e-05,
+      "loss": 5.125,
+      "step": 10350
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.073686727291484e-05,
+      "loss": 7.4688,
+      "step": 10351
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.073507184198592e-05,
+      "loss": 4.75,
+      "step": 10352
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.073327628952365e-05,
+      "loss": 6.2344,
+      "step": 10353
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.073148061554836e-05,
+      "loss": 3.9219,
+      "step": 10354
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0729684820080374e-05,
+      "loss": 5.4062,
+      "step": 10355
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.072788890314003e-05,
+      "loss": 4.0,
+      "step": 10356
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.072609286474765e-05,
+      "loss": 5.4062,
+      "step": 10357
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0724296704923576e-05,
+      "loss": 3.4062,
+      "step": 10358
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.072250042368813e-05,
+      "loss": 3.4688,
+      "step": 10359
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0720704021061674e-05,
+      "loss": 5.375,
+      "step": 10360
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0718907497064515e-05,
+      "loss": 3.9844,
+      "step": 10361
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0717110851717004e-05,
+      "loss": 3.9844,
+      "step": 10362
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.071531408503949e-05,
+      "loss": 3.0469,
+      "step": 10363
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.0713517197052305e-05,
+      "loss": 5.0,
+      "step": 10364
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.071172018777579e-05,
+      "loss": 4.5156,
+      "step": 10365
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0709923057230294e-05,
+      "loss": 7.0625,
+      "step": 10366
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0708125805436155e-05,
+      "loss": 4.3125,
+      "step": 10367
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.070632843241374e-05,
+      "loss": 5.125,
+      "step": 10368
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0704530938183375e-05,
+      "loss": 6.5781,
+      "step": 10369
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.070273332276541e-05,
+      "loss": 3.6562,
+      "step": 10370
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.070093558618022e-05,
+      "loss": 4.9688,
+      "step": 10371
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.069913772844813e-05,
+      "loss": 2.8125,
+      "step": 10372
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.069733974958951e-05,
+      "loss": 4.5938,
+      "step": 10373
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0695541649624705e-05,
+      "loss": 5.1875,
+      "step": 10374
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0693743428574085e-05,
+      "loss": 2.9844,
+      "step": 10375
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0691945086458e-05,
+      "loss": 1.8906,
+      "step": 10376
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.069014662329681e-05,
+      "loss": 3.0938,
+      "step": 10377
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0688348039110866e-05,
+      "loss": 4.5625,
+      "step": 10378
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0686549333920546e-05,
+      "loss": 3.8594,
+      "step": 10379
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.068475050774621e-05,
+      "loss": 3.7812,
+      "step": 10380
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0682951560608204e-05,
+      "loss": 3.8281,
+      "step": 10381
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.068115249252692e-05,
+      "loss": 2.2188,
+      "step": 10382
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0679353303522725e-05,
+      "loss": 4.1562,
+      "step": 10383
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.067755399361596e-05,
+      "loss": 2.75,
+      "step": 10384
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.067575456282702e-05,
+      "loss": 2.8906,
+      "step": 10385
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.067395501117627e-05,
+      "loss": 3.5156,
+      "step": 10386
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.067215533868408e-05,
+      "loss": 3.9531,
+      "step": 10387
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.067035554537084e-05,
+      "loss": 3.625,
+      "step": 10388
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.06685556312569e-05,
+      "loss": 5.4062,
+      "step": 10389
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.066675559636265e-05,
+      "loss": 3.1562,
+      "step": 10390
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.066495544070847e-05,
+      "loss": 3.2031,
+      "step": 10391
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.066315516431474e-05,
+      "loss": 2.0,
+      "step": 10392
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.0661354767201845e-05,
+      "loss": 2.2031,
+      "step": 10393
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.065955424939015e-05,
+      "loss": 4.375,
+      "step": 10394
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.065775361090006e-05,
+      "loss": 3.0938,
+      "step": 10395
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.065595285175195e-05,
+      "loss": 4.5625,
+      "step": 10396
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.06541519719662e-05,
+      "loss": 5.125,
+      "step": 10397
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.065235097156321e-05,
+      "loss": 3.9219,
+      "step": 10398
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.065054985056337e-05,
+      "loss": 2.875,
+      "step": 10399
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.064874860898706e-05,
+      "loss": 4.2031,
+      "step": 10400
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.064694724685468e-05,
+      "loss": 2.375,
+      "step": 10401
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.0645145764186625e-05,
+      "loss": 2.2188,
+      "step": 10402
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.064334416100328e-05,
+      "loss": 4.1484,
+      "step": 10403
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.064154243732506e-05,
+      "loss": 3.5625,
+      "step": 10404
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.063974059317234e-05,
+      "loss": 4.1562,
+      "step": 10405
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.063793862856553e-05,
+      "loss": 2.4219,
+      "step": 10406
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.063613654352503e-05,
+      "loss": 3.1719,
+      "step": 10407
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.063433433807124e-05,
+      "loss": 2.5625,
+      "step": 10408
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.0632532012224565e-05,
+      "loss": 3.2188,
+      "step": 10409
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.063072956600541e-05,
+      "loss": 2.0625,
+      "step": 10410
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.062892699943419e-05,
+      "loss": 4.125,
+      "step": 10411
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.062712431253129e-05,
+      "loss": 2.7188,
+      "step": 10412
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.0625321505317126e-05,
+      "loss": 4.875,
+      "step": 10413
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.0623518577812116e-05,
+      "loss": 2.5625,
+      "step": 10414
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.062171553003667e-05,
+      "loss": 2.7031,
+      "step": 10415
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.061991236201119e-05,
+      "loss": 2.0938,
+      "step": 10416
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.061810907375611e-05,
+      "loss": 7.0312,
+      "step": 10417
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.061630566529183e-05,
+      "loss": 5.4062,
+      "step": 10418
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.061450213663876e-05,
+      "loss": 4.0,
+      "step": 10419
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.0612698487817334e-05,
+      "loss": 1.7188,
+      "step": 10420
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.061089471884796e-05,
+      "loss": 4.0312,
+      "step": 10421
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.0609090829751074e-05,
+      "loss": 2.2188,
+      "step": 10422
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.0607286820547083e-05,
+      "loss": 2.2656,
+      "step": 10423
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.0605482691256406e-05,
+      "loss": 4.4688,
+      "step": 10424
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.060367844189949e-05,
+      "loss": 2.5938,
+      "step": 10425
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.060187407249674e-05,
+      "loss": 6.1719,
+      "step": 10426
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.0600069583068594e-05,
+      "loss": 3.4375,
+      "step": 10427
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.059826497363548e-05,
+      "loss": 2.8906,
+      "step": 10428
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.059646024421783e-05,
+      "loss": 3.5938,
+      "step": 10429
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.059465539483607e-05,
+      "loss": 2.9688,
+      "step": 10430
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.059285042551063e-05,
+      "loss": 4.5312,
+      "step": 10431
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.059104533626196e-05,
+      "loss": 4.125,
+      "step": 10432
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.058924012711048e-05,
+      "loss": 2.75,
+      "step": 10433
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.0587434798076636e-05,
+      "loss": 3.0312,
+      "step": 10434
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.0585629349180855e-05,
+      "loss": 2.3125,
+      "step": 10435
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.058382378044359e-05,
+      "loss": 4.7031,
+      "step": 10436
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.0582018091885276e-05,
+      "loss": 4.3125,
+      "step": 10437
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.058021228352635e-05,
+      "loss": 2.5312,
+      "step": 10438
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.057840635538727e-05,
+      "loss": 2.4375,
+      "step": 10439
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.057660030748847e-05,
+      "loss": 3.4062,
+      "step": 10440
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.05747941398504e-05,
+      "loss": 3.6562,
+      "step": 10441
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.0572987852493516e-05,
+      "loss": 1.5312,
+      "step": 10442
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.057118144543825e-05,
+      "loss": 7.2812,
+      "step": 10443
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.056937491870507e-05,
+      "loss": 2.7969,
+      "step": 10444
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.056756827231441e-05,
+      "loss": 3.9844,
+      "step": 10445
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.056576150628673e-05,
+      "loss": 3.2812,
+      "step": 10446
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.0563954620642496e-05,
+      "loss": 4.7344,
+      "step": 10447
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.056214761540216e-05,
+      "loss": 6.2969,
+      "step": 10448
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.0560340490586164e-05,
+      "loss": 4.1094,
+      "step": 10449
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.055853324621498e-05,
+      "loss": 3.125,
+      "step": 10450
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.055672588230907e-05,
+      "loss": 2.9688,
+      "step": 10451
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.0554918398888886e-05,
+      "loss": 2.6562,
+      "step": 10452
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.0553110795974904e-05,
+      "loss": 3.2969,
+      "step": 10453
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.055130307358757e-05,
+      "loss": 5.2344,
+      "step": 10454
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.054949523174736e-05,
+      "loss": 5.1875,
+      "step": 10455
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.0547687270474754e-05,
+      "loss": 2.4531,
+      "step": 10456
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.054587918979019e-05,
+      "loss": 4.125,
+      "step": 10457
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0544070989714166e-05,
+      "loss": 1.875,
+      "step": 10458
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0542262670267136e-05,
+      "loss": 2.875,
+      "step": 10459
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.054045423146958e-05,
+      "loss": 3.3438,
+      "step": 10460
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0538645673341966e-05,
+      "loss": 4.1719,
+      "step": 10461
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0536836995904775e-05,
+      "loss": 5.7812,
+      "step": 10462
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.053502819917848e-05,
+      "loss": 3.1875,
+      "step": 10463
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.053321928318355e-05,
+      "loss": 2.7188,
+      "step": 10464
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.053141024794048e-05,
+      "loss": 3.1406,
+      "step": 10465
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.052960109346974e-05,
+      "loss": 3.5938,
+      "step": 10466
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0527791819791824e-05,
+      "loss": 4.5938,
+      "step": 10467
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.05259824269272e-05,
+      "loss": 3.4688,
+      "step": 10468
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0524172914896365e-05,
+      "loss": 5.5312,
+      "step": 10469
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.052236328371979e-05,
+      "loss": 6.3906,
+      "step": 10470
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.052055353341797e-05,
+      "loss": 1.5156,
+      "step": 10471
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0518743664011395e-05,
+      "loss": 2.375,
+      "step": 10472
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.051693367552055e-05,
+      "loss": 4.2656,
+      "step": 10473
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0515123567965937e-05,
+      "loss": 4.4844,
+      "step": 10474
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.051331334136804e-05,
+      "loss": 1.125,
+      "step": 10475
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0511502995747348e-05,
+      "loss": 3.1953,
+      "step": 10476
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0509692531124375e-05,
+      "loss": 3.0469,
+      "step": 10477
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0507881947519587e-05,
+      "loss": 1.7188,
+      "step": 10478
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0506071244953515e-05,
+      "loss": 4.7812,
+      "step": 10479
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0504260423446633e-05,
+      "loss": 2.2344,
+      "step": 10480
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0502449483019453e-05,
+      "loss": 1.4375,
+      "step": 10481
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.050063842369248e-05,
+      "loss": 2.4062,
+      "step": 10482
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0498827245486205e-05,
+      "loss": 3.5312,
+      "step": 10483
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0497015948421152e-05,
+      "loss": 3.9219,
+      "step": 10484
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0495204532517806e-05,
+      "loss": 4.1875,
+      "step": 10485
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.0493392997796687e-05,
+      "loss": 3.5,
+      "step": 10486
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0491581344278303e-05,
+      "loss": 2.0,
+      "step": 10487
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.048976957198315e-05,
+      "loss": 2.5156,
+      "step": 10488
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.048795768093176e-05,
+      "loss": 3.2031,
+      "step": 10489
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0486145671144637e-05,
+      "loss": 7.1406,
+      "step": 10490
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0484333542642292e-05,
+      "loss": 4.0,
+      "step": 10491
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0482521295445244e-05,
+      "loss": 1.6562,
+      "step": 10492
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0480708929574007e-05,
+      "loss": 4.2188,
+      "step": 10493
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0478896445049098e-05,
+      "loss": 3.4062,
+      "step": 10494
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.047708384189104e-05,
+      "loss": 3.25,
+      "step": 10495
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0475271120120355e-05,
+      "loss": 3.2656,
+      "step": 10496
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.047345827975756e-05,
+      "loss": 2.9375,
+      "step": 10497
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0471645320823183e-05,
+      "loss": 5.8281,
+      "step": 10498
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0469832243337744e-05,
+      "loss": 3.7188,
+      "step": 10499
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0468019047321778e-05,
+      "loss": 1.7109,
+      "step": 10500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0466205732795803e-05,
+      "loss": 2.625,
+      "step": 10501
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0464392299780353e-05,
+      "loss": 3.3125,
+      "step": 10502
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.046257874829595e-05,
+      "loss": 2.8438,
+      "step": 10503
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.046076507836314e-05,
+      "loss": 2.3438,
+      "step": 10504
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0458951290002445e-05,
+      "loss": 3.4219,
+      "step": 10505
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0457137383234397e-05,
+      "loss": 2.8125,
+      "step": 10506
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.045532335807954e-05,
+      "loss": 3.2812,
+      "step": 10507
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0453509214558406e-05,
+      "loss": 4.5625,
+      "step": 10508
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.045169495269154e-05,
+      "loss": 3.0312,
+      "step": 10509
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0449880572499472e-05,
+      "loss": 4.0938,
+      "step": 10510
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0448066074002746e-05,
+      "loss": 1.7031,
+      "step": 10511
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.044625145722191e-05,
+      "loss": 3.4688,
+      "step": 10512
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0444436722177493e-05,
+      "loss": 4.7969,
+      "step": 10513
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0442621868890056e-05,
+      "loss": 4.5938,
+      "step": 10514
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0440806897380136e-05,
+      "loss": 2.3125,
+      "step": 10515
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.043899180766829e-05,
+      "loss": 4.9375,
+      "step": 10516
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.0437176599775056e-05,
+      "loss": 5.4062,
+      "step": 10517
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0435361273720984e-05,
+      "loss": 3.7344,
+      "step": 10518
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0433545829526632e-05,
+      "loss": 2.2031,
+      "step": 10519
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0431730267212556e-05,
+      "loss": 3.0625,
+      "step": 10520
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0429914586799306e-05,
+      "loss": 3.1562,
+      "step": 10521
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0428098788307427e-05,
+      "loss": 2.4219,
+      "step": 10522
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0426282871757495e-05,
+      "loss": 2.1562,
+      "step": 10523
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0424466837170055e-05,
+      "loss": 3.7969,
+      "step": 10524
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0422650684565674e-05,
+      "loss": 7.1094,
+      "step": 10525
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.04208344139649e-05,
+      "loss": 5.0625,
+      "step": 10526
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0419018025388312e-05,
+      "loss": 6.3594,
+      "step": 10527
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0417201518856466e-05,
+      "loss": 5.2188,
+      "step": 10528
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.041538489438993e-05,
+      "loss": 2.7812,
+      "step": 10529
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.041356815200926e-05,
+      "loss": 5.0469,
+      "step": 10530
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0411751291735035e-05,
+      "loss": 3.4375,
+      "step": 10531
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.040993431358782e-05,
+      "loss": 5.3594,
+      "step": 10532
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.040811721758819e-05,
+      "loss": 4.625,
+      "step": 10533
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.040630000375671e-05,
+      "loss": 3.9062,
+      "step": 10534
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.040448267211395e-05,
+      "loss": 2.125,
+      "step": 10535
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0402665222680493e-05,
+      "loss": 2.5,
+      "step": 10536
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0400847655476914e-05,
+      "loss": 2.9688,
+      "step": 10537
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.039902997052378e-05,
+      "loss": 4.3906,
+      "step": 10538
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.039721216784168e-05,
+      "loss": 2.4844,
+      "step": 10539
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.039539424745119e-05,
+      "loss": 1.9375,
+      "step": 10540
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.039357620937289e-05,
+      "loss": 2.5781,
+      "step": 10541
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0391758053627365e-05,
+      "loss": 4.5469,
+      "step": 10542
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.038993978023519e-05,
+      "loss": 4.1562,
+      "step": 10543
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.038812138921696e-05,
+      "loss": 1.0156,
+      "step": 10544
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0386302880593263e-05,
+      "loss": 2.5312,
+      "step": 10545
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.038448425438467e-05,
+      "loss": 3.1094,
+      "step": 10546
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.0382665510611793e-05,
+      "loss": 2.9375,
+      "step": 10547
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0380846649295204e-05,
+      "loss": 2.5938,
+      "step": 10548
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0379027670455504e-05,
+      "loss": 6.1094,
+      "step": 10549
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0377208574113287e-05,
+      "loss": 5.25,
+      "step": 10550
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0375389360289137e-05,
+      "loss": 3.375,
+      "step": 10551
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0373570029003656e-05,
+      "loss": 5.2344,
+      "step": 10552
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.037175058027744e-05,
+      "loss": 3.4375,
+      "step": 10553
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.036993101413109e-05,
+      "loss": 2.5625,
+      "step": 10554
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0368111330585207e-05,
+      "loss": 4.6562,
+      "step": 10555
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0366291529660388e-05,
+      "loss": 2.3125,
+      "step": 10556
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.036447161137723e-05,
+      "loss": 2.6094,
+      "step": 10557
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0362651575756347e-05,
+      "loss": 3.4062,
+      "step": 10558
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0360831422818343e-05,
+      "loss": 5.7969,
+      "step": 10559
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0359011152583816e-05,
+      "loss": 1.7812,
+      "step": 10560
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.035719076507338e-05,
+      "loss": 3.25,
+      "step": 10561
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0355370260307643e-05,
+      "loss": 2.625,
+      "step": 10562
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.03535496383072e-05,
+      "loss": 2.5,
+      "step": 10563
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0351728899092693e-05,
+      "loss": 4.3438,
+      "step": 10564
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.034990804268472e-05,
+      "loss": 2.6719,
+      "step": 10565
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.034808706910388e-05,
+      "loss": 2.5625,
+      "step": 10566
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0346265978370815e-05,
+      "loss": 3.4062,
+      "step": 10567
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0344444770506123e-05,
+      "loss": 4.0156,
+      "step": 10568
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0342623445530432e-05,
+      "loss": 3.9062,
+      "step": 10569
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0340802003464353e-05,
+      "loss": 1.6562,
+      "step": 10570
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.033898044432851e-05,
+      "loss": 2.9688,
+      "step": 10571
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0337158768143538e-05,
+      "loss": 3.6875,
+      "step": 10572
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.033533697493004e-05,
+      "loss": 2.0,
+      "step": 10573
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0333515064708653e-05,
+      "loss": 5.0312,
+      "step": 10574
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0331693037499996e-05,
+      "loss": 3.5625,
+      "step": 10575
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0329870893324704e-05,
+      "loss": 3.2812,
+      "step": 10576
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.0328048632203402e-05,
+      "loss": 4.0,
+      "step": 10577
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0326226254156724e-05,
+      "loss": 2.4688,
+      "step": 10578
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.032440375920529e-05,
+      "loss": 3.0,
+      "step": 10579
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0322581147369745e-05,
+      "loss": 1.7188,
+      "step": 10580
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.032075841867072e-05,
+      "loss": 2.8281,
+      "step": 10581
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0318935573128853e-05,
+      "loss": 3.0469,
+      "step": 10582
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0317112610764766e-05,
+      "loss": 2.8125,
+      "step": 10583
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0315289531599112e-05,
+      "loss": 2.7188,
+      "step": 10584
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.031346633565253e-05,
+      "loss": 5.0156,
+      "step": 10585
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0311643022945652e-05,
+      "loss": 3.6406,
+      "step": 10586
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0309819593499124e-05,
+      "loss": 3.125,
+      "step": 10587
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0307996047333598e-05,
+      "loss": 4.2188,
+      "step": 10588
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0306172384469703e-05,
+      "loss": 3.4375,
+      "step": 10589
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0304348604928093e-05,
+      "loss": 2.875,
+      "step": 10590
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0302524708729427e-05,
+      "loss": 4.5781,
+      "step": 10591
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0300700695894324e-05,
+      "loss": 3.25,
+      "step": 10592
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.029887656644346e-05,
+      "loss": 4.7344,
+      "step": 10593
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.029705232039748e-05,
+      "loss": 3.75,
+      "step": 10594
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0295227957777028e-05,
+      "loss": 4.5,
+      "step": 10595
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0293403478602772e-05,
+      "loss": 3.5938,
+      "step": 10596
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0291578882895353e-05,
+      "loss": 3.9219,
+      "step": 10597
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0289754170675435e-05,
+      "loss": 3.375,
+      "step": 10598
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.028792934196368e-05,
+      "loss": 4.1875,
+      "step": 10599
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0286104396780735e-05,
+      "loss": 3.25,
+      "step": 10600
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0284279335147268e-05,
+      "loss": 2.1406,
+      "step": 10601
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.028245415708394e-05,
+      "loss": 3.9062,
+      "step": 10602
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0280628862611417e-05,
+      "loss": 8.1094,
+      "step": 10603
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.027880345175036e-05,
+      "loss": 3.0,
+      "step": 10604
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.027697792452144e-05,
+      "loss": 3.1562,
+      "step": 10605
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0275152280945314e-05,
+      "loss": 5.9062,
+      "step": 10606
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0273326521042663e-05,
+      "loss": 7.25,
+      "step": 10607
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.0271500644834138e-05,
+      "loss": 4.5312,
+      "step": 10608
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0269674652340427e-05,
+      "loss": 3.9062,
+      "step": 10609
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.02678485435822e-05,
+      "loss": 5.0625,
+      "step": 10610
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.026602231858012e-05,
+      "loss": 3.7031,
+      "step": 10611
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.026419597735488e-05,
+      "loss": 3.6875,
+      "step": 10612
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.026236951992714e-05,
+      "loss": 1.8125,
+      "step": 10613
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0260542946317583e-05,
+      "loss": 3.6719,
+      "step": 10614
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0258716256546892e-05,
+      "loss": 3.9688,
+      "step": 10615
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.025688945063574e-05,
+      "loss": 2.375,
+      "step": 10616
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0255062528604813e-05,
+      "loss": 1.1562,
+      "step": 10617
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.025323549047479e-05,
+      "loss": 6.0312,
+      "step": 10618
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.025140833626637e-05,
+      "loss": 2.875,
+      "step": 10619
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.024958106600021e-05,
+      "loss": 2.7031,
+      "step": 10620
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.024775367969702e-05,
+      "loss": 3.5938,
+      "step": 10621
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0245926177377484e-05,
+      "loss": 2.5312,
+      "step": 10622
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0244098559062286e-05,
+      "loss": 1.8125,
+      "step": 10623
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0242270824772127e-05,
+      "loss": 7.1719,
+      "step": 10624
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0240442974527683e-05,
+      "loss": 3.9375,
+      "step": 10625
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0238615008349664e-05,
+      "loss": 4.1562,
+      "step": 10626
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0236786926258754e-05,
+      "loss": 3.5781,
+      "step": 10627
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0234958728275648e-05,
+      "loss": 1.7812,
+      "step": 10628
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.023313041442105e-05,
+      "loss": 3.7188,
+      "step": 10629
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.023130198471565e-05,
+      "loss": 2.8438,
+      "step": 10630
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0229473439180158e-05,
+      "loss": 4.6875,
+      "step": 10631
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.022764477783528e-05,
+      "loss": 4.4375,
+      "step": 10632
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0225816000701694e-05,
+      "loss": 3.4219,
+      "step": 10633
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0223987107800127e-05,
+      "loss": 2.7344,
+      "step": 10634
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0222158099151283e-05,
+      "loss": 3.5,
+      "step": 10635
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.0220328974775852e-05,
+      "loss": 5.8125,
+      "step": 10636
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.021849973469455e-05,
+      "loss": 2.2188,
+      "step": 10637
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.02166703789281e-05,
+      "loss": 1.6875,
+      "step": 10638
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0214840907497188e-05,
+      "loss": 3.4062,
+      "step": 10639
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0213011320422546e-05,
+      "loss": 2.6875,
+      "step": 10640
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0211181617724877e-05,
+      "loss": 3.7812,
+      "step": 10641
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0209351799424893e-05,
+      "loss": 2.3125,
+      "step": 10642
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.020752186554332e-05,
+      "loss": 2.9688,
+      "step": 10643
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0205691816100862e-05,
+      "loss": 1.75,
+      "step": 10644
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0203861651118252e-05,
+      "loss": 3.625,
+      "step": 10645
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0202031370616194e-05,
+      "loss": 2.0156,
+      "step": 10646
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0200200974615417e-05,
+      "loss": 4.4062,
+      "step": 10647
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0198370463136653e-05,
+      "loss": 4.25,
+      "step": 10648
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0196539836200607e-05,
+      "loss": 3.4062,
+      "step": 10649
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0194709093828013e-05,
+      "loss": 3.0,
+      "step": 10650
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.01928782360396e-05,
+      "loss": 2.75,
+      "step": 10651
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0191047262856083e-05,
+      "loss": 4.2031,
+      "step": 10652
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.018921617429821e-05,
+      "loss": 3.2031,
+      "step": 10653
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.01873849703867e-05,
+      "loss": 1.625,
+      "step": 10654
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0185553651142273e-05,
+      "loss": 4.2812,
+      "step": 10655
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0183722216585685e-05,
+      "loss": 4.2812,
+      "step": 10656
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.018189066673765e-05,
+      "loss": 2.5938,
+      "step": 10657
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.018005900161892e-05,
+      "loss": 2.9062,
+      "step": 10658
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.017822722125022e-05,
+      "loss": 5.5781,
+      "step": 10659
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0176395325652287e-05,
+      "loss": 3.2188,
+      "step": 10660
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.017456331484587e-05,
+      "loss": 3.0312,
+      "step": 10661
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.01727311888517e-05,
+      "loss": 3.0938,
+      "step": 10662
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0170898947690518e-05,
+      "loss": 2.9688,
+      "step": 10663
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.016906659138308e-05,
+      "loss": 4.375,
+      "step": 10664
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0167234119950116e-05,
+      "loss": 3.1562,
+      "step": 10665
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0165401533412385e-05,
+      "loss": 4.4688,
+      "step": 10666
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0163568831790616e-05,
+      "loss": 2.7344,
+      "step": 10667
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0161736015105574e-05,
+      "loss": 2.2812,
+      "step": 10668
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.0159903083378004e-05,
+      "loss": 3.0,
+      "step": 10669
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0158070036628655e-05,
+      "loss": 3.2344,
+      "step": 10670
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0156236874878274e-05,
+      "loss": 3.6094,
+      "step": 10671
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0154403598147622e-05,
+      "loss": 2.5625,
+      "step": 10672
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.015257020645746e-05,
+      "loss": 3.9062,
+      "step": 10673
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0150736699828526e-05,
+      "loss": 3.4375,
+      "step": 10674
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.014890307828159e-05,
+      "loss": 4.5625,
+      "step": 10675
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.014706934183741e-05,
+      "loss": 2.6719,
+      "step": 10676
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.014523549051674e-05,
+      "loss": 2.5938,
+      "step": 10677
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0143401524340343e-05,
+      "loss": 3.0938,
+      "step": 10678
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.014156744332899e-05,
+      "loss": 3.9375,
+      "step": 10679
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0139733247503436e-05,
+      "loss": 4.3125,
+      "step": 10680
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0137898936884453e-05,
+      "loss": 5.3281,
+      "step": 10681
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.01360645114928e-05,
+      "loss": 5.2969,
+      "step": 10682
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0134229971349255e-05,
+      "loss": 3.9062,
+      "step": 10683
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0132395316474575e-05,
+      "loss": 4.375,
+      "step": 10684
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.013056054688953e-05,
+      "loss": 2.7812,
+      "step": 10685
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.01287256626149e-05,
+      "loss": 5.375,
+      "step": 10686
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0126890663671457e-05,
+      "loss": 2.9531,
+      "step": 10687
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0125055550079975e-05,
+      "loss": 3.2656,
+      "step": 10688
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0123220321861235e-05,
+      "loss": 2.8438,
+      "step": 10689
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0121384979035997e-05,
+      "loss": 2.5,
+      "step": 10690
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0119549521625057e-05,
+      "loss": 4.0469,
+      "step": 10691
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0117713949649183e-05,
+      "loss": 4.5469,
+      "step": 10692
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0115878263129156e-05,
+      "loss": 1.9688,
+      "step": 10693
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0114042462085765e-05,
+      "loss": 5.4688,
+      "step": 10694
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0112206546539796e-05,
+      "loss": 2.0938,
+      "step": 10695
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0110370516512018e-05,
+      "loss": 2.7031,
+      "step": 10696
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0108534372023235e-05,
+      "loss": 3.75,
+      "step": 10697
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0106698113094222e-05,
+      "loss": 3.6562,
+      "step": 10698
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.0104861739745776e-05,
+      "loss": 3.3438,
+      "step": 10699
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.010302525199868e-05,
+      "loss": 3.625,
+      "step": 10700
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.010118864987372e-05,
+      "loss": 2.1094,
+      "step": 10701
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.00993519333917e-05,
+      "loss": 2.9375,
+      "step": 10702
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.009751510257342e-05,
+      "loss": 5.3125,
+      "step": 10703
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0095678157439658e-05,
+      "loss": 3.2031,
+      "step": 10704
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.009384109801121e-05,
+      "loss": 3.6562,
+      "step": 10705
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0092003924308895e-05,
+      "loss": 3.25,
+      "step": 10706
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.009016663635349e-05,
+      "loss": 2.5312,
+      "step": 10707
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0088329234165806e-05,
+      "loss": 3.0312,
+      "step": 10708
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0086491717766635e-05,
+      "loss": 3.3594,
+      "step": 10709
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.008465408717679e-05,
+      "loss": 3.2188,
+      "step": 10710
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0082816342417077e-05,
+      "loss": 5.3438,
+      "step": 10711
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0080978483508282e-05,
+      "loss": 3.4375,
+      "step": 10712
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0079140510471237e-05,
+      "loss": 2.0625,
+      "step": 10713
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0077302423326738e-05,
+      "loss": 3.375,
+      "step": 10714
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0075464222095595e-05,
+      "loss": 3.2812,
+      "step": 10715
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0073625906798614e-05,
+      "loss": 4.5938,
+      "step": 10716
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.007178747745661e-05,
+      "loss": 3.0312,
+      "step": 10717
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.00699489340904e-05,
+      "loss": 3.0938,
+      "step": 10718
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.00681102767208e-05,
+      "loss": 3.1562,
+      "step": 10719
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0066271505368613e-05,
+      "loss": 4.2656,
+      "step": 10720
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0064432620054667e-05,
+      "loss": 3.3438,
+      "step": 10721
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0062593620799778e-05,
+      "loss": 1.625,
+      "step": 10722
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.006075450762477e-05,
+      "loss": 4.375,
+      "step": 10723
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0058915280550454e-05,
+      "loss": 4.5938,
+      "step": 10724
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.005707593959766e-05,
+      "loss": 4.5781,
+      "step": 10725
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.005523648478721e-05,
+      "loss": 2.75,
+      "step": 10726
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0053396916139924e-05,
+      "loss": 2.8594,
+      "step": 10727
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0051557233676628e-05,
+      "loss": 3.1719,
+      "step": 10728
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0049717437418152e-05,
+      "loss": 2.875,
+      "step": 10729
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.0047877527385333e-05,
+      "loss": 2.4375,
+      "step": 10730
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.004603750359899e-05,
+      "loss": 4.7188,
+      "step": 10731
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0044197366079967e-05,
+      "loss": 3.2031,
+      "step": 10732
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0042357114849073e-05,
+      "loss": 4.1719,
+      "step": 10733
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0040516749927157e-05,
+      "loss": 3.5469,
+      "step": 10734
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.003867627133506e-05,
+      "loss": 4.6094,
+      "step": 10735
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0036835679093604e-05,
+      "loss": 2.5938,
+      "step": 10736
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.003499497322364e-05,
+      "loss": 3.2188,
+      "step": 10737
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0033154153745997e-05,
+      "loss": 2.1875,
+      "step": 10738
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0031313220681526e-05,
+      "loss": 2.625,
+      "step": 10739
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0029472174051055e-05,
+      "loss": 2.6562,
+      "step": 10740
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0027631013875428e-05,
+      "loss": 3.5625,
+      "step": 10741
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0025789740175502e-05,
+      "loss": 5.4062,
+      "step": 10742
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.002394835297211e-05,
+      "loss": 2.6562,
+      "step": 10743
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0022106852286105e-05,
+      "loss": 4.2812,
+      "step": 10744
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0020265238138335e-05,
+      "loss": 1.6406,
+      "step": 10745
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0018423510549644e-05,
+      "loss": 2.9688,
+      "step": 10746
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.001658166954089e-05,
+      "loss": 2.75,
+      "step": 10747
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.001473971513291e-05,
+      "loss": 2.6562,
+      "step": 10748
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.001289764734658e-05,
+      "loss": 2.8594,
+      "step": 10749
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0011055466202738e-05,
+      "loss": 4.25,
+      "step": 10750
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0009213171722238e-05,
+      "loss": 3.1094,
+      "step": 10751
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.000737076392595e-05,
+      "loss": 2.9531,
+      "step": 10752
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.000552824283472e-05,
+      "loss": 4.5938,
+      "step": 10753
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0003685608469405e-05,
+      "loss": 3.6562,
+      "step": 10754
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0001842860850883e-05,
+      "loss": 4.0156,
+      "step": 10755
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.0000000000000004e-05,
+      "loss": 4.1562,
+      "step": 10756
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.9998157025937632e-05,
+      "loss": 2.2812,
+      "step": 10757
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.9996313938684628e-05,
+      "loss": 3.3125,
+      "step": 10758
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.9994470738261872e-05,
+      "loss": 2.6875,
+      "step": 10759
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.9992627424690216e-05,
+      "loss": 3.9375,
+      "step": 10760
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9990783997990527e-05,
+      "loss": 3.125,
+      "step": 10761
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9988940458183692e-05,
+      "loss": 4.75,
+      "step": 10762
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9987096805290566e-05,
+      "loss": 2.2344,
+      "step": 10763
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9985253039332033e-05,
+      "loss": 7.9219,
+      "step": 10764
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9983409160328957e-05,
+      "loss": 1.6875,
+      "step": 10765
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9981565168302213e-05,
+      "loss": 3.6875,
+      "step": 10766
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9979721063272685e-05,
+      "loss": 2.125,
+      "step": 10767
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9977876845261246e-05,
+      "loss": 4.3438,
+      "step": 10768
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.997603251428877e-05,
+      "loss": 3.25,
+      "step": 10769
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9974188070376145e-05,
+      "loss": 3.6094,
+      "step": 10770
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9972343513544247e-05,
+      "loss": 1.9375,
+      "step": 10771
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9970498843813963e-05,
+      "loss": 2.7812,
+      "step": 10772
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.996865406120617e-05,
+      "loss": 2.6094,
+      "step": 10773
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.996680916574176e-05,
+      "loss": 3.3281,
+      "step": 10774
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9964964157441616e-05,
+      "loss": 2.5312,
+      "step": 10775
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9963119036326627e-05,
+      "loss": 2.7656,
+      "step": 10776
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9961273802417676e-05,
+      "loss": 4.0156,
+      "step": 10777
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.995942845573566e-05,
+      "loss": 3.9219,
+      "step": 10778
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.995758299630147e-05,
+      "loss": 6.5,
+      "step": 10779
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9955737424135997e-05,
+      "loss": 4.5156,
+      "step": 10780
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9953891739260138e-05,
+      "loss": 3.0938,
+      "step": 10781
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9952045941694775e-05,
+      "loss": 1.6562,
+      "step": 10782
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9950200031460826e-05,
+      "loss": 4.125,
+      "step": 10783
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9948354008579173e-05,
+      "loss": 2.2812,
+      "step": 10784
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9946507873070724e-05,
+      "loss": 6.1406,
+      "step": 10785
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9944661624956367e-05,
+      "loss": 4.7812,
+      "step": 10786
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9942815264257022e-05,
+      "loss": 1.5312,
+      "step": 10787
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.994096879099358e-05,
+      "loss": 6.1719,
+      "step": 10788
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.993912220518694e-05,
+      "loss": 6.7656,
+      "step": 10789
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.9937275506858018e-05,
+      "loss": 4.4375,
+      "step": 10790
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.993542869602772e-05,
+      "loss": 4.2188,
+      "step": 10791
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9933581772716952e-05,
+      "loss": 5.5156,
+      "step": 10792
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9931734736946616e-05,
+      "loss": 3.8281,
+      "step": 10793
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.992988758873763e-05,
+      "loss": 3.2812,
+      "step": 10794
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9928040328110916e-05,
+      "loss": 1.25,
+      "step": 10795
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9926192955087365e-05,
+      "loss": 1.9688,
+      "step": 10796
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9924345469687908e-05,
+      "loss": 2.0625,
+      "step": 10797
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9922497871933448e-05,
+      "loss": 2.875,
+      "step": 10798
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9920650161844916e-05,
+      "loss": 3.5781,
+      "step": 10799
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.991880233944322e-05,
+      "loss": 2.0,
+      "step": 10800
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9916954404749282e-05,
+      "loss": 2.0312,
+      "step": 10801
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.991510635778403e-05,
+      "loss": 2.5938,
+      "step": 10802
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9913258198568374e-05,
+      "loss": 3.0,
+      "step": 10803
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9911409927123247e-05,
+      "loss": 2.4844,
+      "step": 10804
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9909561543469565e-05,
+      "loss": 1.9531,
+      "step": 10805
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9907713047628263e-05,
+      "loss": 3.125,
+      "step": 10806
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9905864439620257e-05,
+      "loss": 4.4375,
+      "step": 10807
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9904015719466483e-05,
+      "loss": 4.0938,
+      "step": 10808
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.990216688718787e-05,
+      "loss": 2.4375,
+      "step": 10809
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.990031794280535e-05,
+      "loss": 2.6562,
+      "step": 10810
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9898468886339854e-05,
+      "loss": 5.2031,
+      "step": 10811
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9896619717812308e-05,
+      "loss": 2.9062,
+      "step": 10812
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9894770437243664e-05,
+      "loss": 3.5,
+      "step": 10813
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9892921044654842e-05,
+      "loss": 3.2188,
+      "step": 10814
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9891071540066784e-05,
+      "loss": 5.4375,
+      "step": 10815
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9889221923500435e-05,
+      "loss": 4.6562,
+      "step": 10816
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9887372194976723e-05,
+      "loss": 2.9062,
+      "step": 10817
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9885522354516598e-05,
+      "loss": 5.1875,
+      "step": 10818
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9883672402140997e-05,
+      "loss": 7.0312,
+      "step": 10819
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.988182233787087e-05,
+      "loss": 4.0938,
+      "step": 10820
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.9879972161727157e-05,
+      "loss": 2.8438,
+      "step": 10821
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9878121873730797e-05,
+      "loss": 3.0781,
+      "step": 10822
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.987627147390275e-05,
+      "loss": 1.75,
+      "step": 10823
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9874420962263965e-05,
+      "loss": 4.9219,
+      "step": 10824
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9872570338835385e-05,
+      "loss": 2.2188,
+      "step": 10825
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9870719603637956e-05,
+      "loss": 1.25,
+      "step": 10826
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9868868756692645e-05,
+      "loss": 3.1562,
+      "step": 10827
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9867017798020403e-05,
+      "loss": 2.1406,
+      "step": 10828
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.986516672764217e-05,
+      "loss": 2.8906,
+      "step": 10829
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9863315545578916e-05,
+      "loss": 2.3125,
+      "step": 10830
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9861464251851596e-05,
+      "loss": 2.5312,
+      "step": 10831
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9859612846481164e-05,
+      "loss": 3.4062,
+      "step": 10832
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9857761329488594e-05,
+      "loss": 4.0312,
+      "step": 10833
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9855909700894825e-05,
+      "loss": 5.3906,
+      "step": 10834
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9854057960720837e-05,
+      "loss": 3.2188,
+      "step": 10835
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.985220610898759e-05,
+      "loss": 2.4375,
+      "step": 10836
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9850354145716046e-05,
+      "loss": 5.7344,
+      "step": 10837
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9848502070927177e-05,
+      "loss": 3.1562,
+      "step": 10838
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.984664988464194e-05,
+      "loss": 2.75,
+      "step": 10839
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9844797586881317e-05,
+      "loss": 3.0,
+      "step": 10840
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9842945177666263e-05,
+      "loss": 2.2656,
+      "step": 10841
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9841092657017764e-05,
+      "loss": 4.25,
+      "step": 10842
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9839240024956786e-05,
+      "loss": 3.3125,
+      "step": 10843
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9837387281504303e-05,
+      "loss": 3.125,
+      "step": 10844
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9835534426681286e-05,
+      "loss": 3.5469,
+      "step": 10845
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.983368146050873e-05,
+      "loss": 1.6875,
+      "step": 10846
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9831828383007585e-05,
+      "loss": 4.3125,
+      "step": 10847
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9829975194198853e-05,
+      "loss": 3.4062,
+      "step": 10848
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.98281218941035e-05,
+      "loss": 3.0938,
+      "step": 10849
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9826268482742513e-05,
+      "loss": 3.7969,
+      "step": 10850
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.9824414960136874e-05,
+      "loss": 2.8438,
+      "step": 10851
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.982256132630757e-05,
+      "loss": 10.1875,
+      "step": 10852
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.982070758127558e-05,
+      "loss": 4.2812,
+      "step": 10853
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9818853725061902e-05,
+      "loss": 3.0625,
+      "step": 10854
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.981699975768751e-05,
+      "loss": 3.9375,
+      "step": 10855
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9815145679173397e-05,
+      "loss": 3.125,
+      "step": 10856
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.981329148954056e-05,
+      "loss": 1.0781,
+      "step": 10857
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9811437188809978e-05,
+      "loss": 3.7656,
+      "step": 10858
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.980958277700266e-05,
+      "loss": 2.0469,
+      "step": 10859
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.980772825413959e-05,
+      "loss": 1.4219,
+      "step": 10860
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.980587362024176e-05,
+      "loss": 4.625,
+      "step": 10861
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9804018875330184e-05,
+      "loss": 1.125,
+      "step": 10862
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9802164019425836e-05,
+      "loss": 3.625,
+      "step": 10863
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9800309052549732e-05,
+      "loss": 1.6562,
+      "step": 10864
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9798453974722867e-05,
+      "loss": 4.6719,
+      "step": 10865
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9796598785966235e-05,
+      "loss": 2.2812,
+      "step": 10866
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.979474348630085e-05,
+      "loss": 2.7031,
+      "step": 10867
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9792888075747718e-05,
+      "loss": 1.8438,
+      "step": 10868
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9791032554327835e-05,
+      "loss": 2.75,
+      "step": 10869
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9789176922062215e-05,
+      "loss": 3.7812,
+      "step": 10870
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.978732117897186e-05,
+      "loss": 3.7188,
+      "step": 10871
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.978546532507778e-05,
+      "loss": 3.1875,
+      "step": 10872
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9783609360400988e-05,
+      "loss": 5.0312,
+      "step": 10873
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9781753284962495e-05,
+      "loss": 3.4062,
+      "step": 10874
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.977989709878331e-05,
+      "loss": 3.375,
+      "step": 10875
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9778040801884453e-05,
+      "loss": 5.5156,
+      "step": 10876
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9776184394286936e-05,
+      "loss": 3.8438,
+      "step": 10877
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9774327876011785e-05,
+      "loss": 2.8906,
+      "step": 10878
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.977247124708001e-05,
+      "loss": 5.3594,
+      "step": 10879
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9770614507512615e-05,
+      "loss": 2.875,
+      "step": 10880
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9768757657330643e-05,
+      "loss": 3.75,
+      "step": 10881
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.9766900696555113e-05,
+      "loss": 2.1875,
+      "step": 10882
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9765043625207038e-05,
+      "loss": 4.0781,
+      "step": 10883
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9763186443307444e-05,
+      "loss": 2.3438,
+      "step": 10884
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9761329150877366e-05,
+      "loss": 3.375,
+      "step": 10885
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.975947174793782e-05,
+      "loss": 4.3438,
+      "step": 10886
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9757614234509846e-05,
+      "loss": 3.125,
+      "step": 10887
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9755756610614458e-05,
+      "loss": 2.6094,
+      "step": 10888
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9753898876272695e-05,
+      "loss": 1.9219,
+      "step": 10889
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9752041031505588e-05,
+      "loss": 2.6562,
+      "step": 10890
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9750183076334167e-05,
+      "loss": 5.9219,
+      "step": 10891
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.974832501077947e-05,
+      "loss": 3.8281,
+      "step": 10892
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9746466834862535e-05,
+      "loss": 3.6406,
+      "step": 10893
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9744608548604387e-05,
+      "loss": 2.0156,
+      "step": 10894
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9742750152026082e-05,
+      "loss": 3.875,
+      "step": 10895
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.974089164514864e-05,
+      "loss": 3.6719,
+      "step": 10896
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9739033027993115e-05,
+      "loss": 3.6562,
+      "step": 10897
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.973717430058054e-05,
+      "loss": 4.0156,
+      "step": 10898
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9735315462931963e-05,
+      "loss": 4.7969,
+      "step": 10899
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9733456515068432e-05,
+      "loss": 5.3438,
+      "step": 10900
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9731597457010982e-05,
+      "loss": 3.0312,
+      "step": 10901
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9729738288780665e-05,
+      "loss": 3.2656,
+      "step": 10902
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.972787901039854e-05,
+      "loss": 5.1406,
+      "step": 10903
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.972601962188563e-05,
+      "loss": 7.4219,
+      "step": 10904
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.972416012326301e-05,
+      "loss": 2.8281,
+      "step": 10905
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9722300514551717e-05,
+      "loss": 2.2969,
+      "step": 10906
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.972044079577281e-05,
+      "loss": 3.1562,
+      "step": 10907
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9718580966947343e-05,
+      "loss": 3.7812,
+      "step": 10908
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9716721028096375e-05,
+      "loss": 3.4375,
+      "step": 10909
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9714860979240953e-05,
+      "loss": 5.9375,
+      "step": 10910
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.971300082040215e-05,
+      "loss": 2.875,
+      "step": 10911
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.9711140551601005e-05,
+      "loss": 3.5625,
+      "step": 10912
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.970928017285859e-05,
+      "loss": 5.1562,
+      "step": 10913
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.970741968419597e-05,
+      "loss": 3.5469,
+      "step": 10914
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9705559085634198e-05,
+      "loss": 2.6406,
+      "step": 10915
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.970369837719435e-05,
+      "loss": 6.3438,
+      "step": 10916
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.970183755889748e-05,
+      "loss": 4.5938,
+      "step": 10917
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9699976630764658e-05,
+      "loss": 3.8125,
+      "step": 10918
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9698115592816957e-05,
+      "loss": 3.1875,
+      "step": 10919
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.969625444507544e-05,
+      "loss": 4.3125,
+      "step": 10920
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.969439318756118e-05,
+      "loss": 2.8125,
+      "step": 10921
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9692531820295253e-05,
+      "loss": 3.7344,
+      "step": 10922
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.969067034329872e-05,
+      "loss": 2.6875,
+      "step": 10923
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9688808756592666e-05,
+      "loss": 3.75,
+      "step": 10924
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9686947060198165e-05,
+      "loss": 4.1406,
+      "step": 10925
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9685085254136285e-05,
+      "loss": 2.8125,
+      "step": 10926
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.968322333842811e-05,
+      "loss": 4.75,
+      "step": 10927
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.968136131309473e-05,
+      "loss": 4.1562,
+      "step": 10928
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9679499178157202e-05,
+      "loss": 2.4688,
+      "step": 10929
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9677636933636625e-05,
+      "loss": 2.4375,
+      "step": 10930
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.967577457955407e-05,
+      "loss": 5.9688,
+      "step": 10931
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9673912115930635e-05,
+      "loss": 6.1094,
+      "step": 10932
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9672049542787397e-05,
+      "loss": 1.7188,
+      "step": 10933
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9670186860145435e-05,
+      "loss": 3.3125,
+      "step": 10934
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.966832406802585e-05,
+      "loss": 2.5625,
+      "step": 10935
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.966646116644973e-05,
+      "loss": 3.2188,
+      "step": 10936
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9664598155438154e-05,
+      "loss": 4.0,
+      "step": 10937
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9662735035012224e-05,
+      "loss": 2.3438,
+      "step": 10938
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9660871805193026e-05,
+      "loss": 3.5312,
+      "step": 10939
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9659008466001658e-05,
+      "loss": 3.25,
+      "step": 10940
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9657145017459217e-05,
+      "loss": 2.8125,
+      "step": 10941
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.965528145958679e-05,
+      "loss": 4.9219,
+      "step": 10942
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.9653417792405488e-05,
+      "loss": 2.6875,
+      "step": 10943
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.96515540159364e-05,
+      "loss": 3.2031,
+      "step": 10944
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.964969013020063e-05,
+      "loss": 2.6562,
+      "step": 10945
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9647826135219278e-05,
+      "loss": 3.8516,
+      "step": 10946
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9645962031013447e-05,
+      "loss": 2.4844,
+      "step": 10947
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9644097817604236e-05,
+      "loss": 3.0312,
+      "step": 10948
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.964223349501276e-05,
+      "loss": 4.375,
+      "step": 10949
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.964036906326012e-05,
+      "loss": 4.5,
+      "step": 10950
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9638504522367416e-05,
+      "loss": 3.75,
+      "step": 10951
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.963663987235577e-05,
+      "loss": 4.8281,
+      "step": 10952
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9634775113246287e-05,
+      "loss": 2.4062,
+      "step": 10953
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9632910245060075e-05,
+      "loss": 3.5,
+      "step": 10954
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.963104526781825e-05,
+      "loss": 2.5938,
+      "step": 10955
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9629180181541925e-05,
+      "loss": 2.2812,
+      "step": 10956
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9627314986252216e-05,
+      "loss": 3.1562,
+      "step": 10957
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9625449681970234e-05,
+      "loss": 2.4375,
+      "step": 10958
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9623584268717098e-05,
+      "loss": 3.9219,
+      "step": 10959
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9621718746513935e-05,
+      "loss": 2.75,
+      "step": 10960
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9619853115381855e-05,
+      "loss": 2.5312,
+      "step": 10961
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9617987375341984e-05,
+      "loss": 2.8125,
+      "step": 10962
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9616121526415446e-05,
+      "loss": 2.2344,
+      "step": 10963
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9614255568623354e-05,
+      "loss": 2.6562,
+      "step": 10964
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9612389501986844e-05,
+      "loss": 4.4062,
+      "step": 10965
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9610523326527037e-05,
+      "loss": 4.7344,
+      "step": 10966
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.960865704226506e-05,
+      "loss": 3.6406,
+      "step": 10967
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9606790649222045e-05,
+      "loss": 2.5,
+      "step": 10968
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9604924147419126e-05,
+      "loss": 4.3125,
+      "step": 10969
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9603057536877417e-05,
+      "loss": 2.6719,
+      "step": 10970
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.960119081761807e-05,
+      "loss": 2.0938,
+      "step": 10971
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9599323989662202e-05,
+      "loss": 3.3125,
+      "step": 10972
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.9597457053030956e-05,
+      "loss": 3.0312,
+      "step": 10973
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.959559000774547e-05,
+      "loss": 2.4062,
+      "step": 10974
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.959372285382687e-05,
+      "loss": 4.4844,
+      "step": 10975
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9591855591296313e-05,
+      "loss": 3.5938,
+      "step": 10976
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9589988220174922e-05,
+      "loss": 2.5312,
+      "step": 10977
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9588120740483846e-05,
+      "loss": 2.4688,
+      "step": 10978
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9586253152244224e-05,
+      "loss": 3.0,
+      "step": 10979
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9584385455477195e-05,
+      "loss": 4.5938,
+      "step": 10980
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9582517650203913e-05,
+      "loss": 2.0469,
+      "step": 10981
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9580649736445522e-05,
+      "loss": 1.9219,
+      "step": 10982
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.957878171422316e-05,
+      "loss": 4.6406,
+      "step": 10983
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.957691358355798e-05,
+      "loss": 5.2969,
+      "step": 10984
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9575045344471136e-05,
+      "loss": 4.3906,
+      "step": 10985
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9573176996983774e-05,
+      "loss": 2.8125,
+      "step": 10986
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.957130854111705e-05,
+      "loss": 2.7188,
+      "step": 10987
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9569439976892106e-05,
+      "loss": 5.625,
+      "step": 10988
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.956757130433011e-05,
+      "loss": 1.8438,
+      "step": 10989
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9565702523452212e-05,
+      "loss": 4.5469,
+      "step": 10990
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9563833634279562e-05,
+      "loss": 2.875,
+      "step": 10991
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.956196463683333e-05,
+      "loss": 0.5625,
+      "step": 10992
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.956009553113467e-05,
+      "loss": 2.875,
+      "step": 10993
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9558226317204744e-05,
+      "loss": 2.0312,
+      "step": 10994
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.955635699506471e-05,
+      "loss": 2.4062,
+      "step": 10995
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.955448756473573e-05,
+      "loss": 4.0312,
+      "step": 10996
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9552618026238975e-05,
+      "loss": 4.7031,
+      "step": 10997
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9550748379595606e-05,
+      "loss": 3.5312,
+      "step": 10998
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9548878624826777e-05,
+      "loss": 5.5938,
+      "step": 10999
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9547008761953684e-05,
+      "loss": 3.4375,
+      "step": 11000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9545138790997472e-05,
+      "loss": 2.7188,
+      "step": 11001
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.954326871197933e-05,
+      "loss": 5.0938,
+      "step": 11002
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9541398524920408e-05,
+      "loss": 3.6562,
+      "step": 11003
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.9539528229841896e-05,
+      "loss": 3.4062,
+      "step": 11004
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9537657826764952e-05,
+      "loss": 2.9375,
+      "step": 11005
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9535787315710772e-05,
+      "loss": 4.2188,
+      "step": 11006
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9533916696700516e-05,
+      "loss": 2.1875,
+      "step": 11007
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9532045969755364e-05,
+      "loss": 3.7812,
+      "step": 11008
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.95301751348965e-05,
+      "loss": 4.3906,
+      "step": 11009
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.95283041921451e-05,
+      "loss": 4.7188,
+      "step": 11010
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9526433141522352e-05,
+      "loss": 4.625,
+      "step": 11011
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9524561983049432e-05,
+      "loss": 3.4688,
+      "step": 11012
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9522690716747517e-05,
+      "loss": 8.4844,
+      "step": 11013
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.95208193426378e-05,
+      "loss": 8.6875,
+      "step": 11014
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9518947860741476e-05,
+      "loss": 6.0938,
+      "step": 11015
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9517076271079715e-05,
+      "loss": 5.3438,
+      "step": 11016
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9515204573673717e-05,
+      "loss": 6.8594,
+      "step": 11017
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9513332768544667e-05,
+      "loss": 4.0469,
+      "step": 11018
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9511460855713762e-05,
+      "loss": 6.1875,
+      "step": 11019
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9509588835202184e-05,
+      "loss": 4.375,
+      "step": 11020
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9507716707031133e-05,
+      "loss": 3.6562,
+      "step": 11021
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.95058444712218e-05,
+      "loss": 4.375,
+      "step": 11022
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.950397212779539e-05,
+      "loss": 5.8438,
+      "step": 11023
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9502099676773085e-05,
+      "loss": 3.5938,
+      "step": 11024
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9500227118176102e-05,
+      "loss": 3.9688,
+      "step": 11025
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9498354452025623e-05,
+      "loss": 4.7969,
+      "step": 11026
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9496481678342862e-05,
+      "loss": 3.4688,
+      "step": 11027
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9494608797149008e-05,
+      "loss": 3.2656,
+      "step": 11028
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9492735808465273e-05,
+      "loss": 4.6094,
+      "step": 11029
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.949086271231286e-05,
+      "loss": 4.1875,
+      "step": 11030
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9488989508712977e-05,
+      "loss": 4.6406,
+      "step": 11031
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9487116197686822e-05,
+      "loss": 3.9688,
+      "step": 11032
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9485242779255616e-05,
+      "loss": 5.0625,
+      "step": 11033
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.9483369253440557e-05,
+      "loss": 3.4844,
+      "step": 11034
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9481495620262863e-05,
+      "loss": 5.5938,
+      "step": 11035
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.947962187974374e-05,
+      "loss": 2.3281,
+      "step": 11036
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.94777480319044e-05,
+      "loss": 3.8594,
+      "step": 11037
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.947587407676607e-05,
+      "loss": 1.7969,
+      "step": 11038
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9474000014349947e-05,
+      "loss": 2.9375,
+      "step": 11039
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9472125844677254e-05,
+      "loss": 3.1406,
+      "step": 11040
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9470251567769218e-05,
+      "loss": 4.9531,
+      "step": 11041
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.946837718364705e-05,
+      "loss": 3.9062,
+      "step": 11042
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9466502692331973e-05,
+      "loss": 6.5625,
+      "step": 11043
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9464628093845205e-05,
+      "loss": 4.0938,
+      "step": 11044
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9462753388207966e-05,
+      "loss": 5.0781,
+      "step": 11045
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9460878575441486e-05,
+      "loss": 3.4688,
+      "step": 11046
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.945900365556699e-05,
+      "loss": 3.6875,
+      "step": 11047
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9457128628605698e-05,
+      "loss": 4.0156,
+      "step": 11048
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9455253494578845e-05,
+      "loss": 5.0938,
+      "step": 11049
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9453378253507662e-05,
+      "loss": 4.5625,
+      "step": 11050
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9451502905413365e-05,
+      "loss": 5.3125,
+      "step": 11051
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9449627450317196e-05,
+      "loss": 3.9062,
+      "step": 11052
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9447751888240384e-05,
+      "loss": 3.2188,
+      "step": 11053
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.944587621920416e-05,
+      "loss": 3.5,
+      "step": 11054
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.944400044322977e-05,
+      "loss": 2.8125,
+      "step": 11055
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9442124560338437e-05,
+      "loss": 3.125,
+      "step": 11056
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9440248570551406e-05,
+      "loss": 2.7656,
+      "step": 11057
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9438372473889905e-05,
+      "loss": 2.8125,
+      "step": 11058
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.943649627037519e-05,
+      "loss": 4.5312,
+      "step": 11059
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.943461996002849e-05,
+      "loss": 5.3438,
+      "step": 11060
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.943274354287105e-05,
+      "loss": 3.6875,
+      "step": 11061
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.943086701892411e-05,
+      "loss": 3.875,
+      "step": 11062
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9428990388208917e-05,
+      "loss": 5.4531,
+      "step": 11063
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.9427113650746716e-05,
+      "loss": 3.7812,
+      "step": 11064
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9425236806558765e-05,
+      "loss": 5.3438,
+      "step": 11065
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9423359855666295e-05,
+      "loss": 2.0,
+      "step": 11066
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9421482798090557e-05,
+      "loss": 5.4375,
+      "step": 11067
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9419605633852812e-05,
+      "loss": 3.2812,
+      "step": 11068
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.941772836297431e-05,
+      "loss": 4.0469,
+      "step": 11069
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.941585098547629e-05,
+      "loss": 6.0,
+      "step": 11070
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.941397350138002e-05,
+      "loss": 2.5781,
+      "step": 11071
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.941209591070676e-05,
+      "loss": 4.0625,
+      "step": 11072
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9410218213477745e-05,
+      "loss": 3.2969,
+      "step": 11073
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9408340409714252e-05,
+      "loss": 3.5,
+      "step": 11074
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.940646249943753e-05,
+      "loss": 1.3438,
+      "step": 11075
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9404584482668846e-05,
+      "loss": 3.4531,
+      "step": 11076
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.940270635942946e-05,
+      "loss": 2.2188,
+      "step": 11077
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9400828129740627e-05,
+      "loss": 2.7344,
+      "step": 11078
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9398949793623618e-05,
+      "loss": 2.9062,
+      "step": 11079
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.93970713510997e-05,
+      "loss": 2.7188,
+      "step": 11080
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.939519280219013e-05,
+      "loss": 4.8438,
+      "step": 11081
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9393314146916184e-05,
+      "loss": 5.8594,
+      "step": 11082
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9391435385299127e-05,
+      "loss": 2.5156,
+      "step": 11083
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9389556517360227e-05,
+      "loss": 4.2188,
+      "step": 11084
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9387677543120755e-05,
+      "loss": 2.3125,
+      "step": 11085
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9385798462601986e-05,
+      "loss": 3.875,
+      "step": 11086
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9383919275825194e-05,
+      "loss": 1.9844,
+      "step": 11087
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.938203998281165e-05,
+      "loss": 1.9062,
+      "step": 11088
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9380160583582625e-05,
+      "loss": 2.6562,
+      "step": 11089
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9378281078159414e-05,
+      "loss": 5.0469,
+      "step": 11090
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.937640146656328e-05,
+      "loss": 3.125,
+      "step": 11091
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.93745217488155e-05,
+      "loss": 3.3438,
+      "step": 11092
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9372641924937372e-05,
+      "loss": 5.125,
+      "step": 11093
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.9370761994950155e-05,
+      "loss": 3.0312,
+      "step": 11094
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.936888195887515e-05,
+      "loss": 4.1719,
+      "step": 11095
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9367001816733635e-05,
+      "loss": 3.3438,
+      "step": 11096
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.936512156854689e-05,
+      "loss": 4.375,
+      "step": 11097
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9363241214336208e-05,
+      "loss": 3.7812,
+      "step": 11098
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9361360754122876e-05,
+      "loss": 3.125,
+      "step": 11099
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.935948018792818e-05,
+      "loss": 1.6875,
+      "step": 11100
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.935759951577342e-05,
+      "loss": 2.6094,
+      "step": 11101
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.935571873767987e-05,
+      "loss": 1.7656,
+      "step": 11102
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9353837853668835e-05,
+      "loss": 3.5156,
+      "step": 11103
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9351956863761613e-05,
+      "loss": 2.8594,
+      "step": 11104
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.935007576797948e-05,
+      "loss": 9.7812,
+      "step": 11105
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.934819456634375e-05,
+      "loss": 4.1562,
+      "step": 11106
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9346313258875718e-05,
+      "loss": 6.9375,
+      "step": 11107
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9344431845596672e-05,
+      "loss": 2.5625,
+      "step": 11108
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.934255032652793e-05,
+      "loss": 4.8281,
+      "step": 11109
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.934066870169077e-05,
+      "loss": 4.125,
+      "step": 11110
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9338786971106512e-05,
+      "loss": 2.7344,
+      "step": 11111
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9336905134796455e-05,
+      "loss": 4.0312,
+      "step": 11112
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9335023192781896e-05,
+      "loss": 6.4219,
+      "step": 11113
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9333141145084148e-05,
+      "loss": 1.8438,
+      "step": 11114
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.933125899172452e-05,
+      "loss": 2.7188,
+      "step": 11115
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.932937673272431e-05,
+      "loss": 2.7188,
+      "step": 11116
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.932749436810485e-05,
+      "loss": 3.7031,
+      "step": 11117
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9325611897887416e-05,
+      "loss": 4.0,
+      "step": 11118
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.932372932209335e-05,
+      "loss": 3.8125,
+      "step": 11119
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9321846640743948e-05,
+      "loss": 4.4453,
+      "step": 11120
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9319963853860523e-05,
+      "loss": 1.9219,
+      "step": 11121
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9318080961464407e-05,
+      "loss": 1.9688,
+      "step": 11122
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9316197963576906e-05,
+      "loss": 3.6406,
+      "step": 11123
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.931431486021933e-05,
+      "loss": 4.5312,
+      "step": 11124
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.9312431651413015e-05,
+      "loss": 4.2031,
+      "step": 11125
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.931054833717927e-05,
+      "loss": 3.1875,
+      "step": 11126
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.930866491753941e-05,
+      "loss": 3.75,
+      "step": 11127
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9306781392514773e-05,
+      "loss": 5.5938,
+      "step": 11128
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9304897762126673e-05,
+      "loss": 4.3594,
+      "step": 11129
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9303014026396435e-05,
+      "loss": 4.375,
+      "step": 11130
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.930113018534539e-05,
+      "loss": 4.8281,
+      "step": 11131
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9299246238994857e-05,
+      "loss": 2.25,
+      "step": 11132
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.929736218736618e-05,
+      "loss": 4.2812,
+      "step": 11133
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9295478030480678e-05,
+      "loss": 1.8125,
+      "step": 11134
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.929359376835967e-05,
+      "loss": 3.4375,
+      "step": 11135
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9291709401024507e-05,
+      "loss": 4.125,
+      "step": 11136
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9289824928496518e-05,
+      "loss": 3.9688,
+      "step": 11137
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9287940350797032e-05,
+      "loss": 2.5625,
+      "step": 11138
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.928605566794739e-05,
+      "loss": 7.1875,
+      "step": 11139
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9284170879968923e-05,
+      "loss": 3.5156,
+      "step": 11140
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.928228598688297e-05,
+      "loss": 3.3125,
+      "step": 11141
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9280400988710885e-05,
+      "loss": 4.8438,
+      "step": 11142
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.927851588547398e-05,
+      "loss": 1.8125,
+      "step": 11143
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.927663067719362e-05,
+      "loss": 2.7656,
+      "step": 11144
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9274745363891144e-05,
+      "loss": 2.75,
+      "step": 11145
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9272859945587884e-05,
+      "loss": 3.5,
+      "step": 11146
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.92709744223052e-05,
+      "loss": 2.2188,
+      "step": 11147
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9269088794064428e-05,
+      "loss": 2.8438,
+      "step": 11148
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.926720306088691e-05,
+      "loss": 3.3281,
+      "step": 11149
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.926531722279402e-05,
+      "loss": 3.0,
+      "step": 11150
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.926343127980708e-05,
+      "loss": 3.2812,
+      "step": 11151
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9261545231947454e-05,
+      "loss": 3.6875,
+      "step": 11152
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9259659079236495e-05,
+      "loss": 3.25,
+      "step": 11153
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9257772821695552e-05,
+      "loss": 5.4844,
+      "step": 11154
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9255886459345985e-05,
+      "loss": 2.9219,
+      "step": 11155
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.9253999992209145e-05,
+      "loss": 3.6406,
+      "step": 11156
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9252113420306384e-05,
+      "loss": 2.125,
+      "step": 11157
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9250226743659076e-05,
+      "loss": 2.875,
+      "step": 11158
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9248339962288567e-05,
+      "loss": 4.0469,
+      "step": 11159
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9246453076216218e-05,
+      "loss": 4.9375,
+      "step": 11160
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.92445660854634e-05,
+      "loss": 2.0,
+      "step": 11161
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9242678990051462e-05,
+      "loss": 1.6719,
+      "step": 11162
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9240791790001784e-05,
+      "loss": 2.5312,
+      "step": 11163
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.923890448533572e-05,
+      "loss": 4.7656,
+      "step": 11164
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.923701707607464e-05,
+      "loss": 2.9375,
+      "step": 11165
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9235129562239915e-05,
+      "loss": 2.4062,
+      "step": 11166
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9233241943852905e-05,
+      "loss": 2.7969,
+      "step": 11167
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9231354220934985e-05,
+      "loss": 3.8594,
+      "step": 11168
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9229466393507532e-05,
+      "loss": 3.4688,
+      "step": 11169
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9227578461591907e-05,
+      "loss": 4.6562,
+      "step": 11170
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.922569042520949e-05,
+      "loss": 6.75,
+      "step": 11171
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9223802284381658e-05,
+      "loss": 4.25,
+      "step": 11172
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9221914039129782e-05,
+      "loss": 1.875,
+      "step": 11173
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9220025689475244e-05,
+      "loss": 3.3125,
+      "step": 11174
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9218137235439416e-05,
+      "loss": 3.6094,
+      "step": 11175
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9216248677043685e-05,
+      "loss": 2.2188,
+      "step": 11176
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9214360014309426e-05,
+      "loss": 2.5938,
+      "step": 11177
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.921247124725802e-05,
+      "loss": 3.4375,
+      "step": 11178
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9210582375910856e-05,
+      "loss": 1.5938,
+      "step": 11179
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9208693400289317e-05,
+      "loss": 4.0938,
+      "step": 11180
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.920680432041478e-05,
+      "loss": 3.75,
+      "step": 11181
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9204915136308635e-05,
+      "loss": 3.9844,
+      "step": 11182
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9203025847992283e-05,
+      "loss": 3.9062,
+      "step": 11183
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9201136455487094e-05,
+      "loss": 2.5938,
+      "step": 11184
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9199246958814468e-05,
+      "loss": 3.25,
+      "step": 11185
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.9197357357995793e-05,
+      "loss": 3.4375,
+      "step": 11186
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.919546765305247e-05,
+      "loss": 4.875,
+      "step": 11187
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.919357784400588e-05,
+      "loss": 2.7188,
+      "step": 11188
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9191687930877423e-05,
+      "loss": 3.1875,
+      "step": 11189
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.91897979136885e-05,
+      "loss": 3.0312,
+      "step": 11190
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9187907792460506e-05,
+      "loss": 2.4688,
+      "step": 11191
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.918601756721483e-05,
+      "loss": 3.5156,
+      "step": 11192
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9184127237972878e-05,
+      "loss": 5.375,
+      "step": 11193
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.918223680475605e-05,
+      "loss": 4.4062,
+      "step": 11194
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9180346267585756e-05,
+      "loss": 2.0,
+      "step": 11195
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9178455626483385e-05,
+      "loss": 3.4219,
+      "step": 11196
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9176564881470352e-05,
+      "loss": 2.7812,
+      "step": 11197
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9174674032568064e-05,
+      "loss": 4.3906,
+      "step": 11198
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9172783079797917e-05,
+      "loss": 4.4062,
+      "step": 11199
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.917089202318132e-05,
+      "loss": 6.8125,
+      "step": 11200
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9169000862739697e-05,
+      "loss": 3.3125,
+      "step": 11201
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9167109598494438e-05,
+      "loss": 2.8125,
+      "step": 11202
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9165218230466964e-05,
+      "loss": 4.1094,
+      "step": 11203
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.916332675867869e-05,
+      "loss": 2.6875,
+      "step": 11204
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9161435183151025e-05,
+      "loss": 2.5312,
+      "step": 11205
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9159543503905384e-05,
+      "loss": 3.3438,
+      "step": 11206
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9157651720963192e-05,
+      "loss": 2.1094,
+      "step": 11207
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9155759834345848e-05,
+      "loss": 2.5,
+      "step": 11208
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9153867844074785e-05,
+      "loss": 4.125,
+      "step": 11209
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9151975750171423e-05,
+      "loss": 3.0938,
+      "step": 11210
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9150083552657167e-05,
+      "loss": 2.6562,
+      "step": 11211
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.914819125155346e-05,
+      "loss": 2.4844,
+      "step": 11212
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9146298846881715e-05,
+      "loss": 3.375,
+      "step": 11213
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9144406338663347e-05,
+      "loss": 4.1875,
+      "step": 11214
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.91425137269198e-05,
+      "loss": 2.6875,
+      "step": 11215
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.9140621011672484e-05,
+      "loss": 4.8125,
+      "step": 11216
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.913872819294284e-05,
+      "loss": 2.9688,
+      "step": 11217
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9136835270752287e-05,
+      "loss": 2.4375,
+      "step": 11218
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.913494224512226e-05,
+      "loss": 3.0625,
+      "step": 11219
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9133049116074188e-05,
+      "loss": 3.0938,
+      "step": 11220
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9131155883629504e-05,
+      "loss": 6.4375,
+      "step": 11221
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.912926254780964e-05,
+      "loss": 1.75,
+      "step": 11222
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.912736910863603e-05,
+      "loss": 2.3438,
+      "step": 11223
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9125475566130123e-05,
+      "loss": 3.6875,
+      "step": 11224
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.912358192031334e-05,
+      "loss": 2.9062,
+      "step": 11225
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9121688171207124e-05,
+      "loss": 3.5938,
+      "step": 11226
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9119794318832914e-05,
+      "loss": 2.8125,
+      "step": 11227
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9117900363212154e-05,
+      "loss": 3.625,
+      "step": 11228
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9116006304366284e-05,
+      "loss": 4.0312,
+      "step": 11229
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.911411214231674e-05,
+      "loss": 3.8281,
+      "step": 11230
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.911221787708498e-05,
+      "loss": 3.0,
+      "step": 11231
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.911032350869244e-05,
+      "loss": 7.1562,
+      "step": 11232
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9108429037160572e-05,
+      "loss": 4.0781,
+      "step": 11233
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9106534462510817e-05,
+      "loss": 3.2031,
+      "step": 11234
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9104639784764624e-05,
+      "loss": 3.2188,
+      "step": 11235
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9102745003943448e-05,
+      "loss": 2.4688,
+      "step": 11236
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9100850120068734e-05,
+      "loss": 2.125,
+      "step": 11237
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.909895513316194e-05,
+      "loss": 3.6562,
+      "step": 11238
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9097060043244522e-05,
+      "loss": 3.0,
+      "step": 11239
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.909516485033792e-05,
+      "loss": 3.0,
+      "step": 11240
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.909326955446361e-05,
+      "loss": 3.1875,
+      "step": 11241
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9091374155643035e-05,
+      "loss": 3.0312,
+      "step": 11242
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9089478653897654e-05,
+      "loss": 3.5625,
+      "step": 11243
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9087583049248928e-05,
+      "loss": 2.8125,
+      "step": 11244
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9085687341718322e-05,
+      "loss": 3.2188,
+      "step": 11245
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.908379153132729e-05,
+      "loss": 3.6875,
+      "step": 11246
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.9081895618097306e-05,
+      "loss": 3.0,
+      "step": 11247
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9079999602049823e-05,
+      "loss": 5.2031,
+      "step": 11248
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.907810348320631e-05,
+      "loss": 4.4375,
+      "step": 11249
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9076207261588233e-05,
+      "loss": 3.6094,
+      "step": 11250
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9074310937217055e-05,
+      "loss": 2.625,
+      "step": 11251
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9072414510114254e-05,
+      "loss": 4.625,
+      "step": 11252
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9070517980301295e-05,
+      "loss": 2.2812,
+      "step": 11253
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9068621347799645e-05,
+      "loss": 4.0156,
+      "step": 11254
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9066724612630782e-05,
+      "loss": 2.875,
+      "step": 11255
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9064827774816176e-05,
+      "loss": 3.5,
+      "step": 11256
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9062930834377304e-05,
+      "loss": 7.3438,
+      "step": 11257
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9061033791335637e-05,
+      "loss": 3.0312,
+      "step": 11258
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9059136645712656e-05,
+      "loss": 3.5469,
+      "step": 11259
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.905723939752984e-05,
+      "loss": 4.7656,
+      "step": 11260
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.905534204680866e-05,
+      "loss": 2.125,
+      "step": 11261
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9053444593570605e-05,
+      "loss": 2.3906,
+      "step": 11262
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9051547037837148e-05,
+      "loss": 4.6562,
+      "step": 11263
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9049649379629784e-05,
+      "loss": 4.0781,
+      "step": 11264
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9047751618969986e-05,
+      "loss": 2.4375,
+      "step": 11265
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.904585375587924e-05,
+      "loss": 5.6562,
+      "step": 11266
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9043955790379035e-05,
+      "loss": 5.6719,
+      "step": 11267
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.904205772249086e-05,
+      "loss": 1.7812,
+      "step": 11268
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.90401595522362e-05,
+      "loss": 2.0938,
+      "step": 11269
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9038261279636543e-05,
+      "loss": 2.0781,
+      "step": 11270
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9036362904713375e-05,
+      "loss": 2.0938,
+      "step": 11271
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9034464427488204e-05,
+      "loss": 4.9688,
+      "step": 11272
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9032565847982507e-05,
+      "loss": 4.7344,
+      "step": 11273
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.903066716621779e-05,
+      "loss": 2.3438,
+      "step": 11274
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.902876838221554e-05,
+      "loss": 2.0938,
+      "step": 11275
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9026869495997256e-05,
+      "loss": 4.4219,
+      "step": 11276
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9024970507584435e-05,
+      "loss": 4.2031,
+      "step": 11277
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.902307141699858e-05,
+      "loss": 5.6016,
+      "step": 11278
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.9021172224261178e-05,
+      "loss": 2.2656,
+      "step": 11279
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.901927292939375e-05,
+      "loss": 3.0312,
+      "step": 11280
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.9017373532417782e-05,
+      "loss": 2.125,
+      "step": 11281
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.9015474033354784e-05,
+      "loss": 4.375,
+      "step": 11282
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.9013574432226258e-05,
+      "loss": 6.2969,
+      "step": 11283
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.901167472905371e-05,
+      "loss": 4.0156,
+      "step": 11284
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.900977492385865e-05,
+      "loss": 4.5625,
+      "step": 11285
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.9007875016662584e-05,
+      "loss": 3.6875,
+      "step": 11286
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.9005975007487017e-05,
+      "loss": 2.4375,
+      "step": 11287
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.9004074896353468e-05,
+      "loss": 2.5,
+      "step": 11288
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.9002174683283444e-05,
+      "loss": 2.875,
+      "step": 11289
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.9000274368298456e-05,
+      "loss": 3.9844,
+      "step": 11290
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.8998373951420017e-05,
+      "loss": 3.2188,
+      "step": 11291
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.8996473432669643e-05,
+      "loss": 3.625,
+      "step": 11292
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.8994572812068856e-05,
+      "loss": 3.1875,
+      "step": 11293
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.8992672089639167e-05,
+      "loss": 2.4062,
+      "step": 11294
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.8990771265402088e-05,
+      "loss": 3.9375,
+      "step": 11295
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.898887033937915e-05,
+      "loss": 3.125,
+      "step": 11296
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.8986969311591875e-05,
+      "loss": 4.0625,
+      "step": 11297
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.898506818206178e-05,
+      "loss": 5.6719,
+      "step": 11298
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.898316695081038e-05,
+      "loss": 4.1719,
+      "step": 11299
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.8981265617859212e-05,
+      "loss": 1.9062,
+      "step": 11300
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.897936418322979e-05,
+      "loss": 3.125,
+      "step": 11301
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.8977462646943654e-05,
+      "loss": 1.9062,
+      "step": 11302
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.8975561009022318e-05,
+      "loss": 2.6406,
+      "step": 11303
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.897365926948732e-05,
+      "loss": 3.6875,
+      "step": 11304
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.8971757428360186e-05,
+      "loss": 6.4062,
+      "step": 11305
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.896985548566245e-05,
+      "loss": 2.6406,
+      "step": 11306
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.8967953441415634e-05,
+      "loss": 2.0,
+      "step": 11307
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.8966051295641285e-05,
+      "loss": 4.375,
+      "step": 11308
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.896414904836093e-05,
+      "loss": 4.3438,
+      "step": 11309
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8962246699596107e-05,
+      "loss": 4.0781,
+      "step": 11310
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8960344249368348e-05,
+      "loss": 4.1094,
+      "step": 11311
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.89584416976992e-05,
+      "loss": 6.2812,
+      "step": 11312
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8956539044610196e-05,
+      "loss": 4.75,
+      "step": 11313
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8954636290122876e-05,
+      "loss": 4.9688,
+      "step": 11314
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8952733434258788e-05,
+      "loss": 2.3125,
+      "step": 11315
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.895083047703946e-05,
+      "loss": 2.3125,
+      "step": 11316
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8948927418486447e-05,
+      "loss": 4.625,
+      "step": 11317
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.894702425862129e-05,
+      "loss": 2.8125,
+      "step": 11318
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8945120997465538e-05,
+      "loss": 5.1875,
+      "step": 11319
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8943217635040737e-05,
+      "loss": 3.8125,
+      "step": 11320
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8941314171368438e-05,
+      "loss": 3.3438,
+      "step": 11321
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8939410606470183e-05,
+      "loss": 4.3125,
+      "step": 11322
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.893750694036753e-05,
+      "loss": 2.7969,
+      "step": 11323
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.893560317308203e-05,
+      "loss": 5.0,
+      "step": 11324
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8933699304635224e-05,
+      "loss": 1.8125,
+      "step": 11325
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.893179533504868e-05,
+      "loss": 3.625,
+      "step": 11326
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8929891264343947e-05,
+      "loss": 6.375,
+      "step": 11327
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.892798709254258e-05,
+      "loss": 5.0156,
+      "step": 11328
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8926082819666148e-05,
+      "loss": 4.125,
+      "step": 11329
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8924178445736197e-05,
+      "loss": 3.4844,
+      "step": 11330
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8922273970774292e-05,
+      "loss": 2.625,
+      "step": 11331
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.892036939480199e-05,
+      "loss": 2.7188,
+      "step": 11332
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8918464717840852e-05,
+      "loss": 1.6875,
+      "step": 11333
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8916559939912442e-05,
+      "loss": 3.8125,
+      "step": 11334
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8914655061038336e-05,
+      "loss": 3.5,
+      "step": 11335
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8912750081240077e-05,
+      "loss": 2.5312,
+      "step": 11336
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.891084500053926e-05,
+      "loss": 2.1875,
+      "step": 11337
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.8908939818957427e-05,
+      "loss": 4.2188,
+      "step": 11338
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8907034536516164e-05,
+      "loss": 6.0,
+      "step": 11339
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8905129153237027e-05,
+      "loss": 3.0,
+      "step": 11340
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8903223669141596e-05,
+      "loss": 7.125,
+      "step": 11341
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8901318084251443e-05,
+      "loss": 4.1719,
+      "step": 11342
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8899412398588137e-05,
+      "loss": 2.25,
+      "step": 11343
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8897506612173258e-05,
+      "loss": 2.375,
+      "step": 11344
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8895600725028376e-05,
+      "loss": 3.0625,
+      "step": 11345
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8893694737175076e-05,
+      "loss": 2.375,
+      "step": 11346
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8891788648634928e-05,
+      "loss": 5.8594,
+      "step": 11347
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.888988245942952e-05,
+      "loss": 3.9688,
+      "step": 11348
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8887976169580417e-05,
+      "loss": 2.0312,
+      "step": 11349
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8886069779109212e-05,
+      "loss": 3.75,
+      "step": 11350
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8884163288037486e-05,
+      "loss": 5.4219,
+      "step": 11351
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.888225669638682e-05,
+      "loss": 2.5312,
+      "step": 11352
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8880350004178805e-05,
+      "loss": 2.5625,
+      "step": 11353
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.887844321143502e-05,
+      "loss": 3.5312,
+      "step": 11354
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8876536318177053e-05,
+      "loss": 2.3125,
+      "step": 11355
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8874629324426503e-05,
+      "loss": 5.8438,
+      "step": 11356
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8872722230204943e-05,
+      "loss": 1.4062,
+      "step": 11357
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8870815035533966e-05,
+      "loss": 3.5938,
+      "step": 11358
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8868907740435175e-05,
+      "loss": 7.4688,
+      "step": 11359
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8867000344930153e-05,
+      "loss": 1.875,
+      "step": 11360
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.88650928490405e-05,
+      "loss": 3.0,
+      "step": 11361
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.886318525278781e-05,
+      "loss": 2.0625,
+      "step": 11362
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.886127755619367e-05,
+      "loss": 5.9531,
+      "step": 11363
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8859369759279696e-05,
+      "loss": 2.5,
+      "step": 11364
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8857461862067463e-05,
+      "loss": 2.8594,
+      "step": 11365
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8855553864578588e-05,
+      "loss": 3.4531,
+      "step": 11366
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.885364576683467e-05,
+      "loss": 3.6562,
+      "step": 11367
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.88517375688573e-05,
+      "loss": 3.7344,
+      "step": 11368
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.8849829270668093e-05,
+      "loss": 4.375,
+      "step": 11369
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8847920872288648e-05,
+      "loss": 3.7188,
+      "step": 11370
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8846012373740566e-05,
+      "loss": 3.5312,
+      "step": 11371
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.884410377504547e-05,
+      "loss": 3.25,
+      "step": 11372
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8842195076224945e-05,
+      "loss": 1.625,
+      "step": 11373
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.884028627730061e-05,
+      "loss": 3.7656,
+      "step": 11374
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8838377378294078e-05,
+      "loss": 1.875,
+      "step": 11375
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8836468379226957e-05,
+      "loss": 2.2422,
+      "step": 11376
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8834559280120858e-05,
+      "loss": 3.8594,
+      "step": 11377
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8832650080997396e-05,
+      "loss": 2.9531,
+      "step": 11378
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8830740781878178e-05,
+      "loss": 4.2656,
+      "step": 11379
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.882883138278484e-05,
+      "loss": 2.8125,
+      "step": 11380
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8826921883738982e-05,
+      "loss": 6.1875,
+      "step": 11381
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8825012284762213e-05,
+      "loss": 2.1875,
+      "step": 11382
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8823102585876173e-05,
+      "loss": 4.2969,
+      "step": 11383
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.882119278710247e-05,
+      "loss": 6.4375,
+      "step": 11384
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8819282888462722e-05,
+      "loss": 2.6406,
+      "step": 11385
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8817372889978564e-05,
+      "loss": 3.75,
+      "step": 11386
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.88154627916716e-05,
+      "loss": 2.9844,
+      "step": 11387
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8813552593563484e-05,
+      "loss": 3.4219,
+      "step": 11388
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8811642295675812e-05,
+      "loss": 4.3438,
+      "step": 11389
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8809731898030224e-05,
+      "loss": 3.5469,
+      "step": 11390
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.880782140064835e-05,
+      "loss": 4.1562,
+      "step": 11391
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8805910803551818e-05,
+      "loss": 4.375,
+      "step": 11392
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.880400010676225e-05,
+      "loss": 4.1562,
+      "step": 11393
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8802089310301284e-05,
+      "loss": 4.8281,
+      "step": 11394
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8800178414190552e-05,
+      "loss": 3.7656,
+      "step": 11395
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8798267418451686e-05,
+      "loss": 2.9375,
+      "step": 11396
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.879635632310633e-05,
+      "loss": 3.2812,
+      "step": 11397
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8794445128176102e-05,
+      "loss": 3.1875,
+      "step": 11398
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.8792533833682657e-05,
+      "loss": 3.4688,
+      "step": 11399
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.879062243964762e-05,
+      "loss": 2.1406,
+      "step": 11400
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8788710946092632e-05,
+      "loss": 3.375,
+      "step": 11401
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8786799353039335e-05,
+      "loss": 2.6406,
+      "step": 11402
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8784887660509374e-05,
+      "loss": 2.0312,
+      "step": 11403
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8782975868524393e-05,
+      "loss": 2.3438,
+      "step": 11404
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.878106397710603e-05,
+      "loss": 9.5,
+      "step": 11405
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.877915198627593e-05,
+      "loss": 3.7656,
+      "step": 11406
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.877723989605574e-05,
+      "loss": 2.7344,
+      "step": 11407
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.877532770646711e-05,
+      "loss": 4.3594,
+      "step": 11408
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8773415417531683e-05,
+      "loss": 1.0625,
+      "step": 11409
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.877150302927111e-05,
+      "loss": 2.6875,
+      "step": 11410
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.876959054170705e-05,
+      "loss": 3.0625,
+      "step": 11411
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.876767795486114e-05,
+      "loss": 4.9219,
+      "step": 11412
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8765765268755042e-05,
+      "loss": 4.2812,
+      "step": 11413
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8763852483410403e-05,
+      "loss": 3.4062,
+      "step": 11414
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.876193959884889e-05,
+      "loss": 3.5312,
+      "step": 11415
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8760026615092145e-05,
+      "loss": 2.1562,
+      "step": 11416
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8758113532161833e-05,
+      "loss": 3.7344,
+      "step": 11417
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8756200350079612e-05,
+      "loss": 3.125,
+      "step": 11418
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.875428706886714e-05,
+      "loss": 2.875,
+      "step": 11419
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.875237368854608e-05,
+      "loss": 4.6562,
+      "step": 11420
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.875046020913809e-05,
+      "loss": 3.6719,
+      "step": 11421
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8748546630664832e-05,
+      "loss": 2.3438,
+      "step": 11422
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8746632953147976e-05,
+      "loss": 4.9375,
+      "step": 11423
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8744719176609183e-05,
+      "loss": 3.3125,
+      "step": 11424
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8742805301070115e-05,
+      "loss": 4.2812,
+      "step": 11425
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8740891326552448e-05,
+      "loss": 5.0312,
+      "step": 11426
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8738977253077847e-05,
+      "loss": 2.4688,
+      "step": 11427
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8737063080667974e-05,
+      "loss": 3.3281,
+      "step": 11428
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8735148809344514e-05,
+      "loss": 2.7812,
+      "step": 11429
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.8733234439129128e-05,
+      "loss": 2.4688,
+      "step": 11430
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8731319970043493e-05,
+      "loss": 2.4062,
+      "step": 11431
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8729405402109285e-05,
+      "loss": 2.3594,
+      "step": 11432
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8727490735348165e-05,
+      "loss": 2.0,
+      "step": 11433
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.872557596978183e-05,
+      "loss": 3.7812,
+      "step": 11434
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8723661105431947e-05,
+      "loss": 2.8125,
+      "step": 11435
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8721746142320186e-05,
+      "loss": 3.5938,
+      "step": 11436
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8719831080468242e-05,
+      "loss": 4.7188,
+      "step": 11437
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8717915919897797e-05,
+      "loss": 3.7969,
+      "step": 11438
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8716000660630517e-05,
+      "loss": 3.8438,
+      "step": 11439
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8714085302688093e-05,
+      "loss": 1.875,
+      "step": 11440
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8712169846092207e-05,
+      "loss": 3.4219,
+      "step": 11441
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.871025429086455e-05,
+      "loss": 2.3125,
+      "step": 11442
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8708338637026803e-05,
+      "loss": 3.0,
+      "step": 11443
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.870642288460065e-05,
+      "loss": 6.0,
+      "step": 11444
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8704507033607796e-05,
+      "loss": 5.0625,
+      "step": 11445
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8702591084069915e-05,
+      "loss": 2.4531,
+      "step": 11446
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8700675036008698e-05,
+      "loss": 4.8906,
+      "step": 11447
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8698758889445844e-05,
+      "loss": 5.2812,
+      "step": 11448
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.869684264440304e-05,
+      "loss": 6.5312,
+      "step": 11449
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8694926300901983e-05,
+      "loss": 2.2031,
+      "step": 11450
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.869300985896437e-05,
+      "loss": 3.125,
+      "step": 11451
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8691093318611895e-05,
+      "loss": 3.5,
+      "step": 11452
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8689176679866252e-05,
+      "loss": 1.9062,
+      "step": 11453
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8687259942749154e-05,
+      "loss": 4.2188,
+      "step": 11454
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8685343107282278e-05,
+      "loss": 3.6406,
+      "step": 11455
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8683426173487342e-05,
+      "loss": 3.6406,
+      "step": 11456
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8681509141386043e-05,
+      "loss": 1.7812,
+      "step": 11457
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.867959201100008e-05,
+      "loss": 3.5,
+      "step": 11458
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8677674782351164e-05,
+      "loss": 2.1406,
+      "step": 11459
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.8675757455461003e-05,
+      "loss": 1.4844,
+      "step": 11460
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8673840030351284e-05,
+      "loss": 2.9062,
+      "step": 11461
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.867192250704374e-05,
+      "loss": 2.9531,
+      "step": 11462
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.867000488556006e-05,
+      "loss": 4.375,
+      "step": 11463
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8668087165921962e-05,
+      "loss": 3.2969,
+      "step": 11464
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8666169348151157e-05,
+      "loss": 3.0625,
+      "step": 11465
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8664251432269355e-05,
+      "loss": 2.5625,
+      "step": 11466
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.866233341829827e-05,
+      "loss": 2.8281,
+      "step": 11467
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.866041530625962e-05,
+      "loss": 3.4375,
+      "step": 11468
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.865849709617511e-05,
+      "loss": 3.9375,
+      "step": 11469
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.865657878806647e-05,
+      "loss": 2.1719,
+      "step": 11470
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8654660381955405e-05,
+      "loss": 3.25,
+      "step": 11471
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.865274187786364e-05,
+      "loss": 4.2344,
+      "step": 11472
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.865082327581289e-05,
+      "loss": 4.9062,
+      "step": 11473
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8648904575824877e-05,
+      "loss": 4.6719,
+      "step": 11474
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.864698577792133e-05,
+      "loss": 2.1875,
+      "step": 11475
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8645066882123965e-05,
+      "loss": 3.4375,
+      "step": 11476
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8643147888454507e-05,
+      "loss": 2.4219,
+      "step": 11477
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8641228796934685e-05,
+      "loss": 2.5312,
+      "step": 11478
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.863930960758622e-05,
+      "loss": 2.5625,
+      "step": 11479
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8637390320430847e-05,
+      "loss": 5.2812,
+      "step": 11480
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8635470935490287e-05,
+      "loss": 2.5312,
+      "step": 11481
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8633551452786265e-05,
+      "loss": 2.7188,
+      "step": 11482
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8631631872340527e-05,
+      "loss": 6.3594,
+      "step": 11483
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8629712194174796e-05,
+      "loss": 3.4375,
+      "step": 11484
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.86277924183108e-05,
+      "loss": 3.5156,
+      "step": 11485
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.862587254477028e-05,
+      "loss": 2.9375,
+      "step": 11486
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8623952573574975e-05,
+      "loss": 3.7344,
+      "step": 11487
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8622032504746617e-05,
+      "loss": 2.8438,
+      "step": 11488
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.862011233830694e-05,
+      "loss": 2.7812,
+      "step": 11489
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8618192074277683e-05,
+      "loss": 2.6875,
+      "step": 11490
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.8616271712680586e-05,
+      "loss": 3.0781,
+      "step": 11491
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8614351253537393e-05,
+      "loss": 5.4375,
+      "step": 11492
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8612430696869845e-05,
+      "loss": 4.375,
+      "step": 11493
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8610510042699684e-05,
+      "loss": 4.6562,
+      "step": 11494
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8608589291048655e-05,
+      "loss": 3.1875,
+      "step": 11495
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8606668441938503e-05,
+      "loss": 2.9688,
+      "step": 11496
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8604747495390976e-05,
+      "loss": 2.2812,
+      "step": 11497
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.860282645142781e-05,
+      "loss": 4.3125,
+      "step": 11498
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.860090531007077e-05,
+      "loss": 3.4844,
+      "step": 11499
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8598984071341592e-05,
+      "loss": 1.375,
+      "step": 11500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8597062735262034e-05,
+      "loss": 4.1875,
+      "step": 11501
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8595141301853847e-05,
+      "loss": 2.8125,
+      "step": 11502
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8593219771138786e-05,
+      "loss": 3.5312,
+      "step": 11503
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8591298143138597e-05,
+      "loss": 3.875,
+      "step": 11504
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8589376417875043e-05,
+      "loss": 5.875,
+      "step": 11505
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8587454595369878e-05,
+      "loss": 3.625,
+      "step": 11506
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.858553267564485e-05,
+      "loss": 1.6875,
+      "step": 11507
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8583610658721734e-05,
+      "loss": 4.5078,
+      "step": 11508
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8581688544622275e-05,
+      "loss": 3.2344,
+      "step": 11509
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8579766333368245e-05,
+      "loss": 3.6094,
+      "step": 11510
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8577844024981397e-05,
+      "loss": 2.25,
+      "step": 11511
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8575921619483498e-05,
+      "loss": 3.7812,
+      "step": 11512
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8573999116896312e-05,
+      "loss": 2.2969,
+      "step": 11513
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8572076517241606e-05,
+      "loss": 2.7031,
+      "step": 11514
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8570153820541132e-05,
+      "loss": 4.3906,
+      "step": 11515
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8568231026816674e-05,
+      "loss": 3.2188,
+      "step": 11516
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8566308136089995e-05,
+      "loss": 2.4688,
+      "step": 11517
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8564385148382862e-05,
+      "loss": 2.0938,
+      "step": 11518
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.856246206371705e-05,
+      "loss": 2.7031,
+      "step": 11519
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.8560538882114324e-05,
+      "loss": 4.2812,
+      "step": 11520
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.855861560359646e-05,
+      "loss": 3.375,
+      "step": 11521
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8556692228185237e-05,
+      "loss": 2.5,
+      "step": 11522
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8554768755902416e-05,
+      "loss": 3.4219,
+      "step": 11523
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8552845186769788e-05,
+      "loss": 3.4375,
+      "step": 11524
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.855092152080912e-05,
+      "loss": 2.3438,
+      "step": 11525
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8548997758042194e-05,
+      "loss": 2.5,
+      "step": 11526
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8547073898490788e-05,
+      "loss": 1.8438,
+      "step": 11527
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.854514994217669e-05,
+      "loss": 2.875,
+      "step": 11528
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8543225889121672e-05,
+      "loss": 3.0938,
+      "step": 11529
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8541301739347514e-05,
+      "loss": 3.9219,
+      "step": 11530
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8539377492876004e-05,
+      "loss": 3.5,
+      "step": 11531
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8537453149728934e-05,
+      "loss": 3.3438,
+      "step": 11532
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8535528709928076e-05,
+      "loss": 2.9375,
+      "step": 11533
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8533604173495226e-05,
+      "loss": 3.2969,
+      "step": 11534
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8531679540452173e-05,
+      "loss": 2.0,
+      "step": 11535
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8529754810820703e-05,
+      "loss": 3.2656,
+      "step": 11536
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.85278299846226e-05,
+      "loss": 8.7031,
+      "step": 11537
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.852590506187967e-05,
+      "loss": 4.3438,
+      "step": 11538
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8523980042613687e-05,
+      "loss": 2.7656,
+      "step": 11539
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8522054926846463e-05,
+      "loss": 3.9688,
+      "step": 11540
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8520129714599784e-05,
+      "loss": 5.3281,
+      "step": 11541
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.851820440589544e-05,
+      "loss": 3.125,
+      "step": 11542
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.851627900075523e-05,
+      "loss": 3.2656,
+      "step": 11543
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8514353499200962e-05,
+      "loss": 3.2812,
+      "step": 11544
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.851242790125443e-05,
+      "loss": 3.9062,
+      "step": 11545
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.851050220693742e-05,
+      "loss": 3.0625,
+      "step": 11546
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.850857641627175e-05,
+      "loss": 3.0,
+      "step": 11547
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.850665052927922e-05,
+      "loss": 2.875,
+      "step": 11548
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.850472454598163e-05,
+      "loss": 2.2109,
+      "step": 11549
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8502798466400778e-05,
+      "loss": 2.125,
+      "step": 11550
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.850087229055848e-05,
+      "loss": 2.625,
+      "step": 11551
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.849894601847654e-05,
+      "loss": 4.25,
+      "step": 11552
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.849701965017676e-05,
+      "loss": 1.75,
+      "step": 11553
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.849509318568096e-05,
+      "loss": 3.6562,
+      "step": 11554
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.849316662501094e-05,
+      "loss": 2.375,
+      "step": 11555
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8491239968188507e-05,
+      "loss": 2.4844,
+      "step": 11556
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8489313215235487e-05,
+      "loss": 3.3438,
+      "step": 11557
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.848738636617368e-05,
+      "loss": 1.2812,
+      "step": 11558
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.848545942102491e-05,
+      "loss": 1.6562,
+      "step": 11559
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8483532379810987e-05,
+      "loss": 4.4688,
+      "step": 11560
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8481605242553728e-05,
+      "loss": 2.7188,
+      "step": 11561
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.847967800927495e-05,
+      "loss": 3.5781,
+      "step": 11562
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8477750679996465e-05,
+      "loss": 5.8594,
+      "step": 11563
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8475823254740106e-05,
+      "loss": 5.2656,
+      "step": 11564
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.847389573352769e-05,
+      "loss": 4.8281,
+      "step": 11565
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.847196811638103e-05,
+      "loss": 2.2812,
+      "step": 11566
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.847004040332196e-05,
+      "loss": 3.2031,
+      "step": 11567
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8468112594372298e-05,
+      "loss": 4.5156,
+      "step": 11568
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.846618468955387e-05,
+      "loss": 2.0156,
+      "step": 11569
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8464256688888502e-05,
+      "loss": 2.75,
+      "step": 11570
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.846232859239802e-05,
+      "loss": 2.2188,
+      "step": 11571
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8460400400104252e-05,
+      "loss": 3.7188,
+      "step": 11572
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8458472112029034e-05,
+      "loss": 2.1562,
+      "step": 11573
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8456543728194183e-05,
+      "loss": 4.0781,
+      "step": 11574
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8454615248621542e-05,
+      "loss": 2.25,
+      "step": 11575
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8452686673332946e-05,
+      "loss": 3.5547,
+      "step": 11576
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8450758002350212e-05,
+      "loss": 3.2812,
+      "step": 11577
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.84488292356952e-05,
+      "loss": 2.0,
+      "step": 11578
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.844690037338972e-05,
+      "loss": 1.6406,
+      "step": 11579
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8444971415455627e-05,
+      "loss": 4.0625,
+      "step": 11580
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.844304236191475e-05,
+      "loss": 2.6562,
+      "step": 11581
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.844111321278893e-05,
+      "loss": 3.7969,
+      "step": 11582
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8439183968100007e-05,
+      "loss": 2.5781,
+      "step": 11583
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8437254627869824e-05,
+      "loss": 4.6719,
+      "step": 11584
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8435325192120222e-05,
+      "loss": 3.4375,
+      "step": 11585
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8433395660873046e-05,
+      "loss": 1.4375,
+      "step": 11586
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.843146603415014e-05,
+      "loss": 3.6406,
+      "step": 11587
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8429536311973345e-05,
+      "loss": 3.0,
+      "step": 11588
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8427606494364517e-05,
+      "loss": 2.9375,
+      "step": 11589
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.84256765813455e-05,
+      "loss": 3.0,
+      "step": 11590
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8423746572938133e-05,
+      "loss": 5.3906,
+      "step": 11591
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8421816469164282e-05,
+      "loss": 3.8516,
+      "step": 11592
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.841988627004579e-05,
+      "loss": 3.6719,
+      "step": 11593
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8417955975604507e-05,
+      "loss": 4.4688,
+      "step": 11594
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8416025585862287e-05,
+      "loss": 3.1562,
+      "step": 11595
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8414095100840983e-05,
+      "loss": 4.0312,
+      "step": 11596
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.841216452056246e-05,
+      "loss": 2.6094,
+      "step": 11597
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.841023384504856e-05,
+      "loss": 4.2812,
+      "step": 11598
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.840830307432115e-05,
+      "loss": 1.625,
+      "step": 11599
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8406372208402085e-05,
+      "loss": 2.5625,
+      "step": 11600
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.840444124731323e-05,
+      "loss": 5.375,
+      "step": 11601
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8402510191076445e-05,
+      "loss": 2.8125,
+      "step": 11602
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8400579039713583e-05,
+      "loss": 6.5156,
+      "step": 11603
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8398647793246512e-05,
+      "loss": 3.4844,
+      "step": 11604
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.83967164516971e-05,
+      "loss": 3.2812,
+      "step": 11605
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8394785015087202e-05,
+      "loss": 3.4219,
+      "step": 11606
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8392853483438693e-05,
+      "loss": 3.1875,
+      "step": 11607
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.839092185677344e-05,
+      "loss": 3.1406,
+      "step": 11608
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8388990135113307e-05,
+      "loss": 1.6875,
+      "step": 11609
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8387058318480166e-05,
+      "loss": 4.1875,
+      "step": 11610
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8385126406895883e-05,
+      "loss": 3.8438,
+      "step": 11611
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8383194400382333e-05,
+      "loss": 3.8906,
+      "step": 11612
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.838126229896139e-05,
+      "loss": 3.9531,
+      "step": 11613
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.837933010265493e-05,
+      "loss": 1.5,
+      "step": 11614
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8377397811484816e-05,
+      "loss": 8.8125,
+      "step": 11615
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8375465425472937e-05,
+      "loss": 2.5312,
+      "step": 11616
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8373532944641168e-05,
+      "loss": 2.5312,
+      "step": 11617
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8371600369011375e-05,
+      "loss": 4.4688,
+      "step": 11618
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8369667698605446e-05,
+      "loss": 3.2812,
+      "step": 11619
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8367734933445262e-05,
+      "loss": 2.6562,
+      "step": 11620
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8365802073552707e-05,
+      "loss": 1.6406,
+      "step": 11621
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8363869118949653e-05,
+      "loss": 3.8594,
+      "step": 11622
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8361936069657986e-05,
+      "loss": 5.25,
+      "step": 11623
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.83600029256996e-05,
+      "loss": 4.3125,
+      "step": 11624
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.835806968709637e-05,
+      "loss": 4.4375,
+      "step": 11625
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8356136353870182e-05,
+      "loss": 2.9375,
+      "step": 11626
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.835420292604294e-05,
+      "loss": 4.4531,
+      "step": 11627
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8352269403636507e-05,
+      "loss": 4.875,
+      "step": 11628
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8350335786672794e-05,
+      "loss": 3.5781,
+      "step": 11629
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8348402075173683e-05,
+      "loss": 3.125,
+      "step": 11630
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.834646826916106e-05,
+      "loss": 1.8125,
+      "step": 11631
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.834453436865683e-05,
+      "loss": 2.125,
+      "step": 11632
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8342600373682878e-05,
+      "loss": 2.75,
+      "step": 11633
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8340666284261104e-05,
+      "loss": 3.9375,
+      "step": 11634
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8338732100413406e-05,
+      "loss": 5.4844,
+      "step": 11635
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8336797822161676e-05,
+      "loss": 3.4375,
+      "step": 11636
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.833486344952781e-05,
+      "loss": 3.5,
+      "step": 11637
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8332928982533717e-05,
+      "loss": 3.9062,
+      "step": 11638
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.833099442120129e-05,
+      "loss": 2.7344,
+      "step": 11639
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8329059765552424e-05,
+      "loss": 2.0625,
+      "step": 11640
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.8327125015609042e-05,
+      "loss": 2.4688,
+      "step": 11641
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.832519017139302e-05,
+      "loss": 3.0938,
+      "step": 11642
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.83232552329263e-05,
+      "loss": 3.8281,
+      "step": 11643
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.832132020023075e-05,
+      "loss": 2.625,
+      "step": 11644
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8319385073328294e-05,
+      "loss": 3.4062,
+      "step": 11645
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8317449852240843e-05,
+      "loss": 6.2031,
+      "step": 11646
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8315514536990298e-05,
+      "loss": 3.625,
+      "step": 11647
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.831357912759857e-05,
+      "loss": 6.5938,
+      "step": 11648
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8311643624087573e-05,
+      "loss": 4.5,
+      "step": 11649
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8309708026479222e-05,
+      "loss": 5.1094,
+      "step": 11650
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.830777233479542e-05,
+      "loss": 3.8438,
+      "step": 11651
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8305836549058098e-05,
+      "loss": 3.8125,
+      "step": 11652
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8303900669289153e-05,
+      "loss": 3.5469,
+      "step": 11653
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8301964695510516e-05,
+      "loss": 4.3984,
+      "step": 11654
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8300028627744093e-05,
+      "loss": 4.8438,
+      "step": 11655
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8298092466011806e-05,
+      "loss": 5.1875,
+      "step": 11656
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8296156210335575e-05,
+      "loss": 3.1875,
+      "step": 11657
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.829421986073733e-05,
+      "loss": 3.9219,
+      "step": 11658
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.829228341723898e-05,
+      "loss": 2.9688,
+      "step": 11659
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8290346879862452e-05,
+      "loss": 2.0625,
+      "step": 11660
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8288410248629668e-05,
+      "loss": 4.0625,
+      "step": 11661
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8286473523562558e-05,
+      "loss": 5.3281,
+      "step": 11662
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.828453670468305e-05,
+      "loss": 5.0312,
+      "step": 11663
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.828259979201305e-05,
+      "loss": 3.3125,
+      "step": 11664
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8280662785574518e-05,
+      "loss": 4.5,
+      "step": 11665
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.827872568538936e-05,
+      "loss": 2.2188,
+      "step": 11666
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.827678849147951e-05,
+      "loss": 3.7188,
+      "step": 11667
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8274851203866914e-05,
+      "loss": 8.5938,
+      "step": 11668
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.827291382257348e-05,
+      "loss": 3.9062,
+      "step": 11669
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.827097634762116e-05,
+      "loss": 3.4375,
+      "step": 11670
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8269038779031882e-05,
+      "loss": 2.8438,
+      "step": 11671
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8267101116827584e-05,
+      "loss": 3.9844,
+      "step": 11672
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.8265163361030196e-05,
+      "loss": 4.9844,
+      "step": 11673
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8263225511661665e-05,
+      "loss": 3.8125,
+      "step": 11674
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.826128756874392e-05,
+      "loss": 2.0625,
+      "step": 11675
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8259349532298916e-05,
+      "loss": 5.8125,
+      "step": 11676
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8257411402348576e-05,
+      "loss": 2.5938,
+      "step": 11677
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8255473178914853e-05,
+      "loss": 2.9531,
+      "step": 11678
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.825353486201968e-05,
+      "loss": 3.5156,
+      "step": 11679
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.825159645168501e-05,
+      "loss": 2.8125,
+      "step": 11680
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8249657947932787e-05,
+      "loss": 3.1094,
+      "step": 11681
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.824771935078496e-05,
+      "loss": 2.9062,
+      "step": 11682
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.824578066026346e-05,
+      "loss": 3.9062,
+      "step": 11683
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8243841876390256e-05,
+      "loss": 4.9375,
+      "step": 11684
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8241902999187285e-05,
+      "loss": 5.0312,
+      "step": 11685
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8239964028676503e-05,
+      "loss": 4.2188,
+      "step": 11686
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8238024964879857e-05,
+      "loss": 2.5625,
+      "step": 11687
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.82360858078193e-05,
+      "loss": 2.4062,
+      "step": 11688
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8234146557516785e-05,
+      "loss": 6.0312,
+      "step": 11689
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8232207213994275e-05,
+      "loss": 2.7812,
+      "step": 11690
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8230267777273712e-05,
+      "loss": 2.6562,
+      "step": 11691
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8228328247377065e-05,
+      "loss": 5.375,
+      "step": 11692
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8226388624326286e-05,
+      "loss": 2.4688,
+      "step": 11693
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8224448908143327e-05,
+      "loss": 3.25,
+      "step": 11694
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8222509098850165e-05,
+      "loss": 2.4375,
+      "step": 11695
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.822056919646874e-05,
+      "loss": 2.25,
+      "step": 11696
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.821862920102103e-05,
+      "loss": 3.0,
+      "step": 11697
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8216689112528995e-05,
+      "loss": 4.4844,
+      "step": 11698
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.821474893101459e-05,
+      "loss": 2.2188,
+      "step": 11699
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.821280865649979e-05,
+      "loss": 4.4375,
+      "step": 11700
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8210868289006565e-05,
+      "loss": 1.9844,
+      "step": 11701
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8208927828556866e-05,
+      "loss": 3.6562,
+      "step": 11702
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8206987275172674e-05,
+      "loss": 2.25,
+      "step": 11703
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.8205046628875954e-05,
+      "loss": 3.4219,
+      "step": 11704
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8203105889688667e-05,
+      "loss": 3.6094,
+      "step": 11705
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8201165057632807e-05,
+      "loss": 2.0625,
+      "step": 11706
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.819922413273033e-05,
+      "loss": 5.1094,
+      "step": 11707
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8197283115003214e-05,
+      "loss": 2.8125,
+      "step": 11708
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8195342004473434e-05,
+      "loss": 4.5312,
+      "step": 11709
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.819340080116296e-05,
+      "loss": 2.4531,
+      "step": 11710
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.819145950509378e-05,
+      "loss": 2.4531,
+      "step": 11711
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.818951811628786e-05,
+      "loss": 5.9375,
+      "step": 11712
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.818757663476718e-05,
+      "loss": 2.6875,
+      "step": 11713
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8185635060553727e-05,
+      "loss": 2.5625,
+      "step": 11714
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.818369339366948e-05,
+      "loss": 3.2656,
+      "step": 11715
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.818175163413641e-05,
+      "loss": 5.375,
+      "step": 11716
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8179809781976523e-05,
+      "loss": 3.0625,
+      "step": 11717
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.817786783721178e-05,
+      "loss": 2.4688,
+      "step": 11718
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8175925799864177e-05,
+      "loss": 5.0938,
+      "step": 11719
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.81739836699557e-05,
+      "loss": 3.2656,
+      "step": 11720
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8172041447508326e-05,
+      "loss": 3.6406,
+      "step": 11721
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.817009913254406e-05,
+      "loss": 2.125,
+      "step": 11722
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8168156725084884e-05,
+      "loss": 2.7812,
+      "step": 11723
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.816621422515278e-05,
+      "loss": 3.25,
+      "step": 11724
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8164271632769754e-05,
+      "loss": 3.125,
+      "step": 11725
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8162328947957783e-05,
+      "loss": 2.7031,
+      "step": 11726
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.816038617073887e-05,
+      "loss": 2.9375,
+      "step": 11727
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8158443301135014e-05,
+      "loss": 3.375,
+      "step": 11728
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8156500339168195e-05,
+      "loss": 3.3438,
+      "step": 11729
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8154557284860425e-05,
+      "loss": 3.4062,
+      "step": 11730
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8152614138233697e-05,
+      "loss": 5.2812,
+      "step": 11731
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8150670899309997e-05,
+      "loss": 5.4375,
+      "step": 11732
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.8148727568111348e-05,
+      "loss": 3.2188,
+      "step": 11733
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.814678414465973e-05,
+      "loss": 2.0625,
+      "step": 11734
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.814484062897716e-05,
+      "loss": 4.2031,
+      "step": 11735
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.814289702108563e-05,
+      "loss": 3.0625,
+      "step": 11736
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.814095332100715e-05,
+      "loss": 1.625,
+      "step": 11737
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8139009528763716e-05,
+      "loss": 8.8281,
+      "step": 11738
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.813706564437735e-05,
+      "loss": 2.25,
+      "step": 11739
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.813512166787004e-05,
+      "loss": 4.1719,
+      "step": 11740
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8133177599263814e-05,
+      "loss": 2.875,
+      "step": 11741
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8131233438580666e-05,
+      "loss": 2.7188,
+      "step": 11742
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.812928918584261e-05,
+      "loss": 4.0156,
+      "step": 11743
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.812734484107166e-05,
+      "loss": 3.6719,
+      "step": 11744
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8125400404289825e-05,
+      "loss": 2.8438,
+      "step": 11745
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.812345587551912e-05,
+      "loss": 3.5156,
+      "step": 11746
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8121511254781563e-05,
+      "loss": 3.2812,
+      "step": 11747
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8119566542099156e-05,
+      "loss": 3.4688,
+      "step": 11748
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8117621737493932e-05,
+      "loss": 1.7812,
+      "step": 11749
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.81156768409879e-05,
+      "loss": 1.6406,
+      "step": 11750
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8113731852603083e-05,
+      "loss": 3.0312,
+      "step": 11751
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.811178677236149e-05,
+      "loss": 3.5312,
+      "step": 11752
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.810984160028515e-05,
+      "loss": 3.3906,
+      "step": 11753
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8107896336396087e-05,
+      "loss": 4.7188,
+      "step": 11754
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.810595098071632e-05,
+      "loss": 1.6562,
+      "step": 11755
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.810400553326787e-05,
+      "loss": 4.625,
+      "step": 11756
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8102059994072762e-05,
+      "loss": 4.3906,
+      "step": 11757
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8100114363153027e-05,
+      "loss": 4.125,
+      "step": 11758
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.809816864053069e-05,
+      "loss": 2.2188,
+      "step": 11759
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.809622282622778e-05,
+      "loss": 1.5938,
+      "step": 11760
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8094276920266315e-05,
+      "loss": 4.5938,
+      "step": 11761
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.809233092266834e-05,
+      "loss": 3.0,
+      "step": 11762
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8090384833455874e-05,
+      "loss": 3.1875,
+      "step": 11763
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.8088438652650952e-05,
+      "loss": 3.0938,
+      "step": 11764
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.808649238027562e-05,
+      "loss": 3.1094,
+      "step": 11765
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.808454601635189e-05,
+      "loss": 2.7812,
+      "step": 11766
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.808259956090182e-05,
+      "loss": 1.4688,
+      "step": 11767
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.808065301394743e-05,
+      "loss": 3.4062,
+      "step": 11768
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8078706375510755e-05,
+      "loss": 2.0625,
+      "step": 11769
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8076759645613845e-05,
+      "loss": 2.1562,
+      "step": 11770
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8074812824278736e-05,
+      "loss": 4.2656,
+      "step": 11771
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.807286591152746e-05,
+      "loss": 3.5938,
+      "step": 11772
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8070918907382068e-05,
+      "loss": 5.1562,
+      "step": 11773
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.80689718118646e-05,
+      "loss": 3.125,
+      "step": 11774
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8067024624997094e-05,
+      "loss": 4.5,
+      "step": 11775
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8065077346801605e-05,
+      "loss": 2.0625,
+      "step": 11776
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8063129977300172e-05,
+      "loss": 3.4688,
+      "step": 11777
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8061182516514832e-05,
+      "loss": 3.4688,
+      "step": 11778
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.805923496446765e-05,
+      "loss": 3.375,
+      "step": 11779
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8057287321180668e-05,
+      "loss": 4.5312,
+      "step": 11780
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8055339586675923e-05,
+      "loss": 2.5312,
+      "step": 11781
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8053391760975488e-05,
+      "loss": 2.9688,
+      "step": 11782
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8051443844101398e-05,
+      "loss": 2.5625,
+      "step": 11783
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8049495836075715e-05,
+      "loss": 2.8281,
+      "step": 11784
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8047547736920484e-05,
+      "loss": 3.4375,
+      "step": 11785
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8045599546657766e-05,
+      "loss": 5.2188,
+      "step": 11786
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8043651265309618e-05,
+      "loss": 6.9375,
+      "step": 11787
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.804170289289809e-05,
+      "loss": 7.6719,
+      "step": 11788
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8039754429445243e-05,
+      "loss": 4.8594,
+      "step": 11789
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8037805874973137e-05,
+      "loss": 2.6562,
+      "step": 11790
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8035857229503834e-05,
+      "loss": 3.3125,
+      "step": 11791
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8033908493059394e-05,
+      "loss": 3.4688,
+      "step": 11792
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8031959665661875e-05,
+      "loss": 2.2188,
+      "step": 11793
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8030010747333336e-05,
+      "loss": 3.2812,
+      "step": 11794
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.8028061738095854e-05,
+      "loss": 3.6562,
+      "step": 11795
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.802611263797149e-05,
+      "loss": 2.5312,
+      "step": 11796
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8024163446982302e-05,
+      "loss": 3.7031,
+      "step": 11797
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8022214165150363e-05,
+      "loss": 3.3125,
+      "step": 11798
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.802026479249774e-05,
+      "loss": 2.1875,
+      "step": 11799
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8018315329046503e-05,
+      "loss": 4.9688,
+      "step": 11800
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8016365774818726e-05,
+      "loss": 2.375,
+      "step": 11801
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8014416129836467e-05,
+      "loss": 2.7812,
+      "step": 11802
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.801246639412182e-05,
+      "loss": 3.0938,
+      "step": 11803
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8010516567696836e-05,
+      "loss": 2.0625,
+      "step": 11804
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8008566650583596e-05,
+      "loss": 5.4688,
+      "step": 11805
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8006616642804187e-05,
+      "loss": 3.0156,
+      "step": 11806
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8004666544380672e-05,
+      "loss": 3.25,
+      "step": 11807
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.8002716355335135e-05,
+      "loss": 5.5312,
+      "step": 11808
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.800076607568965e-05,
+      "loss": 5.0469,
+      "step": 11809
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.79988157054663e-05,
+      "loss": 1.1562,
+      "step": 11810
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.799686524468716e-05,
+      "loss": 3.4062,
+      "step": 11811
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.799491469337432e-05,
+      "loss": 3.25,
+      "step": 11812
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.7992964051549852e-05,
+      "loss": 2.8125,
+      "step": 11813
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.799101331923585e-05,
+      "loss": 2.7188,
+      "step": 11814
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.7989062496454395e-05,
+      "loss": 3.4062,
+      "step": 11815
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.798711158322757e-05,
+      "loss": 2.4219,
+      "step": 11816
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.7985160579577468e-05,
+      "loss": 2.5625,
+      "step": 11817
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.7983209485526158e-05,
+      "loss": 3.5312,
+      "step": 11818
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.798125830109575e-05,
+      "loss": 3.8281,
+      "step": 11819
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.797930702630833e-05,
+      "loss": 3.9375,
+      "step": 11820
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.7977355661185978e-05,
+      "loss": 3.1875,
+      "step": 11821
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.7975404205750795e-05,
+      "loss": 5.8281,
+      "step": 11822
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.7973452660024872e-05,
+      "loss": 5.0625,
+      "step": 11823
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.7971501024030305e-05,
+      "loss": 2.4062,
+      "step": 11824
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.7969549297789186e-05,
+      "loss": 4.5312,
+      "step": 11825
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7967597481323604e-05,
+      "loss": 4.2188,
+      "step": 11826
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.796564557465566e-05,
+      "loss": 2.5,
+      "step": 11827
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.796369357780746e-05,
+      "loss": 3.2812,
+      "step": 11828
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.796174149080109e-05,
+      "loss": 2.2188,
+      "step": 11829
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.795978931365866e-05,
+      "loss": 2.5,
+      "step": 11830
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7957837046402274e-05,
+      "loss": 3.5938,
+      "step": 11831
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.795588468905402e-05,
+      "loss": 2.2812,
+      "step": 11832
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7953932241636014e-05,
+      "loss": 5.5703,
+      "step": 11833
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7951979704170353e-05,
+      "loss": 3.2188,
+      "step": 11834
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.795002707667914e-05,
+      "loss": 3.4375,
+      "step": 11835
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7948074359184487e-05,
+      "loss": 2.9219,
+      "step": 11836
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7946121551708497e-05,
+      "loss": 3.2188,
+      "step": 11837
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7944168654273275e-05,
+      "loss": 4.4062,
+      "step": 11838
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.794221566690094e-05,
+      "loss": 3.2812,
+      "step": 11839
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7940262589613597e-05,
+      "loss": 3.75,
+      "step": 11840
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7938309422433357e-05,
+      "loss": 4.3125,
+      "step": 11841
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7936356165382328e-05,
+      "loss": 3.9844,
+      "step": 11842
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7934402818482628e-05,
+      "loss": 2.4688,
+      "step": 11843
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.793244938175637e-05,
+      "loss": 3.375,
+      "step": 11844
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.793049585522567e-05,
+      "loss": 2.3125,
+      "step": 11845
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7928542238912636e-05,
+      "loss": 2.125,
+      "step": 11846
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.79265885328394e-05,
+      "loss": 5.25,
+      "step": 11847
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.792463473702807e-05,
+      "loss": 7.875,
+      "step": 11848
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7922680851500773e-05,
+      "loss": 4.0469,
+      "step": 11849
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.792072687627962e-05,
+      "loss": 2.8906,
+      "step": 11850
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.791877281138673e-05,
+      "loss": 2.4062,
+      "step": 11851
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.791681865684424e-05,
+      "loss": 4.1875,
+      "step": 11852
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7914864412674263e-05,
+      "loss": 4.625,
+      "step": 11853
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.791291007889892e-05,
+      "loss": 3.5625,
+      "step": 11854
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7910955655540348e-05,
+      "loss": 4.3438,
+      "step": 11855
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.7909001142620663e-05,
+      "loss": 5.2656,
+      "step": 11856
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7907046540162e-05,
+      "loss": 3.1875,
+      "step": 11857
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7905091848186478e-05,
+      "loss": 2.1875,
+      "step": 11858
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7903137066716233e-05,
+      "loss": 3.4062,
+      "step": 11859
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7901182195773398e-05,
+      "loss": 3.1875,
+      "step": 11860
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.78992272353801e-05,
+      "loss": 3.25,
+      "step": 11861
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7897272185558466e-05,
+      "loss": 1.7344,
+      "step": 11862
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.789531704633064e-05,
+      "loss": 2.375,
+      "step": 11863
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7893361817718746e-05,
+      "loss": 3.2812,
+      "step": 11864
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7891406499744934e-05,
+      "loss": 4.125,
+      "step": 11865
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.788945109243133e-05,
+      "loss": 3.0,
+      "step": 11866
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7887495595800068e-05,
+      "loss": 2.2812,
+      "step": 11867
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7885540009873293e-05,
+      "loss": 3.2031,
+      "step": 11868
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7883584334673144e-05,
+      "loss": 3.6875,
+      "step": 11869
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7881628570221754e-05,
+      "loss": 2.3438,
+      "step": 11870
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.787967271654128e-05,
+      "loss": 2.75,
+      "step": 11871
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7877716773653856e-05,
+      "loss": 2.4688,
+      "step": 11872
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7875760741581617e-05,
+      "loss": 2.4688,
+      "step": 11873
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.787380462034673e-05,
+      "loss": 3.1406,
+      "step": 11874
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7871848409971307e-05,
+      "loss": 2.0938,
+      "step": 11875
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7869892110477525e-05,
+      "loss": 3.625,
+      "step": 11876
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.786793572188752e-05,
+      "loss": 4.25,
+      "step": 11877
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7865979244223434e-05,
+      "loss": 3.3594,
+      "step": 11878
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.786402267750743e-05,
+      "loss": 2.5,
+      "step": 11879
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.786206602176165e-05,
+      "loss": 3.8281,
+      "step": 11880
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7860109277008243e-05,
+      "loss": 4.0312,
+      "step": 11881
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.785815244326938e-05,
+      "loss": 4.7188,
+      "step": 11882
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7856195520567183e-05,
+      "loss": 3.5312,
+      "step": 11883
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7854238508923838e-05,
+      "loss": 4.375,
+      "step": 11884
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.7852281408361484e-05,
+      "loss": 3.5,
+      "step": 11885
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.785032421890227e-05,
+      "loss": 2.625,
+      "step": 11886
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7848366940568373e-05,
+      "loss": 2.375,
+      "step": 11887
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7846409573381948e-05,
+      "loss": 2.6562,
+      "step": 11888
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.784445211736514e-05,
+      "loss": 4.875,
+      "step": 11889
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7842494572540127e-05,
+      "loss": 3.0938,
+      "step": 11890
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.784053693892906e-05,
+      "loss": 3.4375,
+      "step": 11891
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7838579216554104e-05,
+      "loss": 2.3125,
+      "step": 11892
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7836621405437417e-05,
+      "loss": 3.8438,
+      "step": 11893
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7834663505601178e-05,
+      "loss": 3.5,
+      "step": 11894
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.783270551706754e-05,
+      "loss": 2.5,
+      "step": 11895
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.783074743985868e-05,
+      "loss": 3.5625,
+      "step": 11896
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7828789273996748e-05,
+      "loss": 3.0625,
+      "step": 11897
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7826831019503937e-05,
+      "loss": 4.375,
+      "step": 11898
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.78248726764024e-05,
+      "loss": 4.4531,
+      "step": 11899
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.782291424471431e-05,
+      "loss": 4.5312,
+      "step": 11900
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.782095572446184e-05,
+      "loss": 2.3906,
+      "step": 11901
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.781899711566717e-05,
+      "loss": 4.1562,
+      "step": 11902
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7817038418352455e-05,
+      "loss": 1.8125,
+      "step": 11903
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7815079632539892e-05,
+      "loss": 4.4062,
+      "step": 11904
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7813120758251647e-05,
+      "loss": 2.6562,
+      "step": 11905
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.781116179550989e-05,
+      "loss": 3.1875,
+      "step": 11906
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.780920274433681e-05,
+      "loss": 2.8438,
+      "step": 11907
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7807243604754573e-05,
+      "loss": 3.9688,
+      "step": 11908
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7805284376785374e-05,
+      "loss": 4.625,
+      "step": 11909
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7803325060451387e-05,
+      "loss": 2.1562,
+      "step": 11910
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.780136565577478e-05,
+      "loss": 2.75,
+      "step": 11911
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7799406162777764e-05,
+      "loss": 2.3438,
+      "step": 11912
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.77974465814825e-05,
+      "loss": 2.8125,
+      "step": 11913
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7795486911911175e-05,
+      "loss": 3.2969,
+      "step": 11914
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.779352715408599e-05,
+      "loss": 5.3594,
+      "step": 11915
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.779156730802911e-05,
+      "loss": 4.9375,
+      "step": 11916
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7789607373762746e-05,
+      "loss": 2.0312,
+      "step": 11917
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.778764735130907e-05,
+      "loss": 5.625,
+      "step": 11918
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7785687240690272e-05,
+      "loss": 4.8125,
+      "step": 11919
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.778372704192855e-05,
+      "loss": 2.0938,
+      "step": 11920
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7781766755046097e-05,
+      "loss": 3.2656,
+      "step": 11921
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.777980638006509e-05,
+      "loss": 2.0469,
+      "step": 11922
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.777784591700775e-05,
+      "loss": 4.2812,
+      "step": 11923
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7775885365896243e-05,
+      "loss": 2.2188,
+      "step": 11924
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7773924726752786e-05,
+      "loss": 3.0625,
+      "step": 11925
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.777196399959956e-05,
+      "loss": 4.4219,
+      "step": 11926
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.777000318445877e-05,
+      "loss": 3.25,
+      "step": 11927
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7768042281352623e-05,
+      "loss": 2.6719,
+      "step": 11928
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.77660812903033e-05,
+      "loss": 2.3438,
+      "step": 11929
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7764120211333018e-05,
+      "loss": 5.2656,
+      "step": 11930
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7762159044463976e-05,
+      "loss": 2.8438,
+      "step": 11931
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7760197789718365e-05,
+      "loss": 4.2344,
+      "step": 11932
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7758236447118406e-05,
+      "loss": 3.3438,
+      "step": 11933
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7756275016686287e-05,
+      "loss": 3.3125,
+      "step": 11934
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7754313498444224e-05,
+      "loss": 2.6875,
+      "step": 11935
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7752351892414416e-05,
+      "loss": 4.2812,
+      "step": 11936
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7750390198619083e-05,
+      "loss": 5.4375,
+      "step": 11937
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7748428417080423e-05,
+      "loss": 4.2812,
+      "step": 11938
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7746466547820656e-05,
+      "loss": 2.7109,
+      "step": 11939
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7744504590861977e-05,
+      "loss": 1.5625,
+      "step": 11940
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7742542546226607e-05,
+      "loss": 4.4375,
+      "step": 11941
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7740580413936757e-05,
+      "loss": 2.2812,
+      "step": 11942
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.773861819401464e-05,
+      "loss": 2.9375,
+      "step": 11943
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7736655886482475e-05,
+      "loss": 4.0,
+      "step": 11944
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7734693491362476e-05,
+      "loss": 3.25,
+      "step": 11945
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.773273100867685e-05,
+      "loss": 2.5938,
+      "step": 11946
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.7730768438447833e-05,
+      "loss": 3.6875,
+      "step": 11947
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.772880578069763e-05,
+      "loss": 3.8906,
+      "step": 11948
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.772684303544846e-05,
+      "loss": 2.9375,
+      "step": 11949
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7724880202722548e-05,
+      "loss": 3.2031,
+      "step": 11950
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7722917282542115e-05,
+      "loss": 2.9688,
+      "step": 11951
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7720954274929383e-05,
+      "loss": 2.8438,
+      "step": 11952
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7718991179906578e-05,
+      "loss": 2.8438,
+      "step": 11953
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7717027997495915e-05,
+      "loss": 2.5625,
+      "step": 11954
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7715064727719635e-05,
+      "loss": 5.0312,
+      "step": 11955
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7713101370599957e-05,
+      "loss": 2.4062,
+      "step": 11956
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.77111379261591e-05,
+      "loss": 3.375,
+      "step": 11957
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.770917439441931e-05,
+      "loss": 3.4688,
+      "step": 11958
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7707210775402798e-05,
+      "loss": 3.2188,
+      "step": 11959
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7705247069131802e-05,
+      "loss": 3.3438,
+      "step": 11960
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.770328327562856e-05,
+      "loss": 2.2656,
+      "step": 11961
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7701319394915292e-05,
+      "loss": 2.5312,
+      "step": 11962
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7699355427014243e-05,
+      "loss": 2.5,
+      "step": 11963
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7697391371947646e-05,
+      "loss": 5.6094,
+      "step": 11964
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.769542722973772e-05,
+      "loss": 3.1562,
+      "step": 11965
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7693463000406725e-05,
+      "loss": 2.4062,
+      "step": 11966
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7691498683976888e-05,
+      "loss": 2.2812,
+      "step": 11967
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.768953428047044e-05,
+      "loss": 2.25,
+      "step": 11968
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.768756978990963e-05,
+      "loss": 3.9062,
+      "step": 11969
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.76856052123167e-05,
+      "loss": 2.7812,
+      "step": 11970
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7683640547713875e-05,
+      "loss": 1.9375,
+      "step": 11971
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7681675796123422e-05,
+      "loss": 3.4688,
+      "step": 11972
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7679710957567568e-05,
+      "loss": 3.0938,
+      "step": 11973
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7677746032068556e-05,
+      "loss": 3.5,
+      "step": 11974
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.767578101964864e-05,
+      "loss": 2.7188,
+      "step": 11975
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.7673815920330053e-05,
+      "loss": 5.0312,
+      "step": 11976
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.767185073413506e-05,
+      "loss": 6.2344,
+      "step": 11977
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.76698854610859e-05,
+      "loss": 4.0781,
+      "step": 11978
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7667920101204816e-05,
+      "loss": 3.5,
+      "step": 11979
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.766595465451407e-05,
+      "loss": 2.5312,
+      "step": 11980
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7663989121035907e-05,
+      "loss": 3.25,
+      "step": 11981
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.766202350079258e-05,
+      "loss": 2.75,
+      "step": 11982
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7660057793806344e-05,
+      "loss": 2.4219,
+      "step": 11983
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7658092000099445e-05,
+      "loss": 5.3594,
+      "step": 11984
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7656126119694143e-05,
+      "loss": 3.1875,
+      "step": 11985
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.76541601526127e-05,
+      "loss": 5.9062,
+      "step": 11986
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7652194098877364e-05,
+      "loss": 3.4375,
+      "step": 11987
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7650227958510397e-05,
+      "loss": 2.7344,
+      "step": 11988
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7648261731534063e-05,
+      "loss": 3.1875,
+      "step": 11989
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7646295417970613e-05,
+      "loss": 4.0938,
+      "step": 11990
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7644329017842314e-05,
+      "loss": 5.0781,
+      "step": 11991
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.764236253117142e-05,
+      "loss": 2.4062,
+      "step": 11992
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7640395957980203e-05,
+      "loss": 4.3125,
+      "step": 11993
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7638429298290925e-05,
+      "loss": 5.8125,
+      "step": 11994
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.763646255212584e-05,
+      "loss": 3.4375,
+      "step": 11995
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7634495719507235e-05,
+      "loss": 2.9688,
+      "step": 11996
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7632528800457367e-05,
+      "loss": 3.1875,
+      "step": 11997
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7630561794998493e-05,
+      "loss": 4.1875,
+      "step": 11998
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.7628594703152894e-05,
+      "loss": 3.3125,
+      "step": 11999
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.762662752494283e-05,
+      "loss": 2.9062,
+      "step": 12000
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 30440,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 3000,
+  "total_flos": 2.165426633229258e+22,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}