diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,22117 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 3682,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.801801801801802e-07,
+      "loss": 2.4043,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.603603603603604e-07,
+      "loss": 2.3711,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-07,
+      "loss": 2.248,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.207207207207208e-07,
+      "loss": 2.3691,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.00900900900901e-07,
+      "loss": 2.3359,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 2.3066,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2612612612612613e-06,
+      "loss": 2.2695,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4414414414414416e-06,
+      "loss": 2.1719,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6216216216216219e-06,
+      "loss": 1.8145,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.801801801801802e-06,
+      "loss": 1.6426,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9819819819819822e-06,
+      "loss": 0.8184,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 0.7773,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3423423423423424e-06,
+      "loss": 0.6592,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5225225225225225e-06,
+      "loss": 0.5962,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 0.1986,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.882882882882883e-06,
+      "loss": 0.2389,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.063063063063063e-06,
+      "loss": 0.2103,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 0.1631,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.423423423423424e-06,
+      "loss": 0.1381,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.603603603603604e-06,
+      "loss": 0.1158,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 0.1486,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9639639639639645e-06,
+      "loss": 0.1462,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1441441441441446e-06,
+      "loss": 0.1148,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 0.0973,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.504504504504505e-06,
+      "loss": 0.1111,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.684684684684685e-06,
+      "loss": 0.1138,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.864864864864866e-06,
+      "loss": 0.1096,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.045045045045045e-06,
+      "loss": 0.1016,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.225225225225226e-06,
+      "loss": 0.0953,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 0.0964,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.585585585585585e-06,
+      "loss": 0.1028,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.765765765765766e-06,
+      "loss": 0.1046,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.945945945945947e-06,
+      "loss": 0.1023,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.126126126126126e-06,
+      "loss": 0.1047,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.3063063063063065e-06,
+      "loss": 0.0932,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 0.1028,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.1071,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.846846846846848e-06,
+      "loss": 0.0964,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.027027027027028e-06,
+      "loss": 0.1052,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.207207207207208e-06,
+      "loss": 0.094,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.387387387387388e-06,
+      "loss": 0.0986,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 0.1064,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.747747747747749e-06,
+      "loss": 0.1033,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.927927927927929e-06,
+      "loss": 0.0998,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 0.0945,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.288288288288289e-06,
+      "loss": 0.0931,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.46846846846847e-06,
+      "loss": 0.0885,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 0.0946,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.82882882882883e-06,
+      "loss": 0.0916,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.00900900900901e-06,
+      "loss": 0.0967,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 0.0933,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.36936936936937e-06,
+      "loss": 0.0955,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.54954954954955e-06,
+      "loss": 0.0777,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 0.0854,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.90990990990991e-06,
+      "loss": 0.0903,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.009009009009009e-05,
+      "loss": 0.0828,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.027027027027027e-05,
+      "loss": 0.0931,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0450450450450452e-05,
+      "loss": 0.0816,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0630630630630632e-05,
+      "loss": 0.0859,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 0.0842,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0990990990990992e-05,
+      "loss": 0.0761,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.117117117117117e-05,
+      "loss": 0.0867,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1351351351351352e-05,
+      "loss": 0.0856,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1531531531531532e-05,
+      "loss": 0.0887,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1711711711711713e-05,
+      "loss": 0.1025,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 0.0818,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2072072072072074e-05,
+      "loss": 0.109,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2252252252252253e-05,
+      "loss": 0.0958,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2432432432432433e-05,
+      "loss": 0.0922,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2612612612612613e-05,
+      "loss": 0.0836,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2792792792792795e-05,
+      "loss": 0.0804,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 0.085,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3153153153153155e-05,
+      "loss": 0.0807,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.0779,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 0.0862,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3693693693693695e-05,
+      "loss": 0.0772,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3873873873873875e-05,
+      "loss": 0.0878,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 0.0738,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4234234234234234e-05,
+      "loss": 0.0849,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4414414414414416e-05,
+      "loss": 0.0928,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 0.0793,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4774774774774776e-05,
+      "loss": 0.0753,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4954954954954957e-05,
+      "loss": 0.0895,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 0.0924,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5315315315315316e-05,
+      "loss": 0.0836,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5495495495495498e-05,
+      "loss": 0.0803,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5675675675675676e-05,
+      "loss": 0.0868,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5855855855855858e-05,
+      "loss": 0.0721,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6036036036036036e-05,
+      "loss": 0.0818,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 0.081,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6396396396396396e-05,
+      "loss": 0.0845,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6576576576576578e-05,
+      "loss": 0.0694,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6756756756756757e-05,
+      "loss": 0.0906,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.693693693693694e-05,
+      "loss": 0.0726,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.711711711711712e-05,
+      "loss": 0.0878,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 0.0779,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7477477477477477e-05,
+      "loss": 0.0817,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.765765765765766e-05,
+      "loss": 0.0872,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.783783783783784e-05,
+      "loss": 0.0722,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.801801801801802e-05,
+      "loss": 0.072,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.81981981981982e-05,
+      "loss": 0.0822,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 0.0721,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.855855855855856e-05,
+      "loss": 0.0809,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.873873873873874e-05,
+      "loss": 0.078,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 0.0819,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.90990990990991e-05,
+      "loss": 0.0836,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.927927927927928e-05,
+      "loss": 0.0917,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 0.0807,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.963963963963964e-05,
+      "loss": 0.0783,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981981981981982e-05,
+      "loss": 0.0913,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.0828,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999996130186627e-05,
+      "loss": 0.0815,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999984520749494e-05,
+      "loss": 0.0815,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999965171697596e-05,
+      "loss": 0.0755,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99999380830459e-05,
+      "loss": 0.0742,
+      "step": 115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999903254815378e-05,
+      "loss": 0.0665,
+      "step": 116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999986068703298e-05,
+      "loss": 0.075,
+      "step": 117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999810379731655e-05,
+      "loss": 0.0863,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999975233295034e-05,
+      "loss": 0.0713,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999968654673396e-05,
+      "loss": 0.0753,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999613021133428e-05,
+      "loss": 0.0864,
+      "step": 121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999531756205654e-05,
+      "loss": 0.0815,
+      "step": 122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999442752013535e-05,
+      "loss": 0.0645,
+      "step": 123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999934600862595e-05,
+      "loss": 0.0721,
+      "step": 124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999241526117787e-05,
+      "loss": 0.0804,
+      "step": 125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999129304569896e-05,
+      "loss": 0.0787,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999009344069144e-05,
+      "loss": 0.0859,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998881644708374e-05,
+      "loss": 0.0852,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998746206586418e-05,
+      "loss": 0.08,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99986030298081e-05,
+      "loss": 0.0739,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998452114484235e-05,
+      "loss": 0.0775,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998293460731624e-05,
+      "loss": 0.0778,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998127068673062e-05,
+      "loss": 0.0818,
+      "step": 133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999795293843733e-05,
+      "loss": 0.0729,
+      "step": 134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997771070159194e-05,
+      "loss": 0.0735,
+      "step": 135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997581463979415e-05,
+      "loss": 0.0749,
+      "step": 136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9997384120044744e-05,
+      "loss": 0.0737,
+      "step": 137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999717903850792e-05,
+      "loss": 0.0763,
+      "step": 138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999696621952766e-05,
+      "loss": 0.0633,
+      "step": 139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999674566326868e-05,
+      "loss": 0.0817,
+      "step": 140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996517369901687e-05,
+      "loss": 0.078,
+      "step": 141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999628133960337e-05,
+      "loss": 0.0687,
+      "step": 142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999603757255641e-05,
+      "loss": 0.0796,
+      "step": 143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999578606894947e-05,
+      "loss": 0.0822,
+      "step": 144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99955268289772e-05,
+      "loss": 0.0807,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999525985284025e-05,
+      "loss": 0.0678,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994985140745247e-05,
+      "loss": 0.0784,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999470269290481e-05,
+      "loss": 0.0744,
+      "step": 148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994412509537534e-05,
+      "loss": 0.0791,
+      "step": 149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994114590868023e-05,
+      "loss": 0.0775,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993808937126848e-05,
+      "loss": 0.0732,
+      "step": 151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999349554855057e-05,
+      "loss": 0.0758,
+      "step": 152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993174425381748e-05,
+      "loss": 0.0841,
+      "step": 153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992845567868912e-05,
+      "loss": 0.0685,
+      "step": 154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992508976266594e-05,
+      "loss": 0.0766,
+      "step": 155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999216465083529e-05,
+      "loss": 0.0661,
+      "step": 156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991812591841514e-05,
+      "loss": 0.0725,
+      "step": 157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991452799557732e-05,
+      "loss": 0.0738,
+      "step": 158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991085274262412e-05,
+      "loss": 0.0804,
+      "step": 159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990710016240006e-05,
+      "loss": 0.0812,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990327025780953e-05,
+      "loss": 0.0794,
+      "step": 161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998993630318167e-05,
+      "loss": 0.0824,
+      "step": 162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998953784874456e-05,
+      "loss": 0.0731,
+      "step": 163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998913166277802e-05,
+      "loss": 0.0712,
+      "step": 164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988717745596412e-05,
+      "loss": 0.0718,
+      "step": 165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988296097520098e-05,
+      "loss": 0.07,
+      "step": 166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987866718875418e-05,
+      "loss": 0.0689,
+      "step": 167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987429609994694e-05,
+      "loss": 0.0753,
+      "step": 168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998698477121623e-05,
+      "loss": 0.0685,
+      "step": 169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998653220288432e-05,
+      "loss": 0.0741,
+      "step": 170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986071905349233e-05,
+      "loss": 0.0639,
+      "step": 171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998560387896722e-05,
+      "loss": 0.0717,
+      "step": 172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9985128124100514e-05,
+      "loss": 0.0738,
+      "step": 173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998464464111734e-05,
+      "loss": 0.0769,
+      "step": 174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9984153430391883e-05,
+      "loss": 0.0747,
+      "step": 175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983654492304335e-05,
+      "loss": 0.0723,
+      "step": 176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983147827240847e-05,
+      "loss": 0.0675,
+      "step": 177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982633435593565e-05,
+      "loss": 0.079,
+      "step": 178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9982111317760598e-05,
+      "loss": 0.0707,
+      "step": 179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998158147414606e-05,
+      "loss": 0.0674,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998104390516002e-05,
+      "loss": 0.0777,
+      "step": 181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998049861121854e-05,
+      "loss": 0.0655,
+      "step": 182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979945592743656e-05,
+      "loss": 0.0732,
+      "step": 183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9979384850163385e-05,
+      "loss": 0.0768,
+      "step": 184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997881638391172e-05,
+      "loss": 0.0709,
+      "step": 185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978240194428635e-05,
+      "loss": 0.0751,
+      "step": 186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997765628216007e-05,
+      "loss": 0.0752,
+      "step": 187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977064647557966e-05,
+      "loss": 0.066,
+      "step": 188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976465291080215e-05,
+      "loss": 0.0775,
+      "step": 189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975858213190703e-05,
+      "loss": 0.0776,
+      "step": 190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975243414359277e-05,
+      "loss": 0.08,
+      "step": 191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974620895061777e-05,
+      "loss": 0.0778,
+      "step": 192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9973990655780007e-05,
+      "loss": 0.0715,
+      "step": 193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9973352697001748e-05,
+      "loss": 0.0738,
+      "step": 194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972707019220755e-05,
+      "loss": 0.068,
+      "step": 195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9972053622936764e-05,
+      "loss": 0.0811,
+      "step": 196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971392508655473e-05,
+      "loss": 0.0829,
+      "step": 197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970723676888563e-05,
+      "loss": 0.0896,
+      "step": 198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997004712815368e-05,
+      "loss": 0.0821,
+      "step": 199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9969362862974455e-05,
+      "loss": 0.0677,
+      "step": 200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9968670881880478e-05,
+      "loss": 0.0671,
+      "step": 201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996797118540732e-05,
+      "loss": 0.076,
+      "step": 202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996726377409652e-05,
+      "loss": 0.0801,
+      "step": 203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966548648495584e-05,
+      "loss": 0.0723,
+      "step": 204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965825809157997e-05,
+      "loss": 0.0739,
+      "step": 205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9965095256643207e-05,
+      "loss": 0.0747,
+      "step": 206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964356991516637e-05,
+      "loss": 0.0848,
+      "step": 207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963611014349673e-05,
+      "loss": 0.0638,
+      "step": 208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9962857325719676e-05,
+      "loss": 0.0732,
+      "step": 209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996209592620997e-05,
+      "loss": 0.0668,
+      "step": 210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9961326816409856e-05,
+      "loss": 0.0804,
+      "step": 211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996054999691459e-05,
+      "loss": 0.0833,
+      "step": 212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959765468325403e-05,
+      "loss": 0.0758,
+      "step": 213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995897323124949e-05,
+      "loss": 0.063,
+      "step": 214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958173286300017e-05,
+      "loss": 0.0897,
+      "step": 215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9957365634096112e-05,
+      "loss": 0.075,
+      "step": 216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995655027526286e-05,
+      "loss": 0.0731,
+      "step": 217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9955727210431322e-05,
+      "loss": 0.0742,
+      "step": 218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9954896440238524e-05,
+      "loss": 0.08,
+      "step": 219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9954057965327443e-05,
+      "loss": 0.07,
+      "step": 220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9953211786347035e-05,
+      "loss": 0.0787,
+      "step": 221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9952357903952207e-05,
+      "loss": 0.0686,
+      "step": 222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9951496318803834e-05,
+      "loss": 0.0816,
+      "step": 223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9950627031568744e-05,
+      "loss": 0.0755,
+      "step": 224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9949750042919745e-05,
+      "loss": 0.0674,
+      "step": 225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9948865353535583e-05,
+      "loss": 0.0745,
+      "step": 226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994797296410098e-05,
+      "loss": 0.0726,
+      "step": 227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994707287530661e-05,
+      "loss": 0.087,
+      "step": 228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994616508784911e-05,
+      "loss": 0.0858,
+      "step": 229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994524960243107e-05,
+      "loss": 0.0737,
+      "step": 230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9944326419761044e-05,
+      "loss": 0.0736,
+      "step": 231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9943395540553543e-05,
+      "loss": 0.072,
+      "step": 232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994245696552903e-05,
+      "loss": 0.0776,
+      "step": 233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9941510695413924e-05,
+      "loss": 0.0743,
+      "step": 234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.994055673094061e-05,
+      "loss": 0.0705,
+      "step": 235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9939595072847413e-05,
+      "loss": 0.0805,
+      "step": 236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993862572187863e-05,
+      "loss": 0.0821,
+      "step": 237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9937648678784494e-05,
+      "loss": 0.0722,
+      "step": 238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99366639443212e-05,
+      "loss": 0.0712,
+      "step": 239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9935671519250906e-05,
+      "loss": 0.0725,
+      "step": 240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9934671404341702e-05,
+      "loss": 0.0832,
+      "step": 241
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993366360036764e-05,
+      "loss": 0.0799,
+      "step": 242
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9932648108108727e-05,
+      "loss": 0.0651,
+      "step": 243
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9931624928350912e-05,
+      "loss": 0.0657,
+      "step": 244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9930594061886103e-05,
+      "loss": 0.07,
+      "step": 245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9929555509512147e-05,
+      "loss": 0.076,
+      "step": 246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992850927203285e-05,
+      "loss": 0.0645,
+      "step": 247
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.992745535025795e-05,
+      "loss": 0.0772,
+      "step": 248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992639374500316e-05,
+      "loss": 0.0707,
+      "step": 249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9925324457090108e-05,
+      "loss": 0.0741,
+      "step": 250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992424748734639e-05,
+      "loss": 0.0739,
+      "step": 251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992316283660554e-05,
+      "loss": 0.0657,
+      "step": 252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9922070505707033e-05,
+      "loss": 0.0662,
+      "step": 253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99209704954963e-05,
+      "loss": 0.0743,
+      "step": 254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99198628068247e-05,
+      "loss": 0.0806,
+      "step": 255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991874744054955e-05,
+      "loss": 0.0753,
+      "step": 256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9917624397534093e-05,
+      "loss": 0.0681,
+      "step": 257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991649367864753e-05,
+      "loss": 0.0813,
+      "step": 258
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991535528476499e-05,
+      "loss": 0.0709,
+      "step": 259
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9914209216767554e-05,
+      "loss": 0.0698,
+      "step": 260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9913055475542228e-05,
+      "loss": 0.0871,
+      "step": 261
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991189406198197e-05,
+      "loss": 0.0726,
+      "step": 262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9910724976985668e-05,
+      "loss": 0.0721,
+      "step": 263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9909548221458154e-05,
+      "loss": 0.0869,
+      "step": 264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9908363796310188e-05,
+      "loss": 0.0863,
+      "step": 265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9907171702458476e-05,
+      "loss": 0.0742,
+      "step": 266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9905971940825647e-05,
+      "loss": 0.0669,
+      "step": 267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9904764512340276e-05,
+      "loss": 0.0714,
+      "step": 268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990354941793687e-05,
+      "loss": 0.0689,
+      "step": 269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.990232665855586e-05,
+      "loss": 0.0789,
+      "step": 270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9901096235143625e-05,
+      "loss": 0.0809,
+      "step": 271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989985814865246e-05,
+      "loss": 0.0826,
+      "step": 272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.98986124000406e-05,
+      "loss": 0.0716,
+      "step": 273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9897358990272208e-05,
+      "loss": 0.0768,
+      "step": 274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9896097920317373e-05,
+      "loss": 0.0674,
+      "step": 275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989482919115212e-05,
+      "loss": 0.0669,
+      "step": 276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9893552803758404e-05,
+      "loss": 0.0676,
+      "step": 277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.989226875912409e-05,
+      "loss": 0.0857,
+      "step": 278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9890977058242985e-05,
+      "loss": 0.0767,
+      "step": 279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9889677702114816e-05,
+      "loss": 0.0687,
+      "step": 280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9888370691745237e-05,
+      "loss": 0.0746,
+      "step": 281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988705602814583e-05,
+      "loss": 0.0776,
+      "step": 282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9885733712334087e-05,
+      "loss": 0.0731,
+      "step": 283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9884403745333435e-05,
+      "loss": 0.0655,
+      "step": 284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9883066128173222e-05,
+      "loss": 0.0728,
+      "step": 285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.988172086188871e-05,
+      "loss": 0.0694,
+      "step": 286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9880367947521084e-05,
+      "loss": 0.0673,
+      "step": 287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987900738611745e-05,
+      "loss": 0.0708,
+      "step": 288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9877639178730837e-05,
+      "loss": 0.0724,
+      "step": 289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987626332642018e-05,
+      "loss": 0.0738,
+      "step": 290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987487983025034e-05,
+      "loss": 0.0697,
+      "step": 291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9873488691292092e-05,
+      "loss": 0.0766,
+      "step": 292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9872089910622122e-05,
+      "loss": 0.0644,
+      "step": 293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987068348932304e-05,
+      "loss": 0.0725,
+      "step": 294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9869269428483354e-05,
+      "loss": 0.0761,
+      "step": 295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9867847729197506e-05,
+      "loss": 0.0797,
+      "step": 296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986641839256583e-05,
+      "loss": 0.0743,
+      "step": 297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986498141969458e-05,
+      "loss": 0.0732,
+      "step": 298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9863536811695928e-05,
+      "loss": 0.0815,
+      "step": 299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9862084569687937e-05,
+      "loss": 0.0639,
+      "step": 300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986062469479459e-05,
+      "loss": 0.0649,
+      "step": 301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9859157188145777e-05,
+      "loss": 0.0593,
+      "step": 302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.985768205087729e-05,
+      "loss": 0.0713,
+      "step": 303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9856199284130838e-05,
+      "loss": 0.0807,
+      "step": 304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985470888905402e-05,
+      "loss": 0.0763,
+      "step": 305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9853210866800347e-05,
+      "loss": 0.0746,
+      "step": 306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9851705218529237e-05,
+      "loss": 0.0791,
+      "step": 307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9850191945406002e-05,
+      "loss": 0.0707,
+      "step": 308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984867104860186e-05,
+      "loss": 0.0788,
+      "step": 309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9847142529293924e-05,
+      "loss": 0.0627,
+      "step": 310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9845606388665217e-05,
+      "loss": 0.0789,
+      "step": 311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984406262790465e-05,
+      "loss": 0.0825,
+      "step": 312
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9842511248207042e-05,
+      "loss": 0.0819,
+      "step": 313
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9840952250773097e-05,
+      "loss": 0.0678,
+      "step": 314
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9839385636809423e-05,
+      "loss": 0.0674,
+      "step": 315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9837811407528522e-05,
+      "loss": 0.0731,
+      "step": 316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983622956414879e-05,
+      "loss": 0.0758,
+      "step": 317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9834640107894505e-05,
+      "loss": 0.0692,
+      "step": 318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983304303999586e-05,
+      "loss": 0.068,
+      "step": 319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9831438361688918e-05,
+      "loss": 0.0742,
+      "step": 320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.982982607421564e-05,
+      "loss": 0.071,
+      "step": 321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9828206178823882e-05,
+      "loss": 0.0629,
+      "step": 322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982657867676738e-05,
+      "loss": 0.0782,
+      "step": 323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9824943569305753e-05,
+      "loss": 0.0606,
+      "step": 324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9823300857704523e-05,
+      "loss": 0.0725,
+      "step": 325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9821650543235084e-05,
+      "loss": 0.0626,
+      "step": 326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981999262717472e-05,
+      "loss": 0.0704,
+      "step": 327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9818327110806587e-05,
+      "loss": 0.0689,
+      "step": 328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9816653995419744e-05,
+      "loss": 0.0648,
+      "step": 329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981497328230911e-05,
+      "loss": 0.0762,
+      "step": 330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9813284972775505e-05,
+      "loss": 0.0674,
+      "step": 331
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9811589068125605e-05,
+      "loss": 0.0683,
+      "step": 332
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980988556967199e-05,
+      "loss": 0.0708,
+      "step": 333
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9808174478733095e-05,
+      "loss": 0.0594,
+      "step": 334
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9806455796633238e-05,
+      "loss": 0.0721,
+      "step": 335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9804729524702626e-05,
+      "loss": 0.0635,
+      "step": 336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980299566427732e-05,
+      "loss": 0.0789,
+      "step": 337
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.980125421669927e-05,
+      "loss": 0.0697,
+      "step": 338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9799505183316286e-05,
+      "loss": 0.0764,
+      "step": 339
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9797748565482055e-05,
+      "loss": 0.0637,
+      "step": 340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9795984364556138e-05,
+      "loss": 0.0731,
+      "step": 341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9794212581903954e-05,
+      "loss": 0.0733,
+      "step": 342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9792433218896804e-05,
+      "loss": 0.0687,
+      "step": 343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979064627691184e-05,
+      "loss": 0.0739,
+      "step": 344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9788851757332096e-05,
+      "loss": 0.0736,
+      "step": 345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978704966154646e-05,
+      "loss": 0.0702,
+      "step": 346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9785239990949688e-05,
+      "loss": 0.0786,
+      "step": 347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9783422746942396e-05,
+      "loss": 0.0704,
+      "step": 348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978159793093106e-05,
+      "loss": 0.0695,
+      "step": 349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977976554432803e-05,
+      "loss": 0.0725,
+      "step": 350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9777925588551496e-05,
+      "loss": 0.0808,
+      "step": 351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977607806502552e-05,
+      "loss": 0.0655,
+      "step": 352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9774222975180008e-05,
+      "loss": 0.0862,
+      "step": 353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9772360320450736e-05,
+      "loss": 0.0793,
+      "step": 354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977049010227933e-05,
+      "loss": 0.0722,
+      "step": 355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.976861232211327e-05,
+      "loss": 0.0739,
+      "step": 356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9766726981405884e-05,
+      "loss": 0.0822,
+      "step": 357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9764834081616358e-05,
+      "loss": 0.0771,
+      "step": 358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9762933624209723e-05,
+      "loss": 0.0672,
+      "step": 359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9761025610656864e-05,
+      "loss": 0.0769,
+      "step": 360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9759110042434514e-05,
+      "loss": 0.0787,
+      "step": 361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975718692102525e-05,
+      "loss": 0.0779,
+      "step": 362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9755256247917494e-05,
+      "loss": 0.075,
+      "step": 363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975331802460552e-05,
+      "loss": 0.0748,
+      "step": 364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9751372252589435e-05,
+      "loss": 0.0748,
+      "step": 365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9749418933375197e-05,
+      "loss": 0.0698,
+      "step": 366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9747458068474596e-05,
+      "loss": 0.0808,
+      "step": 367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9745489659405282e-05,
+      "loss": 0.0731,
+      "step": 368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9743513707690715e-05,
+      "loss": 0.0833,
+      "step": 369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974153021486022e-05,
+      "loss": 0.0729,
+      "step": 370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9739539182448935e-05,
+      "loss": 0.0624,
+      "step": 371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9737540611997853e-05,
+      "loss": 0.0686,
+      "step": 372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9735534505053794e-05,
+      "loss": 0.0694,
+      "step": 373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.97335208631694e-05,
+      "loss": 0.0752,
+      "step": 374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9731499687903168e-05,
+      "loss": 0.067,
+      "step": 375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9729470980819402e-05,
+      "loss": 0.0741,
+      "step": 376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9727434743488248e-05,
+      "loss": 0.074,
+      "step": 377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9725390977485683e-05,
+      "loss": 0.0782,
+      "step": 378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9723339684393495e-05,
+      "loss": 0.0675,
+      "step": 379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.972128086579932e-05,
+      "loss": 0.0759,
+      "step": 380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.97192145232966e-05,
+      "loss": 0.0712,
+      "step": 381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9717140658484607e-05,
+      "loss": 0.0663,
+      "step": 382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9715059272968437e-05,
+      "loss": 0.081,
+      "step": 383
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9712970368359007e-05,
+      "loss": 0.0772,
+      "step": 384
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9710873946273043e-05,
+      "loss": 0.0686,
+      "step": 385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9708770008333105e-05,
+      "loss": 0.0627,
+      "step": 386
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970665855616756e-05,
+      "loss": 0.077,
+      "step": 387
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9704539591410592e-05,
+      "loss": 0.0701,
+      "step": 388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9702413115702202e-05,
+      "loss": 0.0719,
+      "step": 389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9700279130688206e-05,
+      "loss": 0.0832,
+      "step": 390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969813763802022e-05,
+      "loss": 0.076,
+      "step": 391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969598863935569e-05,
+      "loss": 0.0671,
+      "step": 392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9693832136357853e-05,
+      "loss": 0.076,
+      "step": 393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9691668130695763e-05,
+      "loss": 0.0667,
+      "step": 394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9689496624044284e-05,
+      "loss": 0.0741,
+      "step": 395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9687317618084075e-05,
+      "loss": 0.0751,
+      "step": 396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9685131114501613e-05,
+      "loss": 0.0634,
+      "step": 397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9682937114989163e-05,
+      "loss": 0.0699,
+      "step": 398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96807356212448e-05,
+      "loss": 0.0627,
+      "step": 399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96785266349724e-05,
+      "loss": 0.0744,
+      "step": 400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9676310157881636e-05,
+      "loss": 0.0688,
+      "step": 401
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9674086191687978e-05,
+      "loss": 0.0724,
+      "step": 402
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9671854738112694e-05,
+      "loss": 0.0571,
+      "step": 403
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9669615798882837e-05,
+      "loss": 0.0717,
+      "step": 404
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966736937573128e-05,
+      "loss": 0.0724,
+      "step": 405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966511547039665e-05,
+      "loss": 0.0712,
+      "step": 406
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96628540846234e-05,
+      "loss": 0.0731,
+      "step": 407
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9660585220161757e-05,
+      "loss": 0.0605,
+      "step": 408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9658308878767734e-05,
+      "loss": 0.065,
+      "step": 409
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9656025062203133e-05,
+      "loss": 0.0646,
+      "step": 410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965373377223555e-05,
+      "loss": 0.0729,
+      "step": 411
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9651435010638343e-05,
+      "loss": 0.0829,
+      "step": 412
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964912877919068e-05,
+      "loss": 0.0665,
+      "step": 413
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.96468150796775e-05,
+      "loss": 0.0723,
+      "step": 414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9644493913889514e-05,
+      "loss": 0.0825,
+      "step": 415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9642165283623216e-05,
+      "loss": 0.0704,
+      "step": 416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9639829190680886e-05,
+      "loss": 0.0719,
+      "step": 417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9637485636870566e-05,
+      "loss": 0.0683,
+      "step": 418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9635134624006083e-05,
+      "loss": 0.066,
+      "step": 419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9632776153907032e-05,
+      "loss": 0.0651,
+      "step": 420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9630410228398783e-05,
+      "loss": 0.0823,
+      "step": 421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9628036849312466e-05,
+      "loss": 0.0688,
+      "step": 422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9625656018484997e-05,
+      "loss": 0.0803,
+      "step": 423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9623267737759048e-05,
+      "loss": 0.079,
+      "step": 424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9620872008983056e-05,
+      "loss": 0.0824,
+      "step": 425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961846883401123e-05,
+      "loss": 0.0643,
+      "step": 426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961605821470353e-05,
+      "loss": 0.0743,
+      "step": 427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.96136401529257e-05,
+      "loss": 0.0777,
+      "step": 428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9611214650549216e-05,
+      "loss": 0.07,
+      "step": 429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960878170945133e-05,
+      "loss": 0.0598,
+      "step": 430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960634133151505e-05,
+      "loss": 0.0691,
+      "step": 431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9603893518629134e-05,
+      "loss": 0.072,
+      "step": 432
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9601438272688097e-05,
+      "loss": 0.0637,
+      "step": 433
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9598975595592214e-05,
+      "loss": 0.0637,
+      "step": 434
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9596505489247502e-05,
+      "loss": 0.0711,
+      "step": 435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9594027955565728e-05,
+      "loss": 0.0591,
+      "step": 436
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9591542996464415e-05,
+      "loss": 0.071,
+      "step": 437
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9589050613866825e-05,
+      "loss": 0.0682,
+      "step": 438
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9586550809701972e-05,
+      "loss": 0.0831,
+      "step": 439
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.958404358590461e-05,
+      "loss": 0.0734,
+      "step": 440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9581528944415236e-05,
+      "loss": 0.0696,
+      "step": 441
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957900688718009e-05,
+      "loss": 0.0752,
+      "step": 442
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9576477416151148e-05,
+      "loss": 0.0727,
+      "step": 443
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.957394053328613e-05,
+      "loss": 0.0744,
+      "step": 444
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9571396240548486e-05,
+      "loss": 0.082,
+      "step": 445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9568844539907405e-05,
+      "loss": 0.0684,
+      "step": 446
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9566285433337802e-05,
+      "loss": 0.0687,
+      "step": 447
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9563718922820336e-05,
+      "loss": 0.0692,
+      "step": 448
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9561145010341393e-05,
+      "loss": 0.0692,
+      "step": 449
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955856369789308e-05,
+      "loss": 0.0729,
+      "step": 450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9555974987473235e-05,
+      "loss": 0.0709,
+      "step": 451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9553378881085428e-05,
+      "loss": 0.0724,
+      "step": 452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9550775380738946e-05,
+      "loss": 0.0693,
+      "step": 453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95481644884488e-05,
+      "loss": 0.0704,
+      "step": 454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954554620623573e-05,
+      "loss": 0.068,
+      "step": 455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954292053612618e-05,
+      "loss": 0.0656,
+      "step": 456
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9540287480152326e-05,
+      "loss": 0.0667,
+      "step": 457
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953764704035205e-05,
+      "loss": 0.0678,
+      "step": 458
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953499921876896e-05,
+      "loss": 0.0701,
+      "step": 459
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9532344017452368e-05,
+      "loss": 0.0718,
+      "step": 460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95296814384573e-05,
+      "loss": 0.0651,
+      "step": 461
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9527011483844493e-05,
+      "loss": 0.0638,
+      "step": 462
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9524334155680392e-05,
+      "loss": 0.0837,
+      "step": 463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9521649456037155e-05,
+      "loss": 0.0694,
+      "step": 464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9518957386992633e-05,
+      "loss": 0.0711,
+      "step": 465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9516257950630383e-05,
+      "loss": 0.0802,
+      "step": 466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9513551149039677e-05,
+      "loss": 0.0668,
+      "step": 467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9510836984315475e-05,
+      "loss": 0.067,
+      "step": 468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9508115458558435e-05,
+      "loss": 0.0674,
+      "step": 469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9505386573874925e-05,
+      "loss": 0.0776,
+      "step": 470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9502650332376994e-05,
+      "loss": 0.0636,
+      "step": 471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949990673618239e-05,
+      "loss": 0.0806,
+      "step": 472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9497155787414558e-05,
+      "loss": 0.0766,
+      "step": 473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9494397488202624e-05,
+      "loss": 0.068,
+      "step": 474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9491631840681413e-05,
+      "loss": 0.0733,
+      "step": 475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9488858846991433e-05,
+      "loss": 0.0697,
+      "step": 476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9486078509278877e-05,
+      "loss": 0.0717,
+      "step": 477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9483290829695626e-05,
+      "loss": 0.0763,
+      "step": 478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948049581039923e-05,
+      "loss": 0.0624,
+      "step": 479
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947769345355294e-05,
+      "loss": 0.065,
+      "step": 480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947488376132567e-05,
+      "loss": 0.0697,
+      "step": 481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947206673589202e-05,
+      "loss": 0.0712,
+      "step": 482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946924237943226e-05,
+      "loss": 0.079,
+      "step": 483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9466410694132338e-05,
+      "loss": 0.0706,
+      "step": 484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9463571682183868e-05,
+      "loss": 0.0708,
+      "step": 485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9460725345784145e-05,
+      "loss": 0.0658,
+      "step": 486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945787168713613e-05,
+      "loss": 0.0809,
+      "step": 487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.945501070844844e-05,
+      "loss": 0.075,
+      "step": 488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9452142411935373e-05,
+      "loss": 0.0612,
+      "step": 489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944926679981688e-05,
+      "loss": 0.0602,
+      "step": 490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9446383874318573e-05,
+      "loss": 0.0625,
+      "step": 491
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9443493637671734e-05,
+      "loss": 0.0657,
+      "step": 492
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.94405960921133e-05,
+      "loss": 0.0675,
+      "step": 493
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9437691239885857e-05,
+      "loss": 0.0645,
+      "step": 494
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943477908323766e-05,
+      "loss": 0.0656,
+      "step": 495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9431859624422603e-05,
+      "loss": 0.064,
+      "step": 496
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9428932865700243e-05,
+      "loss": 0.0693,
+      "step": 497
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9425998809335778e-05,
+      "loss": 0.07,
+      "step": 498
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9423057457600062e-05,
+      "loss": 0.0705,
+      "step": 499
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9420108812769586e-05,
+      "loss": 0.0723,
+      "step": 500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9417152877126493e-05,
+      "loss": 0.0688,
+      "step": 501
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9414189652958567e-05,
+      "loss": 0.0622,
+      "step": 502
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941121914255924e-05,
+      "loss": 0.0823,
+      "step": 503
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9408241348227564e-05,
+      "loss": 0.0667,
+      "step": 504
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9405256272268253e-05,
+      "loss": 0.0946,
+      "step": 505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9402263916991632e-05,
+      "loss": 0.0762,
+      "step": 506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.939926428471368e-05,
+      "loss": 0.0692,
+      "step": 507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9396257377756e-05,
+      "loss": 0.0628,
+      "step": 508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9393243198445827e-05,
+      "loss": 0.08,
+      "step": 509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9390221749116017e-05,
+      "loss": 0.0728,
+      "step": 510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9387193032105067e-05,
+      "loss": 0.0786,
+      "step": 511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9384157049757082e-05,
+      "loss": 0.0814,
+      "step": 512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9381113804421805e-05,
+      "loss": 0.0712,
+      "step": 513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9378063298454595e-05,
+      "loss": 0.0708,
+      "step": 514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9375005534216423e-05,
+      "loss": 0.0713,
+      "step": 515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9371940514073894e-05,
+      "loss": 0.0726,
+      "step": 516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9368868240399215e-05,
+      "loss": 0.0688,
+      "step": 517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936578871557021e-05,
+      "loss": 0.0836,
+      "step": 518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9362701941970314e-05,
+      "loss": 0.0741,
+      "step": 519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935960792198858e-05,
+      "loss": 0.0617,
+      "step": 520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935650665801966e-05,
+      "loss": 0.0612,
+      "step": 521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935339815246382e-05,
+      "loss": 0.0686,
+      "step": 522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9350282407726923e-05,
+      "loss": 0.0742,
+      "step": 523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9347159426220445e-05,
+      "loss": 0.0674,
+      "step": 524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9344029210361448e-05,
+      "loss": 0.0786,
+      "step": 525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.934089176257261e-05,
+      "loss": 0.0658,
+      "step": 526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.93377470852822e-05,
+      "loss": 0.0757,
+      "step": 527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933459518092407e-05,
+      "loss": 0.0661,
+      "step": 528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.933143605193769e-05,
+      "loss": 0.081,
+      "step": 529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9328269700768095e-05,
+      "loss": 0.064,
+      "step": 530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9325096129865938e-05,
+      "loss": 0.0659,
+      "step": 531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9321915341687426e-05,
+      "loss": 0.0777,
+      "step": 532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9318727338694387e-05,
+      "loss": 0.0681,
+      "step": 533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9315532123354203e-05,
+      "loss": 0.0756,
+      "step": 534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.931232969813986e-05,
+      "loss": 0.0669,
+      "step": 535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9309120065529914e-05,
+      "loss": 0.0737,
+      "step": 536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.93059032280085e-05,
+      "loss": 0.0785,
+      "step": 537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9302679188065326e-05,
+      "loss": 0.0745,
+      "step": 538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9299447948195687e-05,
+      "loss": 0.0704,
+      "step": 539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9296209510900436e-05,
+      "loss": 0.0598,
+      "step": 540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9292963878686005e-05,
+      "loss": 0.0721,
+      "step": 541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.928971105406439e-05,
+      "loss": 0.0677,
+      "step": 542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9286451039553157e-05,
+      "loss": 0.0668,
+      "step": 543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9283183837675438e-05,
+      "loss": 0.068,
+      "step": 544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927990945095992e-05,
+      "loss": 0.0739,
+      "step": 545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9276627881940864e-05,
+      "loss": 0.0685,
+      "step": 546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9273339133158073e-05,
+      "loss": 0.0562,
+      "step": 547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.927004320715692e-05,
+      "loss": 0.0699,
+      "step": 548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9266740106488334e-05,
+      "loss": 0.0795,
+      "step": 549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.926342983370878e-05,
+      "loss": 0.0662,
+      "step": 550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9260112391380296e-05,
+      "loss": 0.0688,
+      "step": 551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9256787782070453e-05,
+      "loss": 0.0753,
+      "step": 552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9253456008352373e-05,
+      "loss": 0.0716,
+      "step": 553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9250117072804732e-05,
+      "loss": 0.0729,
+      "step": 554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9246770978011734e-05,
+      "loss": 0.0655,
+      "step": 555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9243417726563137e-05,
+      "loss": 0.0701,
+      "step": 556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.924005732105423e-05,
+      "loss": 0.0706,
+      "step": 557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9236689764085838e-05,
+      "loss": 0.0658,
+      "step": 558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.923331505826433e-05,
+      "loss": 0.0706,
+      "step": 559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.92299332062016e-05,
+      "loss": 0.0759,
+      "step": 560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9226544210515074e-05,
+      "loss": 0.0711,
+      "step": 561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9223148073827707e-05,
+      "loss": 0.0691,
+      "step": 562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921974479876799e-05,
+      "loss": 0.0692,
+      "step": 563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921633438796992e-05,
+      "loss": 0.0746,
+      "step": 564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.921291684407304e-05,
+      "loss": 0.0695,
+      "step": 565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9209492169722388e-05,
+      "loss": 0.0673,
+      "step": 566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9206060367568544e-05,
+      "loss": 0.0612,
+      "step": 567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9202621440267593e-05,
+      "loss": 0.0782,
+      "step": 568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.919917539048113e-05,
+      "loss": 0.0721,
+      "step": 569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9195722220876275e-05,
+      "loss": 0.0678,
+      "step": 570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9192261934125653e-05,
+      "loss": 0.0702,
+      "step": 571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9188794532907396e-05,
+      "loss": 0.0758,
+      "step": 572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918532001990514e-05,
+      "loss": 0.0746,
+      "step": 573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9181838397808028e-05,
+      "loss": 0.0655,
+      "step": 574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9178349669310713e-05,
+      "loss": 0.0712,
+      "step": 575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917485383711333e-05,
+      "loss": 0.065,
+      "step": 576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.917135090392153e-05,
+      "loss": 0.0686,
+      "step": 577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9167840872446452e-05,
+      "loss": 0.0679,
+      "step": 578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.916432374540473e-05,
+      "loss": 0.0747,
+      "step": 579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9160799525518484e-05,
+      "loss": 0.0714,
+      "step": 580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9157268215515332e-05,
+      "loss": 0.0725,
+      "step": 581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9153729818128373e-05,
+      "loss": 0.0657,
+      "step": 582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9150184336096203e-05,
+      "loss": 0.0676,
+      "step": 583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9146631772162884e-05,
+      "loss": 0.0628,
+      "step": 584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.914307212907797e-05,
+      "loss": 0.0679,
+      "step": 585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.91395054095965e-05,
+      "loss": 0.0717,
+      "step": 586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.913593161647897e-05,
+      "loss": 0.0733,
+      "step": 587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9132350752491368e-05,
+      "loss": 0.0823,
+      "step": 588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9128762820405143e-05,
+      "loss": 0.0735,
+      "step": 589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912516782299723e-05,
+      "loss": 0.0686,
+      "step": 590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.912156576305002e-05,
+      "loss": 0.0664,
+      "step": 591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9117956643351368e-05,
+      "loss": 0.0687,
+      "step": 592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9114340466694602e-05,
+      "loss": 0.0755,
+      "step": 593
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.911071723587851e-05,
+      "loss": 0.0645,
+      "step": 594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9107086953707332e-05,
+      "loss": 0.0746,
+      "step": 595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9103449622990776e-05,
+      "loss": 0.0666,
+      "step": 596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9099805246543996e-05,
+      "loss": 0.065,
+      "step": 597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9096153827187606e-05,
+      "loss": 0.0681,
+      "step": 598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9092495367747665e-05,
+      "loss": 0.074,
+      "step": 599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.908882987105569e-05,
+      "loss": 0.0737,
+      "step": 600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9085157339948634e-05,
+      "loss": 0.0761,
+      "step": 601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.90814777772689e-05,
+      "loss": 0.071,
+      "step": 602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907779118586433e-05,
+      "loss": 0.0695,
+      "step": 603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.907409756858821e-05,
+      "loss": 0.0677,
+      "step": 604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9070396928299265e-05,
+      "loss": 0.0836,
+      "step": 605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9066689267861652e-05,
+      "loss": 0.067,
+      "step": 606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9062974590144953e-05,
+      "loss": 0.0743,
+      "step": 607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.90592528980242e-05,
+      "loss": 0.0706,
+      "step": 608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.905552419437984e-05,
+      "loss": 0.0692,
+      "step": 609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9051788482097753e-05,
+      "loss": 0.0723,
+      "step": 610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9048045764069236e-05,
+      "loss": 0.0792,
+      "step": 611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9044296043191015e-05,
+      "loss": 0.0841,
+      "step": 612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9040539322365234e-05,
+      "loss": 0.0703,
+      "step": 613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9036775604499458e-05,
+      "loss": 0.073,
+      "step": 614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.903300489250666e-05,
+      "loss": 0.0729,
+      "step": 615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9029227189305234e-05,
+      "loss": 0.0713,
+      "step": 616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9025442497818977e-05,
+      "loss": 0.0625,
+      "step": 617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9021650820977097e-05,
+      "loss": 0.0731,
+      "step": 618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.901785216171422e-05,
+      "loss": 0.0793,
+      "step": 619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.901404652297036e-05,
+      "loss": 0.0738,
+      "step": 620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.901023390769094e-05,
+      "loss": 0.0669,
+      "step": 621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.900641431882678e-05,
+      "loss": 0.065,
+      "step": 622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9002587759334105e-05,
+      "loss": 0.0684,
+      "step": 623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8998754232174522e-05,
+      "loss": 0.0643,
+      "step": 624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8994913740315045e-05,
+      "loss": 0.068,
+      "step": 625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8991066286728065e-05,
+      "loss": 0.0651,
+      "step": 626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.898721187439137e-05,
+      "loss": 0.0656,
+      "step": 627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8983350506288132e-05,
+      "loss": 0.0695,
+      "step": 628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8979482185406907e-05,
+      "loss": 0.066,
+      "step": 629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8975606914741625e-05,
+      "loss": 0.0757,
+      "step": 630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.897172469729161e-05,
+      "loss": 0.0691,
+      "step": 631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896783553606155e-05,
+      "loss": 0.0736,
+      "step": 632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8963939434061505e-05,
+      "loss": 0.0636,
+      "step": 633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8960036394306915e-05,
+      "loss": 0.077,
+      "step": 634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8956126419818593e-05,
+      "loss": 0.0833,
+      "step": 635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.895220951362271e-05,
+      "loss": 0.0689,
+      "step": 636
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.89482856787508e-05,
+      "loss": 0.0643,
+      "step": 637
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8944354918239773e-05,
+      "loss": 0.0702,
+      "step": 638
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8940417235131886e-05,
+      "loss": 0.066,
+      "step": 639
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8936472632474756e-05,
+      "loss": 0.0695,
+      "step": 640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8932521113321366e-05,
+      "loss": 0.0664,
+      "step": 641
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8928562680730042e-05,
+      "loss": 0.0757,
+      "step": 642
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8924597337764457e-05,
+      "loss": 0.062,
+      "step": 643
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8920625087493646e-05,
+      "loss": 0.0661,
+      "step": 644
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8916645932991976e-05,
+      "loss": 0.0694,
+      "step": 645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.891265987733917e-05,
+      "loss": 0.0713,
+      "step": 646
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.890866692362028e-05,
+      "loss": 0.0686,
+      "step": 647
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8904667074925714e-05,
+      "loss": 0.0721,
+      "step": 648
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8900660334351193e-05,
+      "loss": 0.0646,
+      "step": 649
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.889664670499779e-05,
+      "loss": 0.0762,
+      "step": 650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8892626189971906e-05,
+      "loss": 0.0693,
+      "step": 651
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.888859879238527e-05,
+      "loss": 0.0629,
+      "step": 652
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8884564515354934e-05,
+      "loss": 0.0739,
+      "step": 653
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8880523362003277e-05,
+      "loss": 0.076,
+      "step": 654
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8876475335458003e-05,
+      "loss": 0.0874,
+      "step": 655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8872420438852134e-05,
+      "loss": 0.078,
+      "step": 656
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.886835867532401e-05,
+      "loss": 0.0634,
+      "step": 657
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.886429004801728e-05,
+      "loss": 0.0771,
+      "step": 658
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8860214560080913e-05,
+      "loss": 0.0654,
+      "step": 659
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8856132214669186e-05,
+      "loss": 0.0739,
+      "step": 660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8852043014941674e-05,
+      "loss": 0.0682,
+      "step": 661
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8847946964063273e-05,
+      "loss": 0.0781,
+      "step": 662
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8843844065204174e-05,
+      "loss": 0.0688,
+      "step": 663
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.883973432153986e-05,
+      "loss": 0.0797,
+      "step": 664
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.883561773625113e-05,
+      "loss": 0.0686,
+      "step": 665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.883149431252406e-05,
+      "loss": 0.0578,
+      "step": 666
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8827364053550026e-05,
+      "loss": 0.0741,
+      "step": 667
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8823226962525694e-05,
+      "loss": 0.062,
+      "step": 668
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.881908304265302e-05,
+      "loss": 0.0768,
+      "step": 669
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8814932297139242e-05,
+      "loss": 0.0721,
+      "step": 670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8810774729196887e-05,
+      "loss": 0.0706,
+      "step": 671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.880661034204375e-05,
+      "loss": 0.0742,
+      "step": 672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.880243913890292e-05,
+      "loss": 0.0573,
+      "step": 673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879826112300274e-05,
+      "loss": 0.0691,
+      "step": 674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8794076297576848e-05,
+      "loss": 0.062,
+      "step": 675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.878988466586414e-05,
+      "loss": 0.0747,
+      "step": 676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8785686231108783e-05,
+      "loss": 0.0757,
+      "step": 677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8781480996560207e-05,
+      "loss": 0.062,
+      "step": 678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8777268965473113e-05,
+      "loss": 0.0753,
+      "step": 679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.877305014110745e-05,
+      "loss": 0.0591,
+      "step": 680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8768824526728425e-05,
+      "loss": 0.0632,
+      "step": 681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876459212560652e-05,
+      "loss": 0.08,
+      "step": 682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8760352941017446e-05,
+      "loss": 0.0696,
+      "step": 683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8756106976242173e-05,
+      "loss": 0.0783,
+      "step": 684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8751854234566923e-05,
+      "loss": 0.0571,
+      "step": 685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8747594719283157e-05,
+      "loss": 0.07,
+      "step": 686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8743328433687585e-05,
+      "loss": 0.0684,
+      "step": 687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8739055381082147e-05,
+      "loss": 0.0667,
+      "step": 688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.873477556477403e-05,
+      "loss": 0.0681,
+      "step": 689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8730488988075653e-05,
+      "loss": 0.0656,
+      "step": 690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.872619565430466e-05,
+      "loss": 0.0671,
+      "step": 691
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.872189556678394e-05,
+      "loss": 0.0571,
+      "step": 692
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8717588728841594e-05,
+      "loss": 0.0753,
+      "step": 693
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8713275143810955e-05,
+      "loss": 0.0662,
+      "step": 694
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8708954815030576e-05,
+      "loss": 0.0757,
+      "step": 695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8704627745844234e-05,
+      "loss": 0.0672,
+      "step": 696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8700293939600914e-05,
+      "loss": 0.0744,
+      "step": 697
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8695953399654827e-05,
+      "loss": 0.0758,
+      "step": 698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8691606129365382e-05,
+      "loss": 0.081,
+      "step": 699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8687252132097206e-05,
+      "loss": 0.0657,
+      "step": 700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8682891411220133e-05,
+      "loss": 0.0659,
+      "step": 701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8678523970109196e-05,
+      "loss": 0.0714,
+      "step": 702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8674149812144635e-05,
+      "loss": 0.0727,
+      "step": 703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8669768940711875e-05,
+      "loss": 0.0733,
+      "step": 704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866538135920156e-05,
+      "loss": 0.0678,
+      "step": 705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8660987071009504e-05,
+      "loss": 0.0683,
+      "step": 706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8656586079536728e-05,
+      "loss": 0.0725,
+      "step": 707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.865217838818943e-05,
+      "loss": 0.072,
+      "step": 708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8647764000379003e-05,
+      "loss": 0.0777,
+      "step": 709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.864334291952202e-05,
+      "loss": 0.0642,
+      "step": 710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8638915149040227e-05,
+      "loss": 0.0674,
+      "step": 711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8634480692360558e-05,
+      "loss": 0.0632,
+      "step": 712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8630039552915115e-05,
+      "loss": 0.0646,
+      "step": 713
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.862559173414117e-05,
+      "loss": 0.0788,
+      "step": 714
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8621137239481176e-05,
+      "loss": 0.0767,
+      "step": 715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.861667607238274e-05,
+      "loss": 0.067,
+      "step": 716
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8612208236298644e-05,
+      "loss": 0.0738,
+      "step": 717
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8607733734686822e-05,
+      "loss": 0.0714,
+      "step": 718
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8603252571010372e-05,
+      "loss": 0.0785,
+      "step": 719
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.859876474873755e-05,
+      "loss": 0.0638,
+      "step": 720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.859427027134176e-05,
+      "loss": 0.0592,
+      "step": 721
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.858976914230156e-05,
+      "loss": 0.0722,
+      "step": 722
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8585261365100656e-05,
+      "loss": 0.0607,
+      "step": 723
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8580746943227903e-05,
+      "loss": 0.0738,
+      "step": 724
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.857622588017729e-05,
+      "loss": 0.0773,
+      "step": 725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8571698179447953e-05,
+      "loss": 0.0703,
+      "step": 726
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8567163844544164e-05,
+      "loss": 0.0568,
+      "step": 727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.856262287897533e-05,
+      "loss": 0.0629,
+      "step": 728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8558075286255987e-05,
+      "loss": 0.0642,
+      "step": 729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.85535210699058e-05,
+      "loss": 0.0709,
+      "step": 730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8548960233449563e-05,
+      "loss": 0.0767,
+      "step": 731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8544392780417197e-05,
+      "loss": 0.0621,
+      "step": 732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.853981871434374e-05,
+      "loss": 0.0726,
+      "step": 733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8535238038769344e-05,
+      "loss": 0.0661,
+      "step": 734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8530650757239285e-05,
+      "loss": 0.0614,
+      "step": 735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8526056873303947e-05,
+      "loss": 0.0806,
+      "step": 736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8521456390518823e-05,
+      "loss": 0.0786,
+      "step": 737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8516849312444517e-05,
+      "loss": 0.0652,
+      "step": 738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.851223564264673e-05,
+      "loss": 0.0704,
+      "step": 739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.850761538469628e-05,
+      "loss": 0.0649,
+      "step": 740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8502988542169062e-05,
+      "loss": 0.0745,
+      "step": 741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8498355118646093e-05,
+      "loss": 0.0646,
+      "step": 742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.849371511771346e-05,
+      "loss": 0.0667,
+      "step": 743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.848906854296235e-05,
+      "loss": 0.07,
+      "step": 744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8484415397989045e-05,
+      "loss": 0.0695,
+      "step": 745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8479755686394906e-05,
+      "loss": 0.0588,
+      "step": 746
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.847508941178637e-05,
+      "loss": 0.074,
+      "step": 747
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.847041657777496e-05,
+      "loss": 0.0673,
+      "step": 748
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8465737187977278e-05,
+      "loss": 0.0672,
+      "step": 749
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8461051246014995e-05,
+      "loss": 0.0813,
+      "step": 750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8456358755514852e-05,
+      "loss": 0.0682,
+      "step": 751
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.845165972010867e-05,
+      "loss": 0.0683,
+      "step": 752
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8446954143433326e-05,
+      "loss": 0.0631,
+      "step": 753
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8442242029130753e-05,
+      "loss": 0.0698,
+      "step": 754
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.843752338084796e-05,
+      "loss": 0.0804,
+      "step": 755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8432798202236996e-05,
+      "loss": 0.0736,
+      "step": 756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8428066496954977e-05,
+      "loss": 0.0638,
+      "step": 757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.842332826866407e-05,
+      "loss": 0.0733,
+      "step": 758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8418583521031477e-05,
+      "loss": 0.077,
+      "step": 759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8413832257729467e-05,
+      "loss": 0.075,
+      "step": 760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8409074482435335e-05,
+      "loss": 0.0649,
+      "step": 761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8404310198831423e-05,
+      "loss": 0.0685,
+      "step": 762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.839953941060511e-05,
+      "loss": 0.072,
+      "step": 763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.83947621214488e-05,
+      "loss": 0.0742,
+      "step": 764
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8389978335059947e-05,
+      "loss": 0.0742,
+      "step": 765
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8385188055141013e-05,
+      "loss": 0.0758,
+      "step": 766
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8380391285399505e-05,
+      "loss": 0.0654,
+      "step": 767
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.837558802954794e-05,
+      "loss": 0.0724,
+      "step": 768
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8370778291303857e-05,
+      "loss": 0.064,
+      "step": 769
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8365962074389814e-05,
+      "loss": 0.0639,
+      "step": 770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8361139382533387e-05,
+      "loss": 0.0708,
+      "step": 771
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8356310219467158e-05,
+      "loss": 0.0603,
+      "step": 772
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8351474588928718e-05,
+      "loss": 0.0669,
+      "step": 773
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.834663249466066e-05,
+      "loss": 0.071,
+      "step": 774
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8341783940410596e-05,
+      "loss": 0.0655,
+      "step": 775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8336928929931114e-05,
+      "loss": 0.0694,
+      "step": 776
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8332067466979818e-05,
+      "loss": 0.0659,
+      "step": 777
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8327199555319297e-05,
+      "loss": 0.07,
+      "step": 778
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.832232519871713e-05,
+      "loss": 0.0737,
+      "step": 779
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.831744440094589e-05,
+      "loss": 0.068,
+      "step": 780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8312557165783132e-05,
+      "loss": 0.0723,
+      "step": 781
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8307663497011398e-05,
+      "loss": 0.0677,
+      "step": 782
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.83027633984182e-05,
+      "loss": 0.0675,
+      "step": 783
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8297856873796023e-05,
+      "loss": 0.0652,
+      "step": 784
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8292943926942354e-05,
+      "loss": 0.072,
+      "step": 785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8288024561659613e-05,
+      "loss": 0.0716,
+      "step": 786
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8283098781755214e-05,
+      "loss": 0.0649,
+      "step": 787
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8278166591041523e-05,
+      "loss": 0.0716,
+      "step": 788
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8273227993335873e-05,
+      "loss": 0.0707,
+      "step": 789
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8268282992460556e-05,
+      "loss": 0.0777,
+      "step": 790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826333159224281e-05,
+      "loss": 0.0693,
+      "step": 791
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8258373796514846e-05,
+      "loss": 0.0674,
+      "step": 792
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8253409609113805e-05,
+      "loss": 0.0728,
+      "step": 793
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8248439033881785e-05,
+      "loss": 0.0683,
+      "step": 794
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.824346207466583e-05,
+      "loss": 0.0635,
+      "step": 795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.823847873531791e-05,
+      "loss": 0.0721,
+      "step": 796
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8233489019694953e-05,
+      "loss": 0.0576,
+      "step": 797
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8228492931658808e-05,
+      "loss": 0.0791,
+      "step": 798
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.822349047507626e-05,
+      "loss": 0.0684,
+      "step": 799
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.821848165381903e-05,
+      "loss": 0.0742,
+      "step": 800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8213466471763753e-05,
+      "loss": 0.0653,
+      "step": 801
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8208444932791994e-05,
+      "loss": 0.0722,
+      "step": 802
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8203417040790236e-05,
+      "loss": 0.0784,
+      "step": 803
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.819838279964988e-05,
+      "loss": 0.0713,
+      "step": 804
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8193342213267245e-05,
+      "loss": 0.0687,
+      "step": 805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.818829528554355e-05,
+      "loss": 0.0786,
+      "step": 806
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8183242020384933e-05,
+      "loss": 0.0746,
+      "step": 807
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.817818242170243e-05,
+      "loss": 0.0627,
+      "step": 808
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8173116493411983e-05,
+      "loss": 0.069,
+      "step": 809
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8168044239434435e-05,
+      "loss": 0.0702,
+      "step": 810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.816296566369551e-05,
+      "loss": 0.0811,
+      "step": 811
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8157880770125852e-05,
+      "loss": 0.067,
+      "step": 812
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8152789562660968e-05,
+      "loss": 0.0622,
+      "step": 813
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8147692045241267e-05,
+      "loss": 0.0782,
+      "step": 814
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.814258822181203e-05,
+      "loss": 0.0742,
+      "step": 815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.813747809632343e-05,
+      "loss": 0.0746,
+      "step": 816
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8132361672730524e-05,
+      "loss": 0.0682,
+      "step": 817
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8127238954993218e-05,
+      "loss": 0.0786,
+      "step": 818
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.812210994707631e-05,
+      "loss": 0.0677,
+      "step": 819
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.811697465294946e-05,
+      "loss": 0.0789,
+      "step": 820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8111833076587193e-05,
+      "loss": 0.0633,
+      "step": 821
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.81066852219689e-05,
+      "loss": 0.0717,
+      "step": 822
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8101531093078825e-05,
+      "loss": 0.066,
+      "step": 823
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8096370693906077e-05,
+      "loss": 0.0638,
+      "step": 824
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8091204028444608e-05,
+      "loss": 0.0737,
+      "step": 825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.808603110069322e-05,
+      "loss": 0.0721,
+      "step": 826
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8080851914655576e-05,
+      "loss": 0.067,
+      "step": 827
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.807566647434017e-05,
+      "loss": 0.0789,
+      "step": 828
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.807047478376033e-05,
+      "loss": 0.0664,
+      "step": 829
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.806527684693424e-05,
+      "loss": 0.0662,
+      "step": 830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8060072667884906e-05,
+      "loss": 0.0671,
+      "step": 831
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.805486225064017e-05,
+      "loss": 0.0672,
+      "step": 832
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80496455992327e-05,
+      "loss": 0.0638,
+      "step": 833
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.804442271769999e-05,
+      "loss": 0.0734,
+      "step": 834
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8039193610084353e-05,
+      "loss": 0.0638,
+      "step": 835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8033958280432923e-05,
+      "loss": 0.069,
+      "step": 836
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8028716732797655e-05,
+      "loss": 0.0633,
+      "step": 837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8023468971235303e-05,
+      "loss": 0.0632,
+      "step": 838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8018214999807443e-05,
+      "loss": 0.0804,
+      "step": 839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8012954822580457e-05,
+      "loss": 0.0667,
+      "step": 840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8007688443625518e-05,
+      "loss": 0.0665,
+      "step": 841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8002415867018612e-05,
+      "loss": 0.0686,
+      "step": 842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7997137096840513e-05,
+      "loss": 0.073,
+      "step": 843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7991852137176798e-05,
+      "loss": 0.0726,
+      "step": 844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7986560992117818e-05,
+      "loss": 0.0724,
+      "step": 845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7981263665758734e-05,
+      "loss": 0.0677,
+      "step": 846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.797596016219947e-05,
+      "loss": 0.0663,
+      "step": 847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7970650485544744e-05,
+      "loss": 0.0739,
+      "step": 848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7965334639904046e-05,
+      "loss": 0.0674,
+      "step": 849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7960012629391643e-05,
+      "loss": 0.0817,
+      "step": 850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.795468445812657e-05,
+      "loss": 0.084,
+      "step": 851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7949350130232638e-05,
+      "loss": 0.0687,
+      "step": 852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7944009649838413e-05,
+      "loss": 0.0724,
+      "step": 853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.793866302107723e-05,
+      "loss": 0.071,
+      "step": 854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7933310248087176e-05,
+      "loss": 0.0737,
+      "step": 855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7927951335011103e-05,
+      "loss": 0.0701,
+      "step": 856
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7922586285996604e-05,
+      "loss": 0.0749,
+      "step": 857
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7917215105196033e-05,
+      "loss": 0.0748,
+      "step": 858
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.791183779676648e-05,
+      "loss": 0.0718,
+      "step": 859
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7906454364869782e-05,
+      "loss": 0.0683,
+      "step": 860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.790106481367251e-05,
+      "loss": 0.0682,
+      "step": 861
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7895669147345983e-05,
+      "loss": 0.067,
+      "step": 862
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7890267370066242e-05,
+      "loss": 0.0781,
+      "step": 863
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7884859486014058e-05,
+      "loss": 0.0701,
+      "step": 864
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7879445499374935e-05,
+      "loss": 0.0676,
+      "step": 865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7874025414339092e-05,
+      "loss": 0.065,
+      "step": 866
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7868599235101483e-05,
+      "loss": 0.0629,
+      "step": 867
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7863166965861755e-05,
+      "loss": 0.0689,
+      "step": 868
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.785772861082429e-05,
+      "loss": 0.0632,
+      "step": 869
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.785228417419817e-05,
+      "loss": 0.0691,
+      "step": 870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7846833660197188e-05,
+      "loss": 0.0789,
+      "step": 871
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7841377073039833e-05,
+      "loss": 0.0618,
+      "step": 872
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.78359144169493e-05,
+      "loss": 0.0686,
+      "step": 873
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.783044569615349e-05,
+      "loss": 0.0841,
+      "step": 874
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7824970914884976e-05,
+      "loss": 0.0699,
+      "step": 875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7819490077381044e-05,
+      "loss": 0.0734,
+      "step": 876
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7814003187883652e-05,
+      "loss": 0.0644,
+      "step": 877
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7808510250639456e-05,
+      "loss": 0.0751,
+      "step": 878
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.780301126989977e-05,
+      "loss": 0.0779,
+      "step": 879
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7797506249920612e-05,
+      "loss": 0.0743,
+      "step": 880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.779199519496266e-05,
+      "loss": 0.0633,
+      "step": 881
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.778647810929126e-05,
+      "loss": 0.065,
+      "step": 882
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7780954997176438e-05,
+      "loss": 0.0696,
+      "step": 883
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7775425862892864e-05,
+      "loss": 0.0761,
+      "step": 884
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7769890710719898e-05,
+      "loss": 0.0694,
+      "step": 885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.776434954494153e-05,
+      "loss": 0.0654,
+      "step": 886
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7758802369846415e-05,
+      "loss": 0.0661,
+      "step": 887
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7753249189727867e-05,
+      "loss": 0.0663,
+      "step": 888
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7747690008883833e-05,
+      "loss": 0.0649,
+      "step": 889
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7742124831616917e-05,
+      "loss": 0.0662,
+      "step": 890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7736553662234353e-05,
+      "loss": 0.0796,
+      "step": 891
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.773097650504802e-05,
+      "loss": 0.0662,
+      "step": 892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7725393364374433e-05,
+      "loss": 0.0647,
+      "step": 893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7719804244534733e-05,
+      "loss": 0.0668,
+      "step": 894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.771420914985469e-05,
+      "loss": 0.084,
+      "step": 895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7708608084664694e-05,
+      "loss": 0.0612,
+      "step": 896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7703001053299768e-05,
+      "loss": 0.0667,
+      "step": 897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7697388060099537e-05,
+      "loss": 0.0701,
+      "step": 898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.769176910940825e-05,
+      "loss": 0.0679,
+      "step": 899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7686144205574775e-05,
+      "loss": 0.0616,
+      "step": 900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7680513352952564e-05,
+      "loss": 0.0711,
+      "step": 901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7674876555899697e-05,
+      "loss": 0.0639,
+      "step": 902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7669233818778836e-05,
+      "loss": 0.0705,
+      "step": 903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7663585145957255e-05,
+      "loss": 0.0688,
+      "step": 904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7657930541806815e-05,
+      "loss": 0.0639,
+      "step": 905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7652270010703967e-05,
+      "loss": 0.0627,
+      "step": 906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.764660355702975e-05,
+      "loss": 0.0733,
+      "step": 907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.764093118516979e-05,
+      "loss": 0.0632,
+      "step": 908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.763525289951429e-05,
+      "loss": 0.0597,
+      "step": 909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7629568704458034e-05,
+      "loss": 0.0727,
+      "step": 910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7623878604400372e-05,
+      "loss": 0.068,
+      "step": 911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7618182603745228e-05,
+      "loss": 0.067,
+      "step": 912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7612480706901104e-05,
+      "loss": 0.0658,
+      "step": 913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7606772918281045e-05,
+      "loss": 0.0698,
+      "step": 914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7601059242302666e-05,
+      "loss": 0.069,
+      "step": 915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7595339683388148e-05,
+      "loss": 0.0679,
+      "step": 916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.758961424596421e-05,
+      "loss": 0.0639,
+      "step": 917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7583882934462123e-05,
+      "loss": 0.0684,
+      "step": 918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.757814575331772e-05,
+      "loss": 0.0664,
+      "step": 919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7572402706971354e-05,
+      "loss": 0.0732,
+      "step": 920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.756665379986793e-05,
+      "loss": 0.0772,
+      "step": 921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7560899036456895e-05,
+      "loss": 0.0612,
+      "step": 922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7555138421192213e-05,
+      "loss": 0.0646,
+      "step": 923
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.754937195853239e-05,
+      "loss": 0.0677,
+      "step": 924
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.754359965294045e-05,
+      "loss": 0.0682,
+      "step": 925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7537821508883945e-05,
+      "loss": 0.071,
+      "step": 926
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7532037530834937e-05,
+      "loss": 0.0661,
+      "step": 927
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7526247723270016e-05,
+      "loss": 0.0736,
+      "step": 928
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7520452090670274e-05,
+      "loss": 0.0717,
+      "step": 929
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7514650637521316e-05,
+      "loss": 0.0804,
+      "step": 930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7508843368313244e-05,
+      "loss": 0.0724,
+      "step": 931
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7503030287540676e-05,
+      "loss": 0.0608,
+      "step": 932
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7497211399702717e-05,
+      "loss": 0.0652,
+      "step": 933
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7491386709302966e-05,
+      "loss": 0.0753,
+      "step": 934
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7485556220849516e-05,
+      "loss": 0.0711,
+      "step": 935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.747971993885495e-05,
+      "loss": 0.0648,
+      "step": 936
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7473877867836333e-05,
+      "loss": 0.0701,
+      "step": 937
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7468030012315207e-05,
+      "loss": 0.0647,
+      "step": 938
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7462176376817597e-05,
+      "loss": 0.0775,
+      "step": 939
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7456316965873995e-05,
+      "loss": 0.0727,
+      "step": 940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7450451784019372e-05,
+      "loss": 0.0632,
+      "step": 941
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.744458083579315e-05,
+      "loss": 0.0712,
+      "step": 942
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7438704125739233e-05,
+      "loss": 0.0669,
+      "step": 943
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.743282165840597e-05,
+      "loss": 0.0697,
+      "step": 944
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.742693343834618e-05,
+      "loss": 0.0717,
+      "step": 945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.742103947011711e-05,
+      "loss": 0.0727,
+      "step": 946
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7415139758280485e-05,
+      "loss": 0.0748,
+      "step": 947
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.740923430740246e-05,
+      "loss": 0.0676,
+      "step": 948
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7403323122053622e-05,
+      "loss": 0.0685,
+      "step": 949
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7397406206809024e-05,
+      "loss": 0.0773,
+      "step": 950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7391483566248127e-05,
+      "loss": 0.0708,
+      "step": 951
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7385555204954833e-05,
+      "loss": 0.0672,
+      "step": 952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.737962112751748e-05,
+      "loss": 0.0636,
+      "step": 953
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7373681338528817e-05,
+      "loss": 0.0692,
+      "step": 954
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.736773584258602e-05,
+      "loss": 0.0724,
+      "step": 955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7361784644290685e-05,
+      "loss": 0.063,
+      "step": 956
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7355827748248812e-05,
+      "loss": 0.0656,
+      "step": 957
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7349865159070815e-05,
+      "loss": 0.0672,
+      "step": 958
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.734389688137152e-05,
+      "loss": 0.0705,
+      "step": 959
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.733792291977015e-05,
+      "loss": 0.062,
+      "step": 960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7331943278890328e-05,
+      "loss": 0.064,
+      "step": 961
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.732595796336007e-05,
+      "loss": 0.0688,
+      "step": 962
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7319966977811792e-05,
+      "loss": 0.0674,
+      "step": 963
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.731397032688229e-05,
+      "loss": 0.0575,
+      "step": 964
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.730796801521275e-05,
+      "loss": 0.0619,
+      "step": 965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7301960047448734e-05,
+      "loss": 0.0673,
+      "step": 966
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7295946428240187e-05,
+      "loss": 0.0628,
+      "step": 967
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7289927162241425e-05,
+      "loss": 0.0667,
+      "step": 968
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.728390225411114e-05,
+      "loss": 0.0733,
+      "step": 969
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7277871708512374e-05,
+      "loss": 0.0611,
+      "step": 970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7271835530112557e-05,
+      "loss": 0.076,
+      "step": 971
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7265793723583463e-05,
+      "loss": 0.0648,
+      "step": 972
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.725974629360122e-05,
+      "loss": 0.07,
+      "step": 973
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7253693244846313e-05,
+      "loss": 0.0608,
+      "step": 974
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7247634582003582e-05,
+      "loss": 0.0793,
+      "step": 975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7241570309762198e-05,
+      "loss": 0.0775,
+      "step": 976
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7235500432815687e-05,
+      "loss": 0.0785,
+      "step": 977
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7229424955861906e-05,
+      "loss": 0.0766,
+      "step": 978
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.722334388360305e-05,
+      "loss": 0.0793,
+      "step": 979
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7217257220745637e-05,
+      "loss": 0.0782,
+      "step": 980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.721116497200052e-05,
+      "loss": 0.0694,
+      "step": 981
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.720506714208287e-05,
+      "loss": 0.0729,
+      "step": 982
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.719896373571218e-05,
+      "loss": 0.0747,
+      "step": 983
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7192854757612262e-05,
+      "loss": 0.0803,
+      "step": 984
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7186740212511234e-05,
+      "loss": 0.0744,
+      "step": 985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7180620105141526e-05,
+      "loss": 0.0689,
+      "step": 986
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7174494440239872e-05,
+      "loss": 0.0858,
+      "step": 987
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7168363222547307e-05,
+      "loss": 0.0706,
+      "step": 988
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.716222645680917e-05,
+      "loss": 0.0771,
+      "step": 989
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7156084147775085e-05,
+      "loss": 0.0636,
+      "step": 990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.714993630019897e-05,
+      "loss": 0.0666,
+      "step": 991
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7143782918839028e-05,
+      "loss": 0.0697,
+      "step": 992
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7137624008457752e-05,
+      "loss": 0.0667,
+      "step": 993
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7131459573821902e-05,
+      "loss": 0.0739,
+      "step": 994
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7125289619702525e-05,
+      "loss": 0.0765,
+      "step": 995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7119114150874933e-05,
+      "loss": 0.0702,
+      "step": 996
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7112933172118708e-05,
+      "loss": 0.0626,
+      "step": 997
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7106746688217698e-05,
+      "loss": 0.0784,
+      "step": 998
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7100554703960014e-05,
+      "loss": 0.0759,
+      "step": 999
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7094357224138012e-05,
+      "loss": 0.0684,
+      "step": 1000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7088154253548316e-05,
+      "loss": 0.071,
+      "step": 1001
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7081945796991796e-05,
+      "loss": 0.067,
+      "step": 1002
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7075731859273557e-05,
+      "loss": 0.0653,
+      "step": 1003
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7069512445202966e-05,
+      "loss": 0.0579,
+      "step": 1004
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.706328755959361e-05,
+      "loss": 0.0633,
+      "step": 1005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.705705720726332e-05,
+      "loss": 0.0679,
+      "step": 1006
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7050821393034154e-05,
+      "loss": 0.0783,
+      "step": 1007
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7044580121732405e-05,
+      "loss": 0.0648,
+      "step": 1008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7038333398188575e-05,
+      "loss": 0.0696,
+      "step": 1009
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7032081227237405e-05,
+      "loss": 0.0746,
+      "step": 1010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7025823613717836e-05,
+      "loss": 0.066,
+      "step": 1011
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7019560562473028e-05,
+      "loss": 0.0695,
+      "step": 1012
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.701329207835035e-05,
+      "loss": 0.0767,
+      "step": 1013
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7007018166201374e-05,
+      "loss": 0.061,
+      "step": 1014
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7000738830881874e-05,
+      "loss": 0.06,
+      "step": 1015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.699445407725182e-05,
+      "loss": 0.0644,
+      "step": 1016
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.698816391017538e-05,
+      "loss": 0.0646,
+      "step": 1017
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6981868334520907e-05,
+      "loss": 0.0697,
+      "step": 1018
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.697556735516094e-05,
+      "loss": 0.0686,
+      "step": 1019
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.69692609769722e-05,
+      "loss": 0.0729,
+      "step": 1020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6962949204835596e-05,
+      "loss": 0.0829,
+      "step": 1021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6956632043636197e-05,
+      "loss": 0.0634,
+      "step": 1022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6950309498263252e-05,
+      "loss": 0.0627,
+      "step": 1023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6943981573610177e-05,
+      "loss": 0.0677,
+      "step": 1024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.693764827457455e-05,
+      "loss": 0.0695,
+      "step": 1025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.69313096060581e-05,
+      "loss": 0.0747,
+      "step": 1026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.692496557296673e-05,
+      "loss": 0.0727,
+      "step": 1027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.691861618021048e-05,
+      "loss": 0.0682,
+      "step": 1028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6912261432703543e-05,
+      "loss": 0.0591,
+      "step": 1029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6905901335364253e-05,
+      "loss": 0.0648,
+      "step": 1030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.68995358931151e-05,
+      "loss": 0.063,
+      "step": 1031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6893165110882685e-05,
+      "loss": 0.0586,
+      "step": 1032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.688678899359776e-05,
+      "loss": 0.0773,
+      "step": 1033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6880407546195207e-05,
+      "loss": 0.0711,
+      "step": 1034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6874020773614025e-05,
+      "loss": 0.068,
+      "step": 1035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6867628680797335e-05,
+      "loss": 0.0692,
+      "step": 1036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.686123127269238e-05,
+      "loss": 0.0675,
+      "step": 1037
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6854828554250513e-05,
+      "loss": 0.0727,
+      "step": 1038
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6848420530427205e-05,
+      "loss": 0.0789,
+      "step": 1039
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6842007206182024e-05,
+      "loss": 0.0626,
+      "step": 1040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.683558858647864e-05,
+      "loss": 0.0624,
+      "step": 1041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6829164676284827e-05,
+      "loss": 0.0651,
+      "step": 1042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6822735480572457e-05,
+      "loss": 0.07,
+      "step": 1043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.681630100431748e-05,
+      "loss": 0.065,
+      "step": 1044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6809861252499945e-05,
+      "loss": 0.0665,
+      "step": 1045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.680341623010398e-05,
+      "loss": 0.0622,
+      "step": 1046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6796965942117786e-05,
+      "loss": 0.065,
+      "step": 1047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6790510393533654e-05,
+      "loss": 0.0662,
+      "step": 1048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.678404958934793e-05,
+      "loss": 0.0739,
+      "step": 1049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.677758353456104e-05,
+      "loss": 0.0686,
+      "step": 1050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6771112234177466e-05,
+      "loss": 0.0701,
+      "step": 1051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6764635693205755e-05,
+      "loss": 0.0727,
+      "step": 1052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6758153916658506e-05,
+      "loss": 0.0666,
+      "step": 1053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6751666909552376e-05,
+      "loss": 0.057,
+      "step": 1054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6745174676908063e-05,
+      "loss": 0.0702,
+      "step": 1055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.673867722375031e-05,
+      "loss": 0.0791,
+      "step": 1056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6732174555107912e-05,
+      "loss": 0.0628,
+      "step": 1057
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6725666676013683e-05,
+      "loss": 0.0734,
+      "step": 1058
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6719153591504485e-05,
+      "loss": 0.0623,
+      "step": 1059
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6712635306621196e-05,
+      "loss": 0.0692,
+      "step": 1060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6706111826408728e-05,
+      "loss": 0.0651,
+      "step": 1061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6699583155916016e-05,
+      "loss": 0.0673,
+      "step": 1062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6693049300196e-05,
+      "loss": 0.0699,
+      "step": 1063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.668651026430564e-05,
+      "loss": 0.0746,
+      "step": 1064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6679966053305914e-05,
+      "loss": 0.069,
+      "step": 1065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6673416672261792e-05,
+      "loss": 0.0604,
+      "step": 1066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6666862126242255e-05,
+      "loss": 0.065,
+      "step": 1067
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6660302420320265e-05,
+      "loss": 0.0628,
+      "step": 1068
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.66537375595728e-05,
+      "loss": 0.0631,
+      "step": 1069
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6647167549080812e-05,
+      "loss": 0.0601,
+      "step": 1070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6640592393929253e-05,
+      "loss": 0.0663,
+      "step": 1071
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6634012099207036e-05,
+      "loss": 0.0666,
+      "step": 1072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.662742667000707e-05,
+      "loss": 0.0753,
+      "step": 1073
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.662083611142623e-05,
+      "loss": 0.0665,
+      "step": 1074
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.661424042856536e-05,
+      "loss": 0.0807,
+      "step": 1075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6607639626529275e-05,
+      "loss": 0.0685,
+      "step": 1076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.660103371042675e-05,
+      "loss": 0.0681,
+      "step": 1077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6594422685370518e-05,
+      "loss": 0.0705,
+      "step": 1078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6587806556477262e-05,
+      "loss": 0.0601,
+      "step": 1079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6581185328867618e-05,
+      "loss": 0.0791,
+      "step": 1080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6574559007666177e-05,
+      "loss": 0.0713,
+      "step": 1081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6567927598001457e-05,
+      "loss": 0.0705,
+      "step": 1082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6561291105005925e-05,
+      "loss": 0.0627,
+      "step": 1083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6554649533815976e-05,
+      "loss": 0.0703,
+      "step": 1084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654800288957194e-05,
+      "loss": 0.0579,
+      "step": 1085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654135117741807e-05,
+      "loss": 0.0705,
+      "step": 1086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6534694402502548e-05,
+      "loss": 0.0723,
+      "step": 1087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6528032569977468e-05,
+      "loss": 0.0612,
+      "step": 1088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6521365684998833e-05,
+      "loss": 0.0732,
+      "step": 1089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.651469375272657e-05,
+      "loss": 0.0681,
+      "step": 1090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.65080167783245e-05,
+      "loss": 0.0724,
+      "step": 1091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.650133476696036e-05,
+      "loss": 0.0684,
+      "step": 1092
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6494647723805766e-05,
+      "loss": 0.0654,
+      "step": 1093
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.648795565403625e-05,
+      "loss": 0.0712,
+      "step": 1094
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6481258562831217e-05,
+      "loss": 0.0714,
+      "step": 1095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6474556455373972e-05,
+      "loss": 0.0648,
+      "step": 1096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6467849336851697e-05,
+      "loss": 0.0722,
+      "step": 1097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6461137212455442e-05,
+      "loss": 0.0674,
+      "step": 1098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6454420087380147e-05,
+      "loss": 0.0693,
+      "step": 1099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6447697966824616e-05,
+      "loss": 0.0641,
+      "step": 1100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.644097085599152e-05,
+      "loss": 0.0654,
+      "step": 1101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.643423876008739e-05,
+      "loss": 0.0667,
+      "step": 1102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6427501684322615e-05,
+      "loss": 0.0735,
+      "step": 1103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6420759633911446e-05,
+      "loss": 0.071,
+      "step": 1104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6414012614071973e-05,
+      "loss": 0.072,
+      "step": 1105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6407260630026137e-05,
+      "loss": 0.082,
+      "step": 1106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6400503686999728e-05,
+      "loss": 0.0632,
+      "step": 1107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6393741790222364e-05,
+      "loss": 0.0693,
+      "step": 1108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.63869749449275e-05,
+      "loss": 0.0679,
+      "step": 1109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6380203156352424e-05,
+      "loss": 0.0706,
+      "step": 1110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.637342642973824e-05,
+      "loss": 0.0686,
+      "step": 1111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6366644770329893e-05,
+      "loss": 0.0688,
+      "step": 1112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.635985818337613e-05,
+      "loss": 0.0641,
+      "step": 1113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.635306667412951e-05,
+      "loss": 0.0719,
+      "step": 1114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.634627024784641e-05,
+      "loss": 0.0765,
+      "step": 1115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6339468909787014e-05,
+      "loss": 0.0627,
+      "step": 1116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6332662665215306e-05,
+      "loss": 0.0683,
+      "step": 1117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.632585151939906e-05,
+      "loss": 0.0562,
+      "step": 1118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6319035477609847e-05,
+      "loss": 0.0725,
+      "step": 1119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6312214545123037e-05,
+      "loss": 0.0702,
+      "step": 1120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6305388727217767e-05,
+      "loss": 0.0704,
+      "step": 1121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6298558029176973e-05,
+      "loss": 0.0634,
+      "step": 1122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.629172245628736e-05,
+      "loss": 0.06,
+      "step": 1123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.62848820138394e-05,
+      "loss": 0.0719,
+      "step": 1124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.627803670712735e-05,
+      "loss": 0.0734,
+      "step": 1125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6271186541449212e-05,
+      "loss": 0.0566,
+      "step": 1126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6264331522106766e-05,
+      "loss": 0.0768,
+      "step": 1127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6257471654405534e-05,
+      "loss": 0.0692,
+      "step": 1128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.62506069436548e-05,
+      "loss": 0.0752,
+      "step": 1129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6243737395167602e-05,
+      "loss": 0.0655,
+      "step": 1130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6236863014260703e-05,
+      "loss": 0.0681,
+      "step": 1131
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.622998380625462e-05,
+      "loss": 0.0737,
+      "step": 1132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.622309977647361e-05,
+      "loss": 0.0892,
+      "step": 1133
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6216210930245643e-05,
+      "loss": 0.0589,
+      "step": 1134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6209317272902442e-05,
+      "loss": 0.0711,
+      "step": 1135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6202418809779434e-05,
+      "loss": 0.061,
+      "step": 1136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6195515546215768e-05,
+      "loss": 0.0677,
+      "step": 1137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.618860748755432e-05,
+      "loss": 0.0631,
+      "step": 1138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6181694639141668e-05,
+      "loss": 0.0771,
+      "step": 1139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6174777006328092e-05,
+      "loss": 0.0652,
+      "step": 1140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.616785459446759e-05,
+      "loss": 0.0612,
+      "step": 1141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6160927408917847e-05,
+      "loss": 0.0717,
+      "step": 1142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6153995455040243e-05,
+      "loss": 0.068,
+      "step": 1143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6147058738199854e-05,
+      "loss": 0.0686,
+      "step": 1144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.614011726376544e-05,
+      "loss": 0.0708,
+      "step": 1145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6133171037109446e-05,
+      "loss": 0.0689,
+      "step": 1146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6126220063607993e-05,
+      "loss": 0.0641,
+      "step": 1147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6119264348640867e-05,
+      "loss": 0.0701,
+      "step": 1148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6112303897591534e-05,
+      "loss": 0.0749,
+      "step": 1149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.610533871584713e-05,
+      "loss": 0.0759,
+      "step": 1150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6098368808798443e-05,
+      "loss": 0.0601,
+      "step": 1151
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6091394181839915e-05,
+      "loss": 0.0694,
+      "step": 1152
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6084414840369654e-05,
+      "loss": 0.0668,
+      "step": 1153
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.607743078978941e-05,
+      "loss": 0.066,
+      "step": 1154
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.607044203550457e-05,
+      "loss": 0.0777,
+      "step": 1155
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6063448582924175e-05,
+      "loss": 0.0722,
+      "step": 1156
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6056450437460897e-05,
+      "loss": 0.0678,
+      "step": 1157
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.604944760453103e-05,
+      "loss": 0.0736,
+      "step": 1158
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.604244008955452e-05,
+      "loss": 0.0785,
+      "step": 1159
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.603542789795491e-05,
+      "loss": 0.0638,
+      "step": 1160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6028411035159383e-05,
+      "loss": 0.0669,
+      "step": 1161
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6021389506598722e-05,
+      "loss": 0.0676,
+      "step": 1162
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.601436331770733e-05,
+      "loss": 0.0639,
+      "step": 1163
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6007332473923217e-05,
+      "loss": 0.0663,
+      "step": 1164
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6000296980687996e-05,
+      "loss": 0.0622,
+      "step": 1165
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5993256843446872e-05,
+      "loss": 0.0682,
+      "step": 1166
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5986212067648644e-05,
+      "loss": 0.0639,
+      "step": 1167
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5979162658745716e-05,
+      "loss": 0.0746,
+      "step": 1168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.597210862219406e-05,
+      "loss": 0.0675,
+      "step": 1169
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.596504996345324e-05,
+      "loss": 0.0684,
+      "step": 1170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5957986687986394e-05,
+      "loss": 0.0649,
+      "step": 1171
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5950918801260235e-05,
+      "loss": 0.0759,
+      "step": 1172
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5943846308745038e-05,
+      "loss": 0.071,
+      "step": 1173
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5936769215914653e-05,
+      "loss": 0.0609,
+      "step": 1174
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5929687528246486e-05,
+      "loss": 0.0731,
+      "step": 1175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5922601251221493e-05,
+      "loss": 0.0692,
+      "step": 1176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5915510390324194e-05,
+      "loss": 0.0744,
+      "step": 1177
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5908414951042654e-05,
+      "loss": 0.0663,
+      "step": 1178
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5901314938868467e-05,
+      "loss": 0.0661,
+      "step": 1179
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5894210359296785e-05,
+      "loss": 0.066,
+      "step": 1180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.588710121782629e-05,
+      "loss": 0.0667,
+      "step": 1181
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.587998751995919e-05,
+      "loss": 0.0684,
+      "step": 1182
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5872869271201214e-05,
+      "loss": 0.0672,
+      "step": 1183
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5865746477061626e-05,
+      "loss": 0.0699,
+      "step": 1184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5858619143053205e-05,
+      "loss": 0.0714,
+      "step": 1185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5851487274692243e-05,
+      "loss": 0.0694,
+      "step": 1186
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5844350877498534e-05,
+      "loss": 0.0732,
+      "step": 1187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.583720995699539e-05,
+      "loss": 0.0647,
+      "step": 1188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.583006451870961e-05,
+      "loss": 0.0655,
+      "step": 1189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5822914568171495e-05,
+      "loss": 0.0699,
+      "step": 1190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5815760110914854e-05,
+      "loss": 0.0643,
+      "step": 1191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5808601152476955e-05,
+      "loss": 0.0796,
+      "step": 1192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.580143769839857e-05,
+      "loss": 0.0656,
+      "step": 1193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5794269754223944e-05,
+      "loss": 0.0563,
+      "step": 1194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5787097325500798e-05,
+      "loss": 0.0643,
+      "step": 1195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.577992041778033e-05,
+      "loss": 0.0679,
+      "step": 1196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.577273903661719e-05,
+      "loss": 0.0678,
+      "step": 1197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5765553187569502e-05,
+      "loss": 0.0778,
+      "step": 1198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5758362876198847e-05,
+      "loss": 0.0649,
+      "step": 1199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5751168108070255e-05,
+      "loss": 0.0735,
+      "step": 1200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5743968888752216e-05,
+      "loss": 0.0746,
+      "step": 1201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5736765223816644e-05,
+      "loss": 0.0607,
+      "step": 1202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5729557118838917e-05,
+      "loss": 0.0656,
+      "step": 1203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5722344579397835e-05,
+      "loss": 0.0777,
+      "step": 1204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.571512761107563e-05,
+      "loss": 0.0698,
+      "step": 1205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.570790621945798e-05,
+      "loss": 0.0731,
+      "step": 1206
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.570068041013396e-05,
+      "loss": 0.0689,
+      "step": 1207
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5693450188696075e-05,
+      "loss": 0.067,
+      "step": 1208
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5686215560740256e-05,
+      "loss": 0.0744,
+      "step": 1209
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.567897653186583e-05,
+      "loss": 0.066,
+      "step": 1210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5671733107675534e-05,
+      "loss": 0.0642,
+      "step": 1211
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5664485293775517e-05,
+      "loss": 0.0708,
+      "step": 1212
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5657233095775303e-05,
+      "loss": 0.0752,
+      "step": 1213
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.564997651928783e-05,
+      "loss": 0.0684,
+      "step": 1214
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5642715569929415e-05,
+      "loss": 0.0648,
+      "step": 1215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5635450253319765e-05,
+      "loss": 0.068,
+      "step": 1216
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.562818057508196e-05,
+      "loss": 0.0746,
+      "step": 1217
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5620906540842464e-05,
+      "loss": 0.0634,
+      "step": 1218
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5613628156231106e-05,
+      "loss": 0.0676,
+      "step": 1219
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.560634542688108e-05,
+      "loss": 0.0708,
+      "step": 1220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5599058358428957e-05,
+      "loss": 0.071,
+      "step": 1221
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.559176695651464e-05,
+      "loss": 0.0714,
+      "step": 1222
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.558447122678142e-05,
+      "loss": 0.0662,
+      "step": 1223
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5577171174875903e-05,
+      "loss": 0.0688,
+      "step": 1224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.556986680644807e-05,
+      "loss": 0.0685,
+      "step": 1225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.556255812715122e-05,
+      "loss": 0.0714,
+      "step": 1226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5555245142642005e-05,
+      "loss": 0.0659,
+      "step": 1227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5547927858580396e-05,
+      "loss": 0.0666,
+      "step": 1228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5540606280629704e-05,
+      "loss": 0.0725,
+      "step": 1229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.553328041445655e-05,
+      "loss": 0.061,
+      "step": 1230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5525950265730888e-05,
+      "loss": 0.0638,
+      "step": 1231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5518615840125975e-05,
+      "loss": 0.0615,
+      "step": 1232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5511277143318383e-05,
+      "loss": 0.0636,
+      "step": 1233
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.550393418098799e-05,
+      "loss": 0.0575,
+      "step": 1234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5496586958817978e-05,
+      "loss": 0.0685,
+      "step": 1235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5489235482494815e-05,
+      "loss": 0.0826,
+      "step": 1236
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.548187975770828e-05,
+      "loss": 0.0692,
+      "step": 1237
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.547451979015142e-05,
+      "loss": 0.065,
+      "step": 1238
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5467155585520578e-05,
+      "loss": 0.0672,
+      "step": 1239
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5459787149515374e-05,
+      "loss": 0.0691,
+      "step": 1240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5452414487838705e-05,
+      "loss": 0.0651,
+      "step": 1241
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5445037606196733e-05,
+      "loss": 0.0716,
+      "step": 1242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5437656510298888e-05,
+      "loss": 0.076,
+      "step": 1243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5430271205857864e-05,
+      "loss": 0.0686,
+      "step": 1244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5422881698589613e-05,
+      "loss": 0.0828,
+      "step": 1245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5415487994213337e-05,
+      "loss": 0.0655,
+      "step": 1246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5408090098451488e-05,
+      "loss": 0.0769,
+      "step": 1247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5400688017029754e-05,
+      "loss": 0.0595,
+      "step": 1248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5393281755677078e-05,
+      "loss": 0.0663,
+      "step": 1249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5385871320125628e-05,
+      "loss": 0.0862,
+      "step": 1250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5378456716110803e-05,
+      "loss": 0.07,
+      "step": 1251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.537103794937123e-05,
+      "loss": 0.07,
+      "step": 1252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5363615025648758e-05,
+      "loss": 0.0636,
+      "step": 1253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5356187950688453e-05,
+      "loss": 0.0679,
+      "step": 1254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5348756730238597e-05,
+      "loss": 0.0706,
+      "step": 1255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.534132137005067e-05,
+      "loss": 0.0593,
+      "step": 1256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.533388187587937e-05,
+      "loss": 0.0671,
+      "step": 1257
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5326438253482582e-05,
+      "loss": 0.0694,
+      "step": 1258
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5318990508621394e-05,
+      "loss": 0.066,
+      "step": 1259
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5311538647060088e-05,
+      "loss": 0.0699,
+      "step": 1260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5304082674566118e-05,
+      "loss": 0.0722,
+      "step": 1261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5296622596910134e-05,
+      "loss": 0.0603,
+      "step": 1262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5289158419865956e-05,
+      "loss": 0.058,
+      "step": 1263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.528169014921058e-05,
+      "loss": 0.0714,
+      "step": 1264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5274217790724165e-05,
+      "loss": 0.0709,
+      "step": 1265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5266741350190044e-05,
+      "loss": 0.0682,
+      "step": 1266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5259260833394692e-05,
+      "loss": 0.0696,
+      "step": 1267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5251776246127765e-05,
+      "loss": 0.0655,
+      "step": 1268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5244287594182039e-05,
+      "loss": 0.0709,
+      "step": 1269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5236794883353462e-05,
+      "loss": 0.0686,
+      "step": 1270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.522929811944111e-05,
+      "loss": 0.0716,
+      "step": 1271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5221797308247193e-05,
+      "loss": 0.0678,
+      "step": 1272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5214292455577065e-05,
+      "loss": 0.0672,
+      "step": 1273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5206783567239198e-05,
+      "loss": 0.0729,
+      "step": 1274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5199270649045193e-05,
+      "loss": 0.0707,
+      "step": 1275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5191753706809769e-05,
+      "loss": 0.0752,
+      "step": 1276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5184232746350759e-05,
+      "loss": 0.0735,
+      "step": 1277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.51767077734891e-05,
+      "loss": 0.07,
+      "step": 1278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.516917879404885e-05,
+      "loss": 0.0663,
+      "step": 1279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5161645813857152e-05,
+      "loss": 0.0608,
+      "step": 1280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.515410883874425e-05,
+      "loss": 0.0692,
+      "step": 1281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5146567874543484e-05,
+      "loss": 0.0709,
+      "step": 1282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.513902292709128e-05,
+      "loss": 0.0794,
+      "step": 1283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5131474002227147e-05,
+      "loss": 0.0654,
+      "step": 1284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5123921105793665e-05,
+      "loss": 0.0733,
+      "step": 1285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5116364243636496e-05,
+      "loss": 0.066,
+      "step": 1286
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5108803421604375e-05,
+      "loss": 0.0747,
+      "step": 1287
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5101238645549087e-05,
+      "loss": 0.0705,
+      "step": 1288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5093669921325493e-05,
+      "loss": 0.069,
+      "step": 1289
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5086097254791501e-05,
+      "loss": 0.0619,
+      "step": 1290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5078520651808073e-05,
+      "loss": 0.0764,
+      "step": 1291
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5070940118239214e-05,
+      "loss": 0.0758,
+      "step": 1292
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5063355659951974e-05,
+      "loss": 0.0797,
+      "step": 1293
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.505576728281645e-05,
+      "loss": 0.0747,
+      "step": 1294
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5048174992705748e-05,
+      "loss": 0.0666,
+      "step": 1295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5040578795496028e-05,
+      "loss": 0.0632,
+      "step": 1296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.503297869706646e-05,
+      "loss": 0.0727,
+      "step": 1297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5025374703299238e-05,
+      "loss": 0.0752,
+      "step": 1298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5017766820079565e-05,
+      "loss": 0.0767,
+      "step": 1299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5010155053295661e-05,
+      "loss": 0.0598,
+      "step": 1300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.500253940883875e-05,
+      "loss": 0.0663,
+      "step": 1301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4994919892603057e-05,
+      "loss": 0.0564,
+      "step": 1302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.49872965104858e-05,
+      "loss": 0.0612,
+      "step": 1303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4979669268387196e-05,
+      "loss": 0.0787,
+      "step": 1304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4972038172210445e-05,
+      "loss": 0.0725,
+      "step": 1305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4964403227861727e-05,
+      "loss": 0.0667,
+      "step": 1306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4956764441250208e-05,
+      "loss": 0.0764,
+      "step": 1307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4949121818288027e-05,
+      "loss": 0.0728,
+      "step": 1308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4941475364890282e-05,
+      "loss": 0.0742,
+      "step": 1309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4933825086975043e-05,
+      "loss": 0.0662,
+      "step": 1310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4926170990463343e-05,
+      "loss": 0.0653,
+      "step": 1311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4918513081279164e-05,
+      "loss": 0.0616,
+      "step": 1312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4910851365349443e-05,
+      "loss": 0.0763,
+      "step": 1313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4903185848604067e-05,
+      "loss": 0.0611,
+      "step": 1314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4895516536975851e-05,
+      "loss": 0.0562,
+      "step": 1315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4887843436400564e-05,
+      "loss": 0.0805,
+      "step": 1316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4880166552816892e-05,
+      "loss": 0.0698,
+      "step": 1317
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.487248589216646e-05,
+      "loss": 0.0657,
+      "step": 1318
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4864801460393818e-05,
+      "loss": 0.0697,
+      "step": 1319
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.485711326344642e-05,
+      "loss": 0.0666,
+      "step": 1320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4849421307274647e-05,
+      "loss": 0.0593,
+      "step": 1321
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4841725597831789e-05,
+      "loss": 0.0667,
+      "step": 1322
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4834026141074035e-05,
+      "loss": 0.0697,
+      "step": 1323
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4826322942960476e-05,
+      "loss": 0.0658,
+      "step": 1324
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4818616009453104e-05,
+      "loss": 0.0682,
+      "step": 1325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4810905346516796e-05,
+      "loss": 0.0645,
+      "step": 1326
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4803190960119312e-05,
+      "loss": 0.068,
+      "step": 1327
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4795472856231308e-05,
+      "loss": 0.0682,
+      "step": 1328
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4787751040826304e-05,
+      "loss": 0.0589,
+      "step": 1329
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4780025519880692e-05,
+      "loss": 0.076,
+      "step": 1330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4772296299373748e-05,
+      "loss": 0.0786,
+      "step": 1331
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.476456338528759e-05,
+      "loss": 0.0721,
+      "step": 1332
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4756826783607214e-05,
+      "loss": 0.0662,
+      "step": 1333
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.474908650032045e-05,
+      "loss": 0.0822,
+      "step": 1334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4741342541417995e-05,
+      "loss": 0.073,
+      "step": 1335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4733594912893387e-05,
+      "loss": 0.0764,
+      "step": 1336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4725843620742996e-05,
+      "loss": 0.0714,
+      "step": 1337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4718088670966034e-05,
+      "loss": 0.0704,
+      "step": 1338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4710330069564545e-05,
+      "loss": 0.0732,
+      "step": 1339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4702567822543392e-05,
+      "loss": 0.0637,
+      "step": 1340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4694801935910268e-05,
+      "loss": 0.0811,
+      "step": 1341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4687032415675682e-05,
+      "loss": 0.0657,
+      "step": 1342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4679259267852948e-05,
+      "loss": 0.0757,
+      "step": 1343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4671482498458192e-05,
+      "loss": 0.068,
+      "step": 1344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4663702113510347e-05,
+      "loss": 0.0706,
+      "step": 1345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4655918119031137e-05,
+      "loss": 0.0667,
+      "step": 1346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4648130521045087e-05,
+      "loss": 0.0629,
+      "step": 1347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4640339325579505e-05,
+      "loss": 0.0724,
+      "step": 1348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4632544538664482e-05,
+      "loss": 0.0644,
+      "step": 1349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.46247461663329e-05,
+      "loss": 0.0729,
+      "step": 1350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4616944214620401e-05,
+      "loss": 0.0718,
+      "step": 1351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4609138689565407e-05,
+      "loss": 0.0657,
+      "step": 1352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4601329597209107e-05,
+      "loss": 0.0598,
+      "step": 1353
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.459351694359544e-05,
+      "loss": 0.0752,
+      "step": 1354
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.458570073477111e-05,
+      "loss": 0.0619,
+      "step": 1355
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4577880976785576e-05,
+      "loss": 0.0693,
+      "step": 1356
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4570057675691032e-05,
+      "loss": 0.0657,
+      "step": 1357
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4562230837542429e-05,
+      "loss": 0.0679,
+      "step": 1358
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4554400468397438e-05,
+      "loss": 0.0588,
+      "step": 1359
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.454656657431648e-05,
+      "loss": 0.06,
+      "step": 1360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4538729161362693e-05,
+      "loss": 0.0656,
+      "step": 1361
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4530888235601944e-05,
+      "loss": 0.0622,
+      "step": 1362
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4523043803102812e-05,
+      "loss": 0.0594,
+      "step": 1363
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4515195869936602e-05,
+      "loss": 0.0666,
+      "step": 1364
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4507344442177319e-05,
+      "loss": 0.0673,
+      "step": 1365
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4499489525901673e-05,
+      "loss": 0.071,
+      "step": 1366
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4491631127189077e-05,
+      "loss": 0.0688,
+      "step": 1367
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.448376925212164e-05,
+      "loss": 0.0856,
+      "step": 1368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4475903906784155e-05,
+      "loss": 0.0777,
+      "step": 1369
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4468035097264112e-05,
+      "loss": 0.0629,
+      "step": 1370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4460162829651672e-05,
+      "loss": 0.0628,
+      "step": 1371
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4452287110039678e-05,
+      "loss": 0.0836,
+      "step": 1372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4444407944523642e-05,
+      "loss": 0.0774,
+      "step": 1373
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4436525339201743e-05,
+      "loss": 0.0708,
+      "step": 1374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4428639300174827e-05,
+      "loss": 0.0615,
+      "step": 1375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.442074983354639e-05,
+      "loss": 0.0727,
+      "step": 1376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4412856945422587e-05,
+      "loss": 0.0784,
+      "step": 1377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4404960641912217e-05,
+      "loss": 0.0603,
+      "step": 1378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4397060929126727e-05,
+      "loss": 0.0651,
+      "step": 1379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4389157813180195e-05,
+      "loss": 0.0681,
+      "step": 1380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4381251300189342e-05,
+      "loss": 0.0733,
+      "step": 1381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4373341396273514e-05,
+      "loss": 0.0659,
+      "step": 1382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4365428107554678e-05,
+      "loss": 0.0666,
+      "step": 1383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4357511440157428e-05,
+      "loss": 0.0678,
+      "step": 1384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4349591400208964e-05,
+      "loss": 0.0809,
+      "step": 1385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.434166799383911e-05,
+      "loss": 0.0681,
+      "step": 1386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4333741227180278e-05,
+      "loss": 0.0754,
+      "step": 1387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4325811106367493e-05,
+      "loss": 0.0679,
+      "step": 1388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4317877637538373e-05,
+      "loss": 0.0645,
+      "step": 1389
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4309940826833127e-05,
+      "loss": 0.0782,
+      "step": 1390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4302000680394547e-05,
+      "loss": 0.0745,
+      "step": 1391
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4294057204368013e-05,
+      "loss": 0.0779,
+      "step": 1392
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.428611040490148e-05,
+      "loss": 0.0657,
+      "step": 1393
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4278160288145473e-05,
+      "loss": 0.0728,
+      "step": 1394
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4270206860253083e-05,
+      "loss": 0.0718,
+      "step": 1395
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4262250127379968e-05,
+      "loss": 0.0739,
+      "step": 1396
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4254290095684344e-05,
+      "loss": 0.0715,
+      "step": 1397
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.424632677132698e-05,
+      "loss": 0.071,
+      "step": 1398
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4238360160471183e-05,
+      "loss": 0.076,
+      "step": 1399
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4230390269282824e-05,
+      "loss": 0.0592,
+      "step": 1400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4222417103930293e-05,
+      "loss": 0.0684,
+      "step": 1401
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4214440670584524e-05,
+      "loss": 0.0661,
+      "step": 1402
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4206460975418984e-05,
+      "loss": 0.0732,
+      "step": 1403
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4198478024609652e-05,
+      "loss": 0.071,
+      "step": 1404
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4190491824335035e-05,
+      "loss": 0.065,
+      "step": 1405
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4182502380776157e-05,
+      "loss": 0.0626,
+      "step": 1406
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4174509700116552e-05,
+      "loss": 0.061,
+      "step": 1407
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4166513788542249e-05,
+      "loss": 0.0698,
+      "step": 1408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4158514652241788e-05,
+      "loss": 0.064,
+      "step": 1409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4150512297406204e-05,
+      "loss": 0.0609,
+      "step": 1410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.414250673022902e-05,
+      "loss": 0.0756,
+      "step": 1411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4134497956906246e-05,
+      "loss": 0.0658,
+      "step": 1412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4126485983636373e-05,
+      "loss": 0.068,
+      "step": 1413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4118470816620368e-05,
+      "loss": 0.064,
+      "step": 1414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4110452462061677e-05,
+      "loss": 0.07,
+      "step": 1415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4102430926166197e-05,
+      "loss": 0.0658,
+      "step": 1416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4094406215142306e-05,
+      "loss": 0.066,
+      "step": 1417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.408637833520083e-05,
+      "loss": 0.0586,
+      "step": 1418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4078347292555043e-05,
+      "loss": 0.0621,
+      "step": 1419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4070313093420675e-05,
+      "loss": 0.0738,
+      "step": 1420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4062275744015903e-05,
+      "loss": 0.0598,
+      "step": 1421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4054235250561323e-05,
+      "loss": 0.0624,
+      "step": 1422
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4046191619279986e-05,
+      "loss": 0.0677,
+      "step": 1423
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.403814485639736e-05,
+      "loss": 0.0576,
+      "step": 1424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4030094968141339e-05,
+      "loss": 0.0734,
+      "step": 1425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4022041960742234e-05,
+      "loss": 0.0668,
+      "step": 1426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4013985840432776e-05,
+      "loss": 0.0581,
+      "step": 1427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.40059266134481e-05,
+      "loss": 0.0627,
+      "step": 1428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3997864286025744e-05,
+      "loss": 0.0717,
+      "step": 1429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.398979886440565e-05,
+      "loss": 0.0766,
+      "step": 1430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3981730354830154e-05,
+      "loss": 0.0695,
+      "step": 1431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3973658763543988e-05,
+      "loss": 0.0789,
+      "step": 1432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3965584096794246e-05,
+      "loss": 0.0768,
+      "step": 1433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3957506360830433e-05,
+      "loss": 0.0742,
+      "step": 1434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3949425561904412e-05,
+      "loss": 0.0704,
+      "step": 1435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3941341706270417e-05,
+      "loss": 0.0724,
+      "step": 1436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.393325480018505e-05,
+      "loss": 0.0618,
+      "step": 1437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3925164849907277e-05,
+      "loss": 0.0656,
+      "step": 1438
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.391707186169842e-05,
+      "loss": 0.0665,
+      "step": 1439
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3908975841822141e-05,
+      "loss": 0.0685,
+      "step": 1440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.390087679654447e-05,
+      "loss": 0.0861,
+      "step": 1441
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3892774732133754e-05,
+      "loss": 0.0745,
+      "step": 1442
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3884669654860692e-05,
+      "loss": 0.0659,
+      "step": 1443
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3876561570998312e-05,
+      "loss": 0.0767,
+      "step": 1444
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3868450486821972e-05,
+      "loss": 0.0696,
+      "step": 1445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3860336408609345e-05,
+      "loss": 0.0657,
+      "step": 1446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.385221934264042e-05,
+      "loss": 0.0601,
+      "step": 1447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.384409929519751e-05,
+      "loss": 0.0722,
+      "step": 1448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3835976272565228e-05,
+      "loss": 0.0665,
+      "step": 1449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3827850281030489e-05,
+      "loss": 0.0774,
+      "step": 1450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3819721326882505e-05,
+      "loss": 0.0665,
+      "step": 1451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3811589416412784e-05,
+      "loss": 0.065,
+      "step": 1452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3803454555915126e-05,
+      "loss": 0.0779,
+      "step": 1453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3795316751685601e-05,
+      "loss": 0.0759,
+      "step": 1454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3787176010022573e-05,
+      "loss": 0.0636,
+      "step": 1455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.377903233722667e-05,
+      "loss": 0.0688,
+      "step": 1456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3770885739600788e-05,
+      "loss": 0.0797,
+      "step": 1457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3762736223450095e-05,
+      "loss": 0.07,
+      "step": 1458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3754583795082007e-05,
+      "loss": 0.0663,
+      "step": 1459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.37464284608062e-05,
+      "loss": 0.0598,
+      "step": 1460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3738270226934602e-05,
+      "loss": 0.0688,
+      "step": 1461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3730109099781376e-05,
+      "loss": 0.072,
+      "step": 1462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3721945085662935e-05,
+      "loss": 0.0812,
+      "step": 1463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3713778190897922e-05,
+      "loss": 0.078,
+      "step": 1464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3705608421807202e-05,
+      "loss": 0.063,
+      "step": 1465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3697435784713875e-05,
+      "loss": 0.0652,
+      "step": 1466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.368926028594326e-05,
+      "loss": 0.0653,
+      "step": 1467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3681081931822883e-05,
+      "loss": 0.0621,
+      "step": 1468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3672900728682488e-05,
+      "loss": 0.073,
+      "step": 1469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.366471668285402e-05,
+      "loss": 0.0666,
+      "step": 1470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3656529800671624e-05,
+      "loss": 0.0684,
+      "step": 1471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3648340088471643e-05,
+      "loss": 0.0668,
+      "step": 1472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3640147552592606e-05,
+      "loss": 0.0614,
+      "step": 1473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3631952199375233e-05,
+      "loss": 0.0585,
+      "step": 1474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3623754035162422e-05,
+      "loss": 0.0692,
+      "step": 1475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3615553066299241e-05,
+      "loss": 0.0605,
+      "step": 1476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.360734929913294e-05,
+      "loss": 0.0607,
+      "step": 1477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3599142740012924e-05,
+      "loss": 0.0696,
+      "step": 1478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3590933395290763e-05,
+      "loss": 0.0662,
+      "step": 1479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.358272127132019e-05,
+      "loss": 0.0731,
+      "step": 1480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3574506374457074e-05,
+      "loss": 0.0581,
+      "step": 1481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3566288711059446e-05,
+      "loss": 0.0658,
+      "step": 1482
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3558068287487465e-05,
+      "loss": 0.0697,
+      "step": 1483
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3549845110103434e-05,
+      "loss": 0.0603,
+      "step": 1484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3541619185271785e-05,
+      "loss": 0.0716,
+      "step": 1485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3533390519359083e-05,
+      "loss": 0.0634,
+      "step": 1486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3525159118733995e-05,
+      "loss": 0.0703,
+      "step": 1487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.351692498976733e-05,
+      "loss": 0.0638,
+      "step": 1488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3508688138831992e-05,
+      "loss": 0.0612,
+      "step": 1489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3500448572302993e-05,
+      "loss": 0.0655,
+      "step": 1490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3492206296557457e-05,
+      "loss": 0.0597,
+      "step": 1491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3483961317974593e-05,
+      "loss": 0.0601,
+      "step": 1492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3475713642935706e-05,
+      "loss": 0.0705,
+      "step": 1493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3467463277824191e-05,
+      "loss": 0.0606,
+      "step": 1494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3459210229025524e-05,
+      "loss": 0.0679,
+      "step": 1495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3450954502927252e-05,
+      "loss": 0.0663,
+      "step": 1496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3442696105919002e-05,
+      "loss": 0.0621,
+      "step": 1497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3434435044392462e-05,
+      "loss": 0.0639,
+      "step": 1498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3426171324741393e-05,
+      "loss": 0.0784,
+      "step": 1499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3417904953361598e-05,
+      "loss": 0.0613,
+      "step": 1500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3409635936650939e-05,
+      "loss": 0.0621,
+      "step": 1501
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3401364281009335e-05,
+      "loss": 0.0726,
+      "step": 1502
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3393089992838733e-05,
+      "loss": 0.0684,
+      "step": 1503
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.338481307854312e-05,
+      "loss": 0.0609,
+      "step": 1504
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3376533544528523e-05,
+      "loss": 0.0712,
+      "step": 1505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3368251397202993e-05,
+      "loss": 0.071,
+      "step": 1506
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3359966642976601e-05,
+      "loss": 0.0687,
+      "step": 1507
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3351679288261435e-05,
+      "loss": 0.068,
+      "step": 1508
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3343389339471606e-05,
+      "loss": 0.063,
+      "step": 1509
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3335096803023217e-05,
+      "loss": 0.0668,
+      "step": 1510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3326801685334387e-05,
+      "loss": 0.067,
+      "step": 1511
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3318503992825223e-05,
+      "loss": 0.0621,
+      "step": 1512
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3310203731917834e-05,
+      "loss": 0.0681,
+      "step": 1513
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.330190090903631e-05,
+      "loss": 0.0778,
+      "step": 1514
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3293595530606723e-05,
+      "loss": 0.0661,
+      "step": 1515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3285287603057129e-05,
+      "loss": 0.0576,
+      "step": 1516
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3276977132817557e-05,
+      "loss": 0.0652,
+      "step": 1517
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3268664126319992e-05,
+      "loss": 0.0819,
+      "step": 1518
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3260348589998398e-05,
+      "loss": 0.073,
+      "step": 1519
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3252030530288692e-05,
+      "loss": 0.0712,
+      "step": 1520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3243709953628731e-05,
+      "loss": 0.0769,
+      "step": 1521
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3235386866458337e-05,
+      "loss": 0.0645,
+      "step": 1522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3227061275219276e-05,
+      "loss": 0.0633,
+      "step": 1523
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3218733186355232e-05,
+      "loss": 0.069,
+      "step": 1524
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3210402606311845e-05,
+      "loss": 0.0701,
+      "step": 1525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3202069541536666e-05,
+      "loss": 0.0753,
+      "step": 1526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.319373399847918e-05,
+      "loss": 0.0668,
+      "step": 1527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3185395983590785e-05,
+      "loss": 0.0756,
+      "step": 1528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3177055503324792e-05,
+      "loss": 0.0711,
+      "step": 1529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3168712564136426e-05,
+      "loss": 0.0744,
+      "step": 1530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3160367172482806e-05,
+      "loss": 0.0657,
+      "step": 1531
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3152019334822955e-05,
+      "loss": 0.077,
+      "step": 1532
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3143669057617784e-05,
+      "loss": 0.0745,
+      "step": 1533
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3135316347330107e-05,
+      "loss": 0.0847,
+      "step": 1534
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3126961210424599e-05,
+      "loss": 0.0731,
+      "step": 1535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3118603653367825e-05,
+      "loss": 0.0721,
+      "step": 1536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.311024368262823e-05,
+      "loss": 0.0683,
+      "step": 1537
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3101881304676112e-05,
+      "loss": 0.0732,
+      "step": 1538
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3093516525983643e-05,
+      "loss": 0.0737,
+      "step": 1539
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3085149353024847e-05,
+      "loss": 0.0648,
+      "step": 1540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3076779792275607e-05,
+      "loss": 0.0634,
+      "step": 1541
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3068407850213647e-05,
+      "loss": 0.0644,
+      "step": 1542
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3060033533318538e-05,
+      "loss": 0.075,
+      "step": 1543
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.305165684807169e-05,
+      "loss": 0.0608,
+      "step": 1544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3043277800956347e-05,
+      "loss": 0.0717,
+      "step": 1545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3034896398457573e-05,
+      "loss": 0.061,
+      "step": 1546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3026512647062264e-05,
+      "loss": 0.0654,
+      "step": 1547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3018126553259135e-05,
+      "loss": 0.0765,
+      "step": 1548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3009738123538702e-05,
+      "loss": 0.0636,
+      "step": 1549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.30013473643933e-05,
+      "loss": 0.0695,
+      "step": 1550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2992954282317062e-05,
+      "loss": 0.0645,
+      "step": 1551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2984558883805925e-05,
+      "loss": 0.0657,
+      "step": 1552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2976161175357607e-05,
+      "loss": 0.0681,
+      "step": 1553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2967761163471626e-05,
+      "loss": 0.0654,
+      "step": 1554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2959358854649278e-05,
+      "loss": 0.0662,
+      "step": 1555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2950954255393631e-05,
+      "loss": 0.0666,
+      "step": 1556
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2942547372209535e-05,
+      "loss": 0.0655,
+      "step": 1557
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2934138211603607e-05,
+      "loss": 0.068,
+      "step": 1558
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2925726780084219e-05,
+      "loss": 0.0594,
+      "step": 1559
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.29173130841615e-05,
+      "loss": 0.0824,
+      "step": 1560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2908897130347345e-05,
+      "loss": 0.0692,
+      "step": 1561
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.290047892515539e-05,
+      "loss": 0.0657,
+      "step": 1562
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2892058475101004e-05,
+      "loss": 0.0723,
+      "step": 1563
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2883635786701303e-05,
+      "loss": 0.0701,
+      "step": 1564
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2875210866475136e-05,
+      "loss": 0.0662,
+      "step": 1565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2866783720943078e-05,
+      "loss": 0.071,
+      "step": 1566
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2858354356627416e-05,
+      "loss": 0.0599,
+      "step": 1567
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2849922780052176e-05,
+      "loss": 0.0633,
+      "step": 1568
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2841488997743076e-05,
+      "loss": 0.0784,
+      "step": 1569
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2833053016227548e-05,
+      "loss": 0.0707,
+      "step": 1570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2824614842034732e-05,
+      "loss": 0.0614,
+      "step": 1571
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2816174481695457e-05,
+      "loss": 0.064,
+      "step": 1572
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2807731941742245e-05,
+      "loss": 0.0634,
+      "step": 1573
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.279928722870931e-05,
+      "loss": 0.0641,
+      "step": 1574
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2790840349132541e-05,
+      "loss": 0.0681,
+      "step": 1575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.278239130954951e-05,
+      "loss": 0.0647,
+      "step": 1576
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.277394011649946e-05,
+      "loss": 0.0639,
+      "step": 1577
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2765486776523297e-05,
+      "loss": 0.0642,
+      "step": 1578
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2757031296163591e-05,
+      "loss": 0.0577,
+      "step": 1579
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2748573681964567e-05,
+      "loss": 0.054,
+      "step": 1580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2740113940472104e-05,
+      "loss": 0.075,
+      "step": 1581
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2731652078233726e-05,
+      "loss": 0.0708,
+      "step": 1582
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2723188101798596e-05,
+      "loss": 0.0639,
+      "step": 1583
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2714722017717521e-05,
+      "loss": 0.0743,
+      "step": 1584
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2706253832542928e-05,
+      "loss": 0.0611,
+      "step": 1585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.269778355282888e-05,
+      "loss": 0.0668,
+      "step": 1586
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2689311185131059e-05,
+      "loss": 0.0716,
+      "step": 1587
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.268083673600676e-05,
+      "loss": 0.0786,
+      "step": 1588
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2672360212014884e-05,
+      "loss": 0.0758,
+      "step": 1589
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2663881619715954e-05,
+      "loss": 0.0685,
+      "step": 1590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2655400965672083e-05,
+      "loss": 0.0634,
+      "step": 1591
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.264691825644697e-05,
+      "loss": 0.0713,
+      "step": 1592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2638433498605927e-05,
+      "loss": 0.0634,
+      "step": 1593
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2629946698715833e-05,
+      "loss": 0.0668,
+      "step": 1594
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2621457863345159e-05,
+      "loss": 0.0713,
+      "step": 1595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2612966999063945e-05,
+      "loss": 0.0611,
+      "step": 1596
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2604474112443802e-05,
+      "loss": 0.0744,
+      "step": 1597
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2595979210057905e-05,
+      "loss": 0.0681,
+      "step": 1598
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2587482298480997e-05,
+      "loss": 0.0745,
+      "step": 1599
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2578983384289364e-05,
+      "loss": 0.074,
+      "step": 1600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2570482474060856e-05,
+      "loss": 0.0756,
+      "step": 1601
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2561979574374858e-05,
+      "loss": 0.0674,
+      "step": 1602
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2553474691812291e-05,
+      "loss": 0.0665,
+      "step": 1603
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2544967832955623e-05,
+      "loss": 0.0753,
+      "step": 1604
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2536459004388844e-05,
+      "loss": 0.0668,
+      "step": 1605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2527948212697468e-05,
+      "loss": 0.0577,
+      "step": 1606
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.251943546446853e-05,
+      "loss": 0.0678,
+      "step": 1607
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2510920766290582e-05,
+      "loss": 0.0714,
+      "step": 1608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2502404124753677e-05,
+      "loss": 0.0751,
+      "step": 1609
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2493885546449387e-05,
+      "loss": 0.0671,
+      "step": 1610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2485365037970762e-05,
+      "loss": 0.0637,
+      "step": 1611
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.247684260591237e-05,
+      "loss": 0.0638,
+      "step": 1612
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2468318256870246e-05,
+      "loss": 0.0758,
+      "step": 1613
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2459791997441922e-05,
+      "loss": 0.0558,
+      "step": 1614
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2451263834226401e-05,
+      "loss": 0.0559,
+      "step": 1615
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.244273377382417e-05,
+      "loss": 0.0609,
+      "step": 1616
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2434201822837168e-05,
+      "loss": 0.0694,
+      "step": 1617
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2425667987868817e-05,
+      "loss": 0.0594,
+      "step": 1618
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2417132275523978e-05,
+      "loss": 0.0667,
+      "step": 1619
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2408594692408977e-05,
+      "loss": 0.0706,
+      "step": 1620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2400055245131584e-05,
+      "loss": 0.0643,
+      "step": 1621
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.239151394030102e-05,
+      "loss": 0.0635,
+      "step": 1622
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.238297078452792e-05,
+      "loss": 0.059,
+      "step": 1623
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2374425784424385e-05,
+      "loss": 0.0766,
+      "step": 1624
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2365878946603911e-05,
+      "loss": 0.067,
+      "step": 1625
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2357330277681444e-05,
+      "loss": 0.0652,
+      "step": 1626
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2348779784273329e-05,
+      "loss": 0.0754,
+      "step": 1627
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2340227472997326e-05,
+      "loss": 0.0769,
+      "step": 1628
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.233167335047261e-05,
+      "loss": 0.0774,
+      "step": 1629
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.232311742331975e-05,
+      "loss": 0.079,
+      "step": 1630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2314559698160713e-05,
+      "loss": 0.0699,
+      "step": 1631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2306000181618862e-05,
+      "loss": 0.072,
+      "step": 1632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2297438880318944e-05,
+      "loss": 0.0664,
+      "step": 1633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2288875800887082e-05,
+      "loss": 0.0649,
+      "step": 1634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.228031094995078e-05,
+      "loss": 0.0628,
+      "step": 1635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.227174433413892e-05,
+      "loss": 0.0649,
+      "step": 1636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2263175960081735e-05,
+      "loss": 0.0633,
+      "step": 1637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2254605834410831e-05,
+      "loss": 0.0748,
+      "step": 1638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2246033963759162e-05,
+      "loss": 0.0682,
+      "step": 1639
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2237460354761042e-05,
+      "loss": 0.0643,
+      "step": 1640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2228885014052119e-05,
+      "loss": 0.0648,
+      "step": 1641
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2220307948269388e-05,
+      "loss": 0.0657,
+      "step": 1642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2211729164051178e-05,
+      "loss": 0.0675,
+      "step": 1643
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2203148668037148e-05,
+      "loss": 0.0693,
+      "step": 1644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2194566466868278e-05,
+      "loss": 0.0689,
+      "step": 1645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.218598256718688e-05,
+      "loss": 0.0605,
+      "step": 1646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2177396975636566e-05,
+      "loss": 0.0874,
+      "step": 1647
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.216880969886226e-05,
+      "loss": 0.0736,
+      "step": 1648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.21602207435102e-05,
+      "loss": 0.0727,
+      "step": 1649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2151630116227917e-05,
+      "loss": 0.0657,
+      "step": 1650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.214303782366423e-05,
+      "loss": 0.0721,
+      "step": 1651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2134443872469256e-05,
+      "loss": 0.0729,
+      "step": 1652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2125848269294398e-05,
+      "loss": 0.067,
+      "step": 1653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2117251020792322e-05,
+      "loss": 0.0719,
+      "step": 1654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2108652133616984e-05,
+      "loss": 0.079,
+      "step": 1655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.21000516144236e-05,
+      "loss": 0.0584,
+      "step": 1656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.209144946986865e-05,
+      "loss": 0.0657,
+      "step": 1657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2082845706609876e-05,
+      "loss": 0.0723,
+      "step": 1658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2074240331306264e-05,
+      "loss": 0.0673,
+      "step": 1659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2065633350618056e-05,
+      "loss": 0.0701,
+      "step": 1660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.205702477120674e-05,
+      "loss": 0.0639,
+      "step": 1661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2048414599735024e-05,
+      "loss": 0.0659,
+      "step": 1662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2039802842866863e-05,
+      "loss": 0.0627,
+      "step": 1663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2031189507267442e-05,
+      "loss": 0.0692,
+      "step": 1664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2022574599603156e-05,
+      "loss": 0.0706,
+      "step": 1665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.201395812654162e-05,
+      "loss": 0.0629,
+      "step": 1666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2005340094751668e-05,
+      "loss": 0.068,
+      "step": 1667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1996720510903331e-05,
+      "loss": 0.0674,
+      "step": 1668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1988099381667844e-05,
+      "loss": 0.064,
+      "step": 1669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1979476713717644e-05,
+      "loss": 0.0667,
+      "step": 1670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.197085251372635e-05,
+      "loss": 0.083,
+      "step": 1671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1962226788368773e-05,
+      "loss": 0.0645,
+      "step": 1672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1953599544320901e-05,
+      "loss": 0.0658,
+      "step": 1673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1944970788259897e-05,
+      "loss": 0.0621,
+      "step": 1674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1936340526864108e-05,
+      "loss": 0.0651,
+      "step": 1675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1927708766813016e-05,
+      "loss": 0.068,
+      "step": 1676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1919075514787293e-05,
+      "loss": 0.0623,
+      "step": 1677
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.191044077746875e-05,
+      "loss": 0.0698,
+      "step": 1678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1901804561540356e-05,
+      "loss": 0.0613,
+      "step": 1679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.189316687368621e-05,
+      "loss": 0.0526,
+      "step": 1680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1884527720591572e-05,
+      "loss": 0.0597,
+      "step": 1681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1875887108942817e-05,
+      "loss": 0.0722,
+      "step": 1682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1867245045427453e-05,
+      "loss": 0.0741,
+      "step": 1683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1858601536734124e-05,
+      "loss": 0.0668,
+      "step": 1684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1849956589552576e-05,
+      "loss": 0.0666,
+      "step": 1685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1841310210573675e-05,
+      "loss": 0.0742,
+      "step": 1686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.18326624064894e-05,
+      "loss": 0.0642,
+      "step": 1687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1824013183992827e-05,
+      "loss": 0.0685,
+      "step": 1688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1815362549778134e-05,
+      "loss": 0.0719,
+      "step": 1689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1806710510540578e-05,
+      "loss": 0.0816,
+      "step": 1690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1798057072976526e-05,
+      "loss": 0.0676,
+      "step": 1691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1789402243783412e-05,
+      "loss": 0.0689,
+      "step": 1692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1780746029659751e-05,
+      "loss": 0.0642,
+      "step": 1693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1772088437305129e-05,
+      "loss": 0.0675,
+      "step": 1694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.17634294734202e-05,
+      "loss": 0.0688,
+      "step": 1695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.175476914470668e-05,
+      "loss": 0.0653,
+      "step": 1696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1746107457867332e-05,
+      "loss": 0.0573,
+      "step": 1697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1737444419605993e-05,
+      "loss": 0.0598,
+      "step": 1698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1728780036627518e-05,
+      "loss": 0.0682,
+      "step": 1699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1720114315637825e-05,
+      "loss": 0.0789,
+      "step": 1700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1711447263343854e-05,
+      "loss": 0.0646,
+      "step": 1701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1702778886453578e-05,
+      "loss": 0.0622,
+      "step": 1702
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1694109191676006e-05,
+      "loss": 0.0724,
+      "step": 1703
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1685438185721148e-05,
+      "loss": 0.0677,
+      "step": 1704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.167676587530004e-05,
+      "loss": 0.0744,
+      "step": 1705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1668092267124736e-05,
+      "loss": 0.068,
+      "step": 1706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1659417367908277e-05,
+      "loss": 0.06,
+      "step": 1707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1650741184364709e-05,
+      "loss": 0.0643,
+      "step": 1708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1642063723209082e-05,
+      "loss": 0.0748,
+      "step": 1709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1633384991157423e-05,
+      "loss": 0.0786,
+      "step": 1710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1624704994926745e-05,
+      "loss": 0.0676,
+      "step": 1711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1616023741235046e-05,
+      "loss": 0.0652,
+      "step": 1712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1607341236801286e-05,
+      "loss": 0.0624,
+      "step": 1713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1598657488345406e-05,
+      "loss": 0.0773,
+      "step": 1714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1589972502588298e-05,
+      "loss": 0.0689,
+      "step": 1715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1581286286251819e-05,
+      "loss": 0.0602,
+      "step": 1716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1572598846058776e-05,
+      "loss": 0.0717,
+      "step": 1717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1563910188732924e-05,
+      "loss": 0.071,
+      "step": 1718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1555220320998955e-05,
+      "loss": 0.0706,
+      "step": 1719
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.154652924958251e-05,
+      "loss": 0.0676,
+      "step": 1720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1537836981210149e-05,
+      "loss": 0.0583,
+      "step": 1721
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1529143522609363e-05,
+      "loss": 0.0627,
+      "step": 1722
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1520448880508567e-05,
+      "loss": 0.0631,
+      "step": 1723
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.151175306163709e-05,
+      "loss": 0.0631,
+      "step": 1724
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1503056072725164e-05,
+      "loss": 0.0658,
+      "step": 1725
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1494357920503945e-05,
+      "loss": 0.0623,
+      "step": 1726
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1485658611705467e-05,
+      "loss": 0.0698,
+      "step": 1727
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1476958153062681e-05,
+      "loss": 0.0676,
+      "step": 1728
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1468256551309411e-05,
+      "loss": 0.0587,
+      "step": 1729
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.145955381318037e-05,
+      "loss": 0.072,
+      "step": 1730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1450849945411158e-05,
+      "loss": 0.067,
+      "step": 1731
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1442144954738242e-05,
+      "loss": 0.0755,
+      "step": 1732
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1433438847898954e-05,
+      "loss": 0.0632,
+      "step": 1733
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1424731631631506e-05,
+      "loss": 0.0577,
+      "step": 1734
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.141602331267495e-05,
+      "loss": 0.065,
+      "step": 1735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1407313897769203e-05,
+      "loss": 0.0662,
+      "step": 1736
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.139860339365503e-05,
+      "loss": 0.0665,
+      "step": 1737
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1389891807074032e-05,
+      "loss": 0.0644,
+      "step": 1738
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1381179144768652e-05,
+      "loss": 0.0641,
+      "step": 1739
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1372465413482168e-05,
+      "loss": 0.0594,
+      "step": 1740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1363750619958677e-05,
+      "loss": 0.0697,
+      "step": 1741
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1355034770943112e-05,
+      "loss": 0.0677,
+      "step": 1742
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1346317873181212e-05,
+      "loss": 0.064,
+      "step": 1743
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1337599933419525e-05,
+      "loss": 0.0645,
+      "step": 1744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.132888095840542e-05,
+      "loss": 0.0661,
+      "step": 1745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1320160954887051e-05,
+      "loss": 0.0651,
+      "step": 1746
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1311439929613377e-05,
+      "loss": 0.061,
+      "step": 1747
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1302717889334148e-05,
+      "loss": 0.056,
+      "step": 1748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1293994840799899e-05,
+      "loss": 0.063,
+      "step": 1749
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1285270790761938e-05,
+      "loss": 0.0799,
+      "step": 1750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1276545745972358e-05,
+      "loss": 0.0712,
+      "step": 1751
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1267819713184018e-05,
+      "loss": 0.0574,
+      "step": 1752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1259092699150544e-05,
+      "loss": 0.0579,
+      "step": 1753
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1250364710626314e-05,
+      "loss": 0.0642,
+      "step": 1754
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1241635754366467e-05,
+      "loss": 0.0732,
+      "step": 1755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1232905837126893e-05,
+      "loss": 0.0678,
+      "step": 1756
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1224174965664218e-05,
+      "loss": 0.0764,
+      "step": 1757
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1215443146735809e-05,
+      "loss": 0.0652,
+      "step": 1758
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1206710387099776e-05,
+      "loss": 0.0629,
+      "step": 1759
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.119797669351494e-05,
+      "loss": 0.0671,
+      "step": 1760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1189242072740858e-05,
+      "loss": 0.0729,
+      "step": 1761
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1180506531537798e-05,
+      "loss": 0.064,
+      "step": 1762
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1171770076666753e-05,
+      "loss": 0.07,
+      "step": 1763
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1163032714889398e-05,
+      "loss": 0.0648,
+      "step": 1764
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.115429445296813e-05,
+      "loss": 0.0683,
+      "step": 1765
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1145555297666044e-05,
+      "loss": 0.0732,
+      "step": 1766
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1136815255746915e-05,
+      "loss": 0.0703,
+      "step": 1767
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.112807433397521e-05,
+      "loss": 0.0633,
+      "step": 1768
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1119332539116074e-05,
+      "loss": 0.072,
+      "step": 1769
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1110589877935333e-05,
+      "loss": 0.073,
+      "step": 1770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.110184635719948e-05,
+      "loss": 0.0681,
+      "step": 1771
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1093101983675669e-05,
+      "loss": 0.0692,
+      "step": 1772
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1084356764131727e-05,
+      "loss": 0.0619,
+      "step": 1773
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1075610705336121e-05,
+      "loss": 0.0643,
+      "step": 1774
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1066863814057976e-05,
+      "loss": 0.0708,
+      "step": 1775
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1058116097067059e-05,
+      "loss": 0.0664,
+      "step": 1776
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1049367561133782e-05,
+      "loss": 0.0588,
+      "step": 1777
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1040618213029176e-05,
+      "loss": 0.0625,
+      "step": 1778
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1031868059524911e-05,
+      "loss": 0.0558,
+      "step": 1779
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1023117107393289e-05,
+      "loss": 0.0656,
+      "step": 1780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.101436536340721e-05,
+      "loss": 0.0724,
+      "step": 1781
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.10056128343402e-05,
+      "loss": 0.0606,
+      "step": 1782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0996859526966392e-05,
+      "loss": 0.056,
+      "step": 1783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.098810544806052e-05,
+      "loss": 0.0552,
+      "step": 1784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0979350604397912e-05,
+      "loss": 0.0729,
+      "step": 1785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0970595002754488e-05,
+      "loss": 0.0686,
+      "step": 1786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0961838649906757e-05,
+      "loss": 0.0562,
+      "step": 1787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0953081552631816e-05,
+      "loss": 0.0666,
+      "step": 1788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0944323717707326e-05,
+      "loss": 0.0729,
+      "step": 1789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0935565151911525e-05,
+      "loss": 0.0627,
+      "step": 1790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0926805862023215e-05,
+      "loss": 0.0576,
+      "step": 1791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0918045854821763e-05,
+      "loss": 0.0648,
+      "step": 1792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0909285137087085e-05,
+      "loss": 0.0731,
+      "step": 1793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.090052371559965e-05,
+      "loss": 0.0732,
+      "step": 1794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0891761597140472e-05,
+      "loss": 0.0603,
+      "step": 1795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0882998788491097e-05,
+      "loss": 0.0637,
+      "step": 1796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0874235296433624e-05,
+      "loss": 0.0681,
+      "step": 1797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0865471127750661e-05,
+      "loss": 0.0652,
+      "step": 1798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0856706289225346e-05,
+      "loss": 0.0632,
+      "step": 1799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0847940787641344e-05,
+      "loss": 0.0742,
+      "step": 1800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0839174629782818e-05,
+      "loss": 0.0751,
+      "step": 1801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0830407822434455e-05,
+      "loss": 0.0666,
+      "step": 1802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0821640372381428e-05,
+      "loss": 0.0697,
+      "step": 1803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.081287228640942e-05,
+      "loss": 0.0665,
+      "step": 1804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0804103571304606e-05,
+      "loss": 0.0664,
+      "step": 1805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0795334233853641e-05,
+      "loss": 0.0667,
+      "step": 1806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0786564280843661e-05,
+      "loss": 0.057,
+      "step": 1807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0777793719062291e-05,
+      "loss": 0.0668,
+      "step": 1808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0769022555297612e-05,
+      "loss": 0.0653,
+      "step": 1809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0760250796338177e-05,
+      "loss": 0.0638,
+      "step": 1810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0751478448973002e-05,
+      "loss": 0.0658,
+      "step": 1811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0742705519991558e-05,
+      "loss": 0.0638,
+      "step": 1812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0733932016183764e-05,
+      "loss": 0.0759,
+      "step": 1813
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0725157944339981e-05,
+      "loss": 0.0696,
+      "step": 1814
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0716383311251015e-05,
+      "loss": 0.0795,
+      "step": 1815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.070760812370811e-05,
+      "loss": 0.0641,
+      "step": 1816
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0698832388502924e-05,
+      "loss": 0.0713,
+      "step": 1817
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0690056112427553e-05,
+      "loss": 0.0666,
+      "step": 1818
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.068127930227451e-05,
+      "loss": 0.0637,
+      "step": 1819
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0672501964836715e-05,
+      "loss": 0.0641,
+      "step": 1820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0663724106907498e-05,
+      "loss": 0.0599,
+      "step": 1821
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0654945735280599e-05,
+      "loss": 0.0672,
+      "step": 1822
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0646166856750147e-05,
+      "loss": 0.0714,
+      "step": 1823
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0637387478110663e-05,
+      "loss": 0.0731,
+      "step": 1824
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0628607606157065e-05,
+      "loss": 0.0672,
+      "step": 1825
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.061982724768464e-05,
+      "loss": 0.0767,
+      "step": 1826
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0611046409489059e-05,
+      "loss": 0.0666,
+      "step": 1827
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0602265098366366e-05,
+      "loss": 0.0662,
+      "step": 1828
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0593483321112965e-05,
+      "loss": 0.0638,
+      "step": 1829
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0584701084525628e-05,
+      "loss": 0.0764,
+      "step": 1830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0575918395401474e-05,
+      "loss": 0.0703,
+      "step": 1831
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0567135260537978e-05,
+      "loss": 0.0684,
+      "step": 1832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.055835168673296e-05,
+      "loss": 0.0735,
+      "step": 1833
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0549567680784576e-05,
+      "loss": 0.0604,
+      "step": 1834
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0540783249491318e-05,
+      "loss": 0.0613,
+      "step": 1835
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0531998399652013e-05,
+      "loss": 0.0661,
+      "step": 1836
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0523213138065801e-05,
+      "loss": 0.0688,
+      "step": 1837
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.051442747153215e-05,
+      "loss": 0.0666,
+      "step": 1838
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0505641406850836e-05,
+      "loss": 0.0657,
+      "step": 1839
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0496854950821946e-05,
+      "loss": 0.0676,
+      "step": 1840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0488068110245871e-05,
+      "loss": 0.0675,
+      "step": 1841
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0479280891923296e-05,
+      "loss": 0.0585,
+      "step": 1842
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0470493302655196e-05,
+      "loss": 0.0561,
+      "step": 1843
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0461705349242845e-05,
+      "loss": 0.0605,
+      "step": 1844
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0452917038487787e-05,
+      "loss": 0.0499,
+      "step": 1845
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0444128377191842e-05,
+      "loss": 0.0599,
+      "step": 1846
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0435339372157114e-05,
+      "loss": 0.058,
+      "step": 1847
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.042655003018596e-05,
+      "loss": 0.0527,
+      "step": 1848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0417760358081001e-05,
+      "loss": 0.0453,
+      "step": 1849
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0408970362645121e-05,
+      "loss": 0.0718,
+      "step": 1850
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0400180050681445e-05,
+      "loss": 0.0481,
+      "step": 1851
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0391389428993346e-05,
+      "loss": 0.0573,
+      "step": 1852
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0382598504384438e-05,
+      "loss": 0.0591,
+      "step": 1853
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0373807283658566e-05,
+      "loss": 0.0566,
+      "step": 1854
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0365015773619812e-05,
+      "loss": 0.0563,
+      "step": 1855
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0356223981072471e-05,
+      "loss": 0.0548,
+      "step": 1856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0347431912821065e-05,
+      "loss": 0.0482,
+      "step": 1857
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0338639575670326e-05,
+      "loss": 0.0544,
+      "step": 1858
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0329846976425195e-05,
+      "loss": 0.0476,
+      "step": 1859
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0321054121890817e-05,
+      "loss": 0.0475,
+      "step": 1860
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0312261018872532e-05,
+      "loss": 0.0579,
+      "step": 1861
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0303467674175873e-05,
+      "loss": 0.054,
+      "step": 1862
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.029467409460656e-05,
+      "loss": 0.065,
+      "step": 1863
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0285880286970495e-05,
+      "loss": 0.0556,
+      "step": 1864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0277086258073765e-05,
+      "loss": 0.0508,
+      "step": 1865
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0268292014722608e-05,
+      "loss": 0.0485,
+      "step": 1866
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0259497563723448e-05,
+      "loss": 0.0562,
+      "step": 1867
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0250702911882857e-05,
+      "loss": 0.0611,
+      "step": 1868
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0241908066007571e-05,
+      "loss": 0.0501,
+      "step": 1869
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0233113032904473e-05,
+      "loss": 0.0577,
+      "step": 1870
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0224317819380585e-05,
+      "loss": 0.0593,
+      "step": 1871
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0215522432243079e-05,
+      "loss": 0.0566,
+      "step": 1872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0206726878299257e-05,
+      "loss": 0.0488,
+      "step": 1873
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0197931164356543e-05,
+      "loss": 0.0548,
+      "step": 1874
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0189135297222497e-05,
+      "loss": 0.0632,
+      "step": 1875
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0180339283704795e-05,
+      "loss": 0.0586,
+      "step": 1876
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0171543130611212e-05,
+      "loss": 0.0625,
+      "step": 1877
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.016274684474965e-05,
+      "loss": 0.0497,
+      "step": 1878
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0153950432928109e-05,
+      "loss": 0.0413,
+      "step": 1879
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0145153901954676e-05,
+      "loss": 0.0711,
+      "step": 1880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0136357258637542e-05,
+      "loss": 0.0473,
+      "step": 1881
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0127560509784977e-05,
+      "loss": 0.0497,
+      "step": 1882
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.011876366220534e-05,
+      "loss": 0.0607,
+      "step": 1883
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0109966722707062e-05,
+      "loss": 0.0436,
+      "step": 1884
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0101169698098644e-05,
+      "loss": 0.0511,
+      "step": 1885
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0092372595188658e-05,
+      "loss": 0.0487,
+      "step": 1886
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.008357542078573e-05,
+      "loss": 0.0566,
+      "step": 1887
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0074778181698544e-05,
+      "loss": 0.058,
+      "step": 1888
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0065980884735836e-05,
+      "loss": 0.055,
+      "step": 1889
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.005718353670639e-05,
+      "loss": 0.052,
+      "step": 1890
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0048386144419016e-05,
+      "loss": 0.0461,
+      "step": 1891
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0039588714682574e-05,
+      "loss": 0.0527,
+      "step": 1892
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0030791254305948e-05,
+      "loss": 0.066,
+      "step": 1893
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0021993770098038e-05,
+      "loss": 0.0568,
+      "step": 1894
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0013196268867768e-05,
+      "loss": 0.0552,
+      "step": 1895
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0004398757424081e-05,
+      "loss": 0.0469,
+      "step": 1896
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.99560124257592e-06,
+      "loss": 0.0604,
+      "step": 1897
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.986803731132234e-06,
+      "loss": 0.0541,
+      "step": 1898
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.978006229901964e-06,
+      "loss": 0.0532,
+      "step": 1899
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.969208745694057e-06,
+      "loss": 0.0505,
+      "step": 1900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.960411285317428e-06,
+      "loss": 0.0553,
+      "step": 1901
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.951613855580986e-06,
+      "loss": 0.0652,
+      "step": 1902
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.942816463293613e-06,
+      "loss": 0.0573,
+      "step": 1903
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.934019115264166e-06,
+      "loss": 0.0579,
+      "step": 1904
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.925221818301463e-06,
+      "loss": 0.0485,
+      "step": 1905
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.916424579214276e-06,
+      "loss": 0.0494,
+      "step": 1906
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.907627404811347e-06,
+      "loss": 0.0506,
+      "step": 1907
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.898830301901358e-06,
+      "loss": 0.0521,
+      "step": 1908
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.89003327729294e-06,
+      "loss": 0.0501,
+      "step": 1909
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.88123633779466e-06,
+      "loss": 0.0455,
+      "step": 1910
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.872439490215027e-06,
+      "loss": 0.0533,
+      "step": 1911
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.863642741362463e-06,
+      "loss": 0.0463,
+      "step": 1912
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.854846098045327e-06,
+      "loss": 0.0556,
+      "step": 1913
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.846049567071895e-06,
+      "loss": 0.0548,
+      "step": 1914
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.837253155250351e-06,
+      "loss": 0.0427,
+      "step": 1915
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.828456869388788e-06,
+      "loss": 0.0577,
+      "step": 1916
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.81966071629521e-06,
+      "loss": 0.0466,
+      "step": 1917
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.810864702777506e-06,
+      "loss": 0.0496,
+      "step": 1918
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.802068835643459e-06,
+      "loss": 0.0542,
+      "step": 1919
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.793273121700746e-06,
+      "loss": 0.049,
+      "step": 1920
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.784477567756923e-06,
+      "loss": 0.0526,
+      "step": 1921
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.77568218061942e-06,
+      "loss": 0.0435,
+      "step": 1922
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.766886967095532e-06,
+      "loss": 0.0565,
+      "step": 1923
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.75809193399243e-06,
+      "loss": 0.0433,
+      "step": 1924
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.749297088117146e-06,
+      "loss": 0.0549,
+      "step": 1925
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.740502436276556e-06,
+      "loss": 0.0527,
+      "step": 1926
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.731707985277392e-06,
+      "loss": 0.0529,
+      "step": 1927
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.72291374192624e-06,
+      "loss": 0.0413,
+      "step": 1928
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.714119713029507e-06,
+      "loss": 0.0391,
+      "step": 1929
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.705325905393443e-06,
+      "loss": 0.0526,
+      "step": 1930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.696532325824129e-06,
+      "loss": 0.0516,
+      "step": 1931
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.687738981127471e-06,
+      "loss": 0.0591,
+      "step": 1932
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.678945878109185e-06,
+      "loss": 0.0562,
+      "step": 1933
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.670153023574808e-06,
+      "loss": 0.044,
+      "step": 1934
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.661360424329677e-06,
+      "loss": 0.0541,
+      "step": 1935
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.652568087178939e-06,
+      "loss": 0.0573,
+      "step": 1936
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.643776018927532e-06,
+      "loss": 0.0551,
+      "step": 1937
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.63498422638019e-06,
+      "loss": 0.0455,
+      "step": 1938
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.626192716341438e-06,
+      "loss": 0.051,
+      "step": 1939
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.617401495615569e-06,
+      "loss": 0.0478,
+      "step": 1940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.608610571006659e-06,
+      "loss": 0.0463,
+      "step": 1941
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.599819949318557e-06,
+      "loss": 0.0511,
+      "step": 1942
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.59102963735488e-06,
+      "loss": 0.0553,
+      "step": 1943
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.582239641918997e-06,
+      "loss": 0.0488,
+      "step": 1944
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.573449969814045e-06,
+      "loss": 0.0523,
+      "step": 1945
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.564660627842888e-06,
+      "loss": 0.0407,
+      "step": 1946
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.555871622808161e-06,
+      "loss": 0.0451,
+      "step": 1947
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.547082961512216e-06,
+      "loss": 0.0573,
+      "step": 1948
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.538294650757158e-06,
+      "loss": 0.0501,
+      "step": 1949
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.529506697344804e-06,
+      "loss": 0.0513,
+      "step": 1950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.520719108076709e-06,
+      "loss": 0.0505,
+      "step": 1951
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.51193188975413e-06,
+      "loss": 0.0516,
+      "step": 1952
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.503145049178056e-06,
+      "loss": 0.0454,
+      "step": 1953
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.494358593149167e-06,
+      "loss": 0.0518,
+      "step": 1954
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.485572528467852e-06,
+      "loss": 0.0609,
+      "step": 1955
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.476786861934202e-06,
+      "loss": 0.0483,
+      "step": 1956
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.468001600347992e-06,
+      "loss": 0.0574,
+      "step": 1957
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.459216750508684e-06,
+      "loss": 0.0443,
+      "step": 1958
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.450432319215426e-06,
+      "loss": 0.0527,
+      "step": 1959
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.441648313267044e-06,
+      "loss": 0.0561,
+      "step": 1960
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.432864739462022e-06,
+      "loss": 0.0449,
+      "step": 1961
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.42408160459853e-06,
+      "loss": 0.0602,
+      "step": 1962
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.415298915474374e-06,
+      "loss": 0.052,
+      "step": 1963
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.406516678887037e-06,
+      "loss": 0.0481,
+      "step": 1964
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.397734901633636e-06,
+      "loss": 0.0544,
+      "step": 1965
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.388953590510941e-06,
+      "loss": 0.0579,
+      "step": 1966
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.380172752315364e-06,
+      "loss": 0.0425,
+      "step": 1967
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.371392393842942e-06,
+      "loss": 0.0534,
+      "step": 1968
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.36261252188934e-06,
+      "loss": 0.0508,
+      "step": 1969
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.353833143249857e-06,
+      "loss": 0.0478,
+      "step": 1970
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.345054264719403e-06,
+      "loss": 0.0598,
+      "step": 1971
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.3362758930925e-06,
+      "loss": 0.046,
+      "step": 1972
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.32749803516329e-06,
+      "loss": 0.0457,
+      "step": 1973
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.318720697725493e-06,
+      "loss": 0.0497,
+      "step": 1974
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.309943887572449e-06,
+      "loss": 0.0553,
+      "step": 1975
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.301167611497079e-06,
+      "loss": 0.0664,
+      "step": 1976
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.292391876291893e-06,
+      "loss": 0.0542,
+      "step": 1977
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.283616688748985e-06,
+      "loss": 0.0427,
+      "step": 1978
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.274842055660024e-06,
+      "loss": 0.0508,
+      "step": 1979
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.26606798381624e-06,
+      "loss": 0.0547,
+      "step": 1980
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.257294480008444e-06,
+      "loss": 0.0467,
+      "step": 1981
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.248521551027e-06,
+      "loss": 0.0497,
+      "step": 1982
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.239749203661825e-06,
+      "loss": 0.0479,
+      "step": 1983
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.230977444702391e-06,
+      "loss": 0.0497,
+      "step": 1984
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.222206280937712e-06,
+      "loss": 0.0588,
+      "step": 1985
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.21343571915634e-06,
+      "loss": 0.0575,
+      "step": 1986
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.204665766146362e-06,
+      "loss": 0.0653,
+      "step": 1987
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.195896428695396e-06,
+      "loss": 0.0558,
+      "step": 1988
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.18712771359058e-06,
+      "loss": 0.0486,
+      "step": 1989
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.178359627618577e-06,
+      "loss": 0.051,
+      "step": 1990
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.16959217756555e-06,
+      "loss": 0.0542,
+      "step": 1991
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.160825370217185e-06,
+      "loss": 0.0617,
+      "step": 1992
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.152059212358659e-06,
+      "loss": 0.0486,
+      "step": 1993
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.143293710774655e-06,
+      "loss": 0.0438,
+      "step": 1994
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.134528872249342e-06,
+      "loss": 0.0437,
+      "step": 1995
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.12576470356638e-06,
+      "loss": 0.0573,
+      "step": 1996
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.117001211508905e-06,
+      "loss": 0.058,
+      "step": 1997
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.108238402859533e-06,
+      "loss": 0.046,
+      "step": 1998
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.099476284400354e-06,
+      "loss": 0.0514,
+      "step": 1999
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.090714862912917e-06,
+      "loss": 0.0508,
+      "step": 2000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.081954145178237e-06,
+      "loss": 0.0562,
+      "step": 2001
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.073194137976787e-06,
+      "loss": 0.0508,
+      "step": 2002
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.06443484808848e-06,
+      "loss": 0.0584,
+      "step": 2003
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.055676282292678e-06,
+      "loss": 0.0477,
+      "step": 2004
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.046918447368185e-06,
+      "loss": 0.0505,
+      "step": 2005
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.038161350093243e-06,
+      "loss": 0.0426,
+      "step": 2006
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.029404997245519e-06,
+      "loss": 0.0451,
+      "step": 2007
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.020649395602093e-06,
+      "loss": 0.0576,
+      "step": 2008
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.011894551939484e-06,
+      "loss": 0.046,
+      "step": 2009
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.00314047303361e-06,
+      "loss": 0.051,
+      "step": 2010
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.994387165659801e-06,
+      "loss": 0.0451,
+      "step": 2011
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.98563463659279e-06,
+      "loss": 0.0716,
+      "step": 2012
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.976882892606716e-06,
+      "loss": 0.0523,
+      "step": 2013
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.96813194047509e-06,
+      "loss": 0.0509,
+      "step": 2014
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.95938178697083e-06,
+      "loss": 0.0411,
+      "step": 2015
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.950632438866221e-06,
+      "loss": 0.0462,
+      "step": 2016
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.941883902932943e-06,
+      "loss": 0.0594,
+      "step": 2017
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.933136185942029e-06,
+      "loss": 0.0509,
+      "step": 2018
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.924389294663884e-06,
+      "loss": 0.0572,
+      "step": 2019
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.915643235868275e-06,
+      "loss": 0.042,
+      "step": 2020
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.906898016324333e-06,
+      "loss": 0.0559,
+      "step": 2021
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.898153642800524e-06,
+      "loss": 0.0638,
+      "step": 2022
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.88941012206467e-06,
+      "loss": 0.063,
+      "step": 2023
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.88066746088393e-06,
+      "loss": 0.0514,
+      "step": 2024
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.871925666024794e-06,
+      "loss": 0.048,
+      "step": 2025
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.863184744253086e-06,
+      "loss": 0.0603,
+      "step": 2026
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.854444702333958e-06,
+      "loss": 0.0509,
+      "step": 2027
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.84570554703187e-06,
+      "loss": 0.0625,
+      "step": 2028
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.836967285110604e-06,
+      "loss": 0.0502,
+      "step": 2029
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.828229923333254e-06,
+      "loss": 0.0536,
+      "step": 2030
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.819493468462203e-06,
+      "loss": 0.0508,
+      "step": 2031
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.810757927259145e-06,
+      "loss": 0.0449,
+      "step": 2032
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.802023306485063e-06,
+      "loss": 0.0562,
+      "step": 2033
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.79328961290023e-06,
+      "loss": 0.0614,
+      "step": 2034
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.784556853264196e-06,
+      "loss": 0.049,
+      "step": 2035
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.775825034335788e-06,
+      "loss": 0.0511,
+      "step": 2036
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.76709416287311e-06,
+      "loss": 0.0542,
+      "step": 2037
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.758364245633536e-06,
+      "loss": 0.0521,
+      "step": 2038
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.749635289373688e-06,
+      "loss": 0.0664,
+      "step": 2039
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.740907300849457e-06,
+      "loss": 0.0536,
+      "step": 2040
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.732180286815984e-06,
+      "loss": 0.0453,
+      "step": 2041
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.723454254027645e-06,
+      "loss": 0.0496,
+      "step": 2042
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.714729209238066e-06,
+      "loss": 0.0592,
+      "step": 2043
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.706005159200103e-06,
+      "loss": 0.0484,
+      "step": 2044
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.697282110665854e-06,
+      "loss": 0.0576,
+      "step": 2045
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.688560070386624e-06,
+      "loss": 0.055,
+      "step": 2046
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.679839045112954e-06,
+      "loss": 0.0685,
+      "step": 2047
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.671119041594584e-06,
+      "loss": 0.0602,
+      "step": 2048
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.662400066580478e-06,
+      "loss": 0.0463,
+      "step": 2049
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.653682126818792e-06,
+      "loss": 0.045,
+      "step": 2050
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.644965229056888e-06,
+      "loss": 0.0542,
+      "step": 2051
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.636249380041325e-06,
+      "loss": 0.0551,
+      "step": 2052
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.627534586517839e-06,
+      "loss": 0.0505,
+      "step": 2053
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.618820855231351e-06,
+      "loss": 0.0471,
+      "step": 2054
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.61010819292597e-06,
+      "loss": 0.0583,
+      "step": 2055
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.601396606344972e-06,
+      "loss": 0.0536,
+      "step": 2056
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.592686102230797e-06,
+      "loss": 0.0377,
+      "step": 2057
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.583976687325053e-06,
+      "loss": 0.0527,
+      "step": 2058
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.575268368368496e-06,
+      "loss": 0.0567,
+      "step": 2059
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.566561152101048e-06,
+      "loss": 0.0587,
+      "step": 2060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.557855045261761e-06,
+      "loss": 0.0475,
+      "step": 2061
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.549150054588845e-06,
+      "loss": 0.0576,
+      "step": 2062
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.540446186819631e-06,
+      "loss": 0.049,
+      "step": 2063
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.531743448690594e-06,
+      "loss": 0.0465,
+      "step": 2064
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.52304184693732e-06,
+      "loss": 0.0598,
+      "step": 2065
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.514341388294534e-06,
+      "loss": 0.0498,
+      "step": 2066
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.505642079496057e-06,
+      "loss": 0.0445,
+      "step": 2067
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.496943927274834e-06,
+      "loss": 0.0434,
+      "step": 2068
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.488246938362915e-06,
+      "loss": 0.0557,
+      "step": 2069
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.479551119491438e-06,
+      "loss": 0.0418,
+      "step": 2070
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.47085647739064e-06,
+      "loss": 0.0453,
+      "step": 2071
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.462163018789853e-06,
+      "loss": 0.0441,
+      "step": 2072
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.453470750417494e-06,
+      "loss": 0.0531,
+      "step": 2073
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.444779679001045e-06,
+      "loss": 0.039,
+      "step": 2074
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.436089811267081e-06,
+      "loss": 0.0497,
+      "step": 2075
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.427401153941227e-06,
+      "loss": 0.0573,
+      "step": 2076
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.418713713748186e-06,
+      "loss": 0.0602,
+      "step": 2077
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.410027497411704e-06,
+      "loss": 0.0447,
+      "step": 2078
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.401342511654596e-06,
+      "loss": 0.0431,
+      "step": 2079
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.392658763198714e-06,
+      "loss": 0.0583,
+      "step": 2080
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.38397625876496e-06,
+      "loss": 0.0531,
+      "step": 2081
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.375295005073257e-06,
+      "loss": 0.0508,
+      "step": 2082
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.366615008842578e-06,
+      "loss": 0.0577,
+      "step": 2083
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.35793627679092e-06,
+      "loss": 0.0582,
+      "step": 2084
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.349258815635291e-06,
+      "loss": 0.0502,
+      "step": 2085
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.34058263209173e-06,
+      "loss": 0.0403,
+      "step": 2086
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.331907732875267e-06,
+      "loss": 0.0537,
+      "step": 2087
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.323234124699962e-06,
+      "loss": 0.0507,
+      "step": 2088
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.314561814278856e-06,
+      "loss": 0.0512,
+      "step": 2089
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.305890808323998e-06,
+      "loss": 0.061,
+      "step": 2090
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.297221113546422e-06,
+      "loss": 0.0555,
+      "step": 2091
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.288552736656151e-06,
+      "loss": 0.0424,
+      "step": 2092
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.27988568436218e-06,
+      "loss": 0.0459,
+      "step": 2093
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.271219963372484e-06,
+      "loss": 0.0608,
+      "step": 2094
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.26255558039401e-06,
+      "loss": 0.0506,
+      "step": 2095
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.253892542132666e-06,
+      "loss": 0.0602,
+      "step": 2096
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.245230855293325e-06,
+      "loss": 0.0526,
+      "step": 2097
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.236570526579805e-06,
+      "loss": 0.0495,
+      "step": 2098
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.227911562694874e-06,
+      "loss": 0.0554,
+      "step": 2099
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.21925397034025e-06,
+      "loss": 0.0523,
+      "step": 2100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.210597756216591e-06,
+      "loss": 0.0406,
+      "step": 2101
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.201942927023474e-06,
+      "loss": 0.0583,
+      "step": 2102
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.193289489459427e-06,
+      "loss": 0.0505,
+      "step": 2103
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.184637450221873e-06,
+      "loss": 0.0488,
+      "step": 2104
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.175986816007177e-06,
+      "loss": 0.0517,
+      "step": 2105
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.1673375935106e-06,
+      "loss": 0.0539,
+      "step": 2106
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.158689789426323e-06,
+      "loss": 0.0517,
+      "step": 2107
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.150043410447425e-06,
+      "loss": 0.0597,
+      "step": 2108
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.14139846326588e-06,
+      "loss": 0.0483,
+      "step": 2109
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.132754954572549e-06,
+      "loss": 0.052,
+      "step": 2110
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.124112891057185e-06,
+      "loss": 0.0437,
+      "step": 2111
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.115472279408431e-06,
+      "loss": 0.0519,
+      "step": 2112
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.106833126313789e-06,
+      "loss": 0.0513,
+      "step": 2113
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.098195438459645e-06,
+      "loss": 0.0546,
+      "step": 2114
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.089559222531253e-06,
+      "loss": 0.0644,
+      "step": 2115
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.080924485212712e-06,
+      "loss": 0.0592,
+      "step": 2116
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.072291233186986e-06,
+      "loss": 0.0525,
+      "step": 2117
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.063659473135896e-06,
+      "loss": 0.0544,
+      "step": 2118
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.055029211740101e-06,
+      "loss": 0.0461,
+      "step": 2119
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.046400455679104e-06,
+      "loss": 0.048,
+      "step": 2120
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.037773211631232e-06,
+      "loss": 0.0457,
+      "step": 2121
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.029147486273652e-06,
+      "loss": 0.0461,
+      "step": 2122
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.02052328628236e-06,
+      "loss": 0.054,
+      "step": 2123
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.011900618332156e-06,
+      "loss": 0.0503,
+      "step": 2124
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.00327948909667e-06,
+      "loss": 0.0395,
+      "step": 2125
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 7.994659905248335e-06,
+      "loss": 0.0562,
+      "step": 2126
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.986041873458383e-06,
+      "loss": 0.0565,
+      "step": 2127
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.977425400396846e-06,
+      "loss": 0.058,
+      "step": 2128
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.96881049273256e-06,
+      "loss": 0.0461,
+      "step": 2129
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.960197157133137e-06,
+      "loss": 0.0549,
+      "step": 2130
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.951585400264981e-06,
+      "loss": 0.0576,
+      "step": 2131
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.942975228793265e-06,
+      "loss": 0.0537,
+      "step": 2132
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.934366649381946e-06,
+      "loss": 0.0438,
+      "step": 2133
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.925759668693739e-06,
+      "loss": 0.0412,
+      "step": 2134
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.917154293390128e-06,
+      "loss": 0.0457,
+      "step": 2135
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.908550530131351e-06,
+      "loss": 0.0542,
+      "step": 2136
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.899948385576404e-06,
+      "loss": 0.0445,
+      "step": 2137
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.891347866383019e-06,
+      "loss": 0.0434,
+      "step": 2138
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.88274897920768e-06,
+      "loss": 0.0431,
+      "step": 2139
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.874151730705606e-06,
+      "loss": 0.0592,
+      "step": 2140
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.865556127530744e-06,
+      "loss": 0.0499,
+      "step": 2141
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.85696217633577e-06,
+      "loss": 0.0491,
+      "step": 2142
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.848369883772088e-06,
+      "loss": 0.0468,
+      "step": 2143
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.839779256489803e-06,
+      "loss": 0.0523,
+      "step": 2144
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.831190301137743e-06,
+      "loss": 0.0527,
+      "step": 2145
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.822603024363438e-06,
+      "loss": 0.0493,
+      "step": 2146
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.814017432813124e-06,
+      "loss": 0.0666,
+      "step": 2147
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.805433533131726e-06,
+      "loss": 0.0685,
+      "step": 2148
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.796851331962857e-06,
+      "loss": 0.0512,
+      "step": 2149
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.788270835948826e-06,
+      "loss": 0.0487,
+      "step": 2150
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.779692051730617e-06,
+      "loss": 0.0511,
+      "step": 2151
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.771114985947885e-06,
+      "loss": 0.0536,
+      "step": 2152
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.76253964523896e-06,
+      "loss": 0.0482,
+      "step": 2153
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.75396603624084e-06,
+      "loss": 0.0575,
+      "step": 2154
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.745394165589174e-06,
+      "loss": 0.0416,
+      "step": 2155
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.736824039918268e-06,
+      "loss": 0.0498,
+      "step": 2156
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.728255665861082e-06,
+      "loss": 0.0497,
+      "step": 2157
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.719689050049222e-06,
+      "loss": 0.0491,
+      "step": 2158
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.71112419911292e-06,
+      "loss": 0.0478,
+      "step": 2159
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.702561119681063e-06,
+      "loss": 0.0466,
+      "step": 2160
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.69399981838114e-06,
+      "loss": 0.0543,
+      "step": 2161
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.68544030183929e-06,
+      "loss": 0.0525,
+      "step": 2162
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.676882576680254e-06,
+      "loss": 0.0562,
+      "step": 2163
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.668326649527394e-06,
+      "loss": 0.0517,
+      "step": 2164
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.659772527002677e-06,
+      "loss": 0.045,
+      "step": 2165
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.651220215726676e-06,
+      "loss": 0.0596,
+      "step": 2166
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.64266972231856e-06,
+      "loss": 0.0485,
+      "step": 2167
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.63412105339609e-06,
+      "loss": 0.0519,
+      "step": 2168
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.6255742155756195e-06,
+      "loss": 0.0465,
+      "step": 2169
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.617029215472079e-06,
+      "loss": 0.0444,
+      "step": 2170
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.608486059698987e-06,
+      "loss": 0.0514,
+      "step": 2171
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.599944754868418e-06,
+      "loss": 0.0474,
+      "step": 2172
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5914053075910254e-06,
+      "loss": 0.0485,
+      "step": 2173
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.582867724476024e-06,
+      "loss": 0.0476,
+      "step": 2174
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.574332012131187e-06,
+      "loss": 0.055,
+      "step": 2175
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.565798177162832e-06,
+      "loss": 0.0478,
+      "step": 2176
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.557266226175835e-06,
+      "loss": 0.0526,
+      "step": 2177
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5487361657736e-06,
+      "loss": 0.0546,
+      "step": 2178
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.540208002558082e-06,
+      "loss": 0.0512,
+      "step": 2179
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.531681743129758e-06,
+      "loss": 0.0487,
+      "step": 2180
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.523157394087632e-06,
+      "loss": 0.0514,
+      "step": 2181
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.51463496202924e-06,
+      "loss": 0.0461,
+      "step": 2182
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.5061144535506195e-06,
+      "loss": 0.0513,
+      "step": 2183
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.497595875246325e-06,
+      "loss": 0.0437,
+      "step": 2184
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.489079233709422e-06,
+      "loss": 0.0529,
+      "step": 2185
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.480564535531474e-06,
+      "loss": 0.0432,
+      "step": 2186
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4720517873025345e-06,
+      "loss": 0.0562,
+      "step": 2187
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.46354099561116e-06,
+      "loss": 0.0495,
+      "step": 2188
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4550321670443795e-06,
+      "loss": 0.0406,
+      "step": 2189
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.446525308187711e-06,
+      "loss": 0.0476,
+      "step": 2190
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.4380204256251455e-06,
+      "loss": 0.0555,
+      "step": 2191
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.429517525939142e-06,
+      "loss": 0.0552,
+      "step": 2192
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.421016615710635e-06,
+      "loss": 0.0523,
+      "step": 2193
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.412517701519009e-06,
+      "loss": 0.053,
+      "step": 2194
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.404020789942098e-06,
+      "loss": 0.0487,
+      "step": 2195
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.395525887556202e-06,
+      "loss": 0.0635,
+      "step": 2196
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.3870330009360595e-06,
+      "loss": 0.0475,
+      "step": 2197
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.378542136654843e-06,
+      "loss": 0.0533,
+      "step": 2198
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.37005330128417e-06,
+      "loss": 0.0549,
+      "step": 2199
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.361566501394077e-06,
+      "loss": 0.056,
+      "step": 2200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.353081743553033e-06,
+      "loss": 0.0525,
+      "step": 2201
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.344599034327922e-06,
+      "loss": 0.0508,
+      "step": 2202
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.3361183802840475e-06,
+      "loss": 0.0582,
+      "step": 2203
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.327639787985115e-06,
+      "loss": 0.0573,
+      "step": 2204
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.319163263993246e-06,
+      "loss": 0.043,
+      "step": 2205
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.310688814868944e-06,
+      "loss": 0.0448,
+      "step": 2206
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.302216447171122e-06,
+      "loss": 0.0466,
+      "step": 2207
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.293746167457074e-06,
+      "loss": 0.0454,
+      "step": 2208
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.285277982282482e-06,
+      "loss": 0.0455,
+      "step": 2209
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.276811898201406e-06,
+      "loss": 0.0554,
+      "step": 2210
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.268347921766281e-06,
+      "loss": 0.0527,
+      "step": 2211
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.259886059527901e-06,
+      "loss": 0.0441,
+      "step": 2212
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.251426318035435e-06,
+      "loss": 0.0457,
+      "step": 2213
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2429687038364124e-06,
+      "loss": 0.046,
+      "step": 2214
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2345132234767035e-06,
+      "loss": 0.0535,
+      "step": 2215
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.226059883500543e-06,
+      "loss": 0.0415,
+      "step": 2216
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.21760869045049e-06,
+      "loss": 0.0513,
+      "step": 2217
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.209159650867462e-06,
+      "loss": 0.0485,
+      "step": 2218
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.2007127712906934e-06,
+      "loss": 0.053,
+      "step": 2219
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.192268058257756e-06,
+      "loss": 0.0569,
+      "step": 2220
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.183825518304546e-06,
+      "loss": 0.0511,
+      "step": 2221
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.175385157965272e-06,
+      "loss": 0.0419,
+      "step": 2222
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.166946983772454e-06,
+      "loss": 0.0504,
+      "step": 2223
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.158511002256926e-06,
+      "loss": 0.0416,
+      "step": 2224
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1500772199478265e-06,
+      "loss": 0.046,
+      "step": 2225
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1416456433725836e-06,
+      "loss": 0.0408,
+      "step": 2226
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.133216279056925e-06,
+      "loss": 0.0631,
+      "step": 2227
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.124789133524867e-06,
+      "loss": 0.0597,
+      "step": 2228
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.116364213298701e-06,
+      "loss": 0.0519,
+      "step": 2229
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1079415248990006e-06,
+      "loss": 0.0498,
+      "step": 2230
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.099521074844613e-06,
+      "loss": 0.0459,
+      "step": 2231
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.091102869652655e-06,
+      "loss": 0.0497,
+      "step": 2232
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.082686915838505e-06,
+      "loss": 0.0399,
+      "step": 2233
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.074273219915788e-06,
+      "loss": 0.0552,
+      "step": 2234
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.065861788396397e-06,
+      "loss": 0.052,
+      "step": 2235
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.057452627790466e-06,
+      "loss": 0.0574,
+      "step": 2236
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.049045744606371e-06,
+      "loss": 0.0406,
+      "step": 2237
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.040641145350726e-06,
+      "loss": 0.0508,
+      "step": 2238
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.032238836528376e-06,
+      "loss": 0.0401,
+      "step": 2239
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.023838824642395e-06,
+      "loss": 0.0429,
+      "step": 2240
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0154411161940775e-06,
+      "loss": 0.0539,
+      "step": 2241
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.007045717682939e-06,
+      "loss": 0.0506,
+      "step": 2242
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9986526356067016e-06,
+      "loss": 0.0513,
+      "step": 2243
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.990261876461303e-06,
+      "loss": 0.0464,
+      "step": 2244
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.981873446740869e-06,
+      "loss": 0.0472,
+      "step": 2245
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9734873529377376e-06,
+      "loss": 0.0484,
+      "step": 2246
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.965103601542429e-06,
+      "loss": 0.0463,
+      "step": 2247
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.956722199043655e-06,
+      "loss": 0.0411,
+      "step": 2248
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.948343151928312e-06,
+      "loss": 0.051,
+      "step": 2249
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.939966466681467e-06,
+      "loss": 0.0568,
+      "step": 2250
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.931592149786358e-06,
+      "loss": 0.0561,
+      "step": 2251
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.923220207724397e-06,
+      "loss": 0.0557,
+      "step": 2252
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9148506469751555e-06,
+      "loss": 0.0497,
+      "step": 2253
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.906483474016359e-06,
+      "loss": 0.0477,
+      "step": 2254
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.898118695323887e-06,
+      "loss": 0.0459,
+      "step": 2255
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.889756317371774e-06,
+      "loss": 0.0482,
+      "step": 2256
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.881396346632176e-06,
+      "loss": 0.0527,
+      "step": 2257
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.873038789575405e-06,
+      "loss": 0.0508,
+      "step": 2258
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.864683652669896e-06,
+      "loss": 0.0388,
+      "step": 2259
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8563309423822165e-06,
+      "loss": 0.0484,
+      "step": 2260
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.847980665177052e-06,
+      "loss": 0.0514,
+      "step": 2261
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8396328275171995e-06,
+      "loss": 0.0398,
+      "step": 2262
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8312874358635784e-06,
+      "loss": 0.0565,
+      "step": 2263
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.82294449667521e-06,
+      "loss": 0.0508,
+      "step": 2264
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.814604016409218e-06,
+      "loss": 0.0581,
+      "step": 2265
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.806266001520821e-06,
+      "loss": 0.062,
+      "step": 2266
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.797930458463338e-06,
+      "loss": 0.0493,
+      "step": 2267
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.789597393688161e-06,
+      "loss": 0.0539,
+      "step": 2268
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.78126681364477e-06,
+      "loss": 0.0446,
+      "step": 2269
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.772938724780727e-06,
+      "loss": 0.0507,
+      "step": 2270
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.764613133541662e-06,
+      "loss": 0.05,
+      "step": 2271
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.75629004637127e-06,
+      "loss": 0.0605,
+      "step": 2272
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.747969469711315e-06,
+      "loss": 0.0535,
+      "step": 2273
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.7396514100016066e-06,
+      "loss": 0.0653,
+      "step": 2274
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.731335873680012e-06,
+      "loss": 0.0508,
+      "step": 2275
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.723022867182447e-06,
+      "loss": 0.0487,
+      "step": 2276
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.714712396942873e-06,
+      "loss": 0.0412,
+      "step": 2277
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.706404469393284e-06,
+      "loss": 0.0642,
+      "step": 2278
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.698099090963697e-06,
+      "loss": 0.0511,
+      "step": 2279
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.689796268082169e-06,
+      "loss": 0.0506,
+      "step": 2280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.681496007174779e-06,
+      "loss": 0.0504,
+      "step": 2281
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.673198314665616e-06,
+      "loss": 0.0428,
+      "step": 2282
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.664903196976783e-06,
+      "loss": 0.048,
+      "step": 2283
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.656610660528398e-06,
+      "loss": 0.0571,
+      "step": 2284
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.648320711738568e-06,
+      "loss": 0.0522,
+      "step": 2285
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.640033357023403e-06,
+      "loss": 0.0409,
+      "step": 2286
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.631748602797009e-06,
+      "loss": 0.0479,
+      "step": 2287
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.623466455471479e-06,
+      "loss": 0.0436,
+      "step": 2288
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.615186921456882e-06,
+      "loss": 0.0481,
+      "step": 2289
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.606910007161273e-06,
+      "loss": 0.054,
+      "step": 2290
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.598635718990668e-06,
+      "loss": 0.0647,
+      "step": 2291
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.590364063349062e-06,
+      "loss": 0.0621,
+      "step": 2292
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.582095046638405e-06,
+      "loss": 0.0541,
+      "step": 2293
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.573828675258609e-06,
+      "loss": 0.0481,
+      "step": 2294
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.56556495560754e-06,
+      "loss": 0.0506,
+      "step": 2295
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.557303894081004e-06,
+      "loss": 0.0518,
+      "step": 2296
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.549045497072752e-06,
+      "loss": 0.0538,
+      "step": 2297
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.540789770974479e-06,
+      "loss": 0.0475,
+      "step": 2298
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.532536722175811e-06,
+      "loss": 0.0552,
+      "step": 2299
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.524286357064296e-06,
+      "loss": 0.0552,
+      "step": 2300
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.516038682025412e-06,
+      "loss": 0.0453,
+      "step": 2301
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.507793703442546e-06,
+      "loss": 0.0495,
+      "step": 2302
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.499551427697009e-06,
+      "loss": 0.0576,
+      "step": 2303
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.49131186116801e-06,
+      "loss": 0.0581,
+      "step": 2304
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.483075010232674e-06,
+      "loss": 0.0499,
+      "step": 2305
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.4748408812660046e-06,
+      "loss": 0.0422,
+      "step": 2306
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.466609480640924e-06,
+      "loss": 0.0441,
+      "step": 2307
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.458380814728215e-06,
+      "loss": 0.0488,
+      "step": 2308
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.450154889896569e-06,
+      "loss": 0.0501,
+      "step": 2309
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.441931712512537e-06,
+      "loss": 0.0537,
+      "step": 2310
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.433711288940557e-06,
+      "loss": 0.0506,
+      "step": 2311
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.425493625542929e-06,
+      "loss": 0.0469,
+      "step": 2312
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.4172787286798145e-06,
+      "loss": 0.0509,
+      "step": 2313
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.409066604709239e-06,
+      "loss": 0.0565,
+      "step": 2314
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.400857259987079e-06,
+      "loss": 0.0507,
+      "step": 2315
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.392650700867064e-06,
+      "loss": 0.0484,
+      "step": 2316
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.38444693370076e-06,
+      "loss": 0.0565,
+      "step": 2317
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.376245964837583e-06,
+      "loss": 0.0559,
+      "step": 2318
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.368047800624768e-06,
+      "loss": 0.0486,
+      "step": 2319
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3598524474073964e-06,
+      "loss": 0.0563,
+      "step": 2320
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.351659911528359e-06,
+      "loss": 0.0521,
+      "step": 2321
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.343470199328376e-06,
+      "loss": 0.0625,
+      "step": 2322
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3352833171459816e-06,
+      "loss": 0.056,
+      "step": 2323
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.327099271317516e-06,
+      "loss": 0.0424,
+      "step": 2324
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.31891806817712e-06,
+      "loss": 0.0521,
+      "step": 2325
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.310739714056743e-06,
+      "loss": 0.0539,
+      "step": 2326
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.302564215286127e-06,
+      "loss": 0.0477,
+      "step": 2327
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.2943915781928e-06,
+      "loss": 0.0413,
+      "step": 2328
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.286221809102084e-06,
+      "loss": 0.051,
+      "step": 2329
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.278054914337066e-06,
+      "loss": 0.0429,
+      "step": 2330
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.269890900218627e-06,
+      "loss": 0.0472,
+      "step": 2331
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.261729773065401e-06,
+      "loss": 0.0444,
+      "step": 2332
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.253571539193801e-06,
+      "loss": 0.0507,
+      "step": 2333
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.245416204917996e-06,
+      "loss": 0.0544,
+      "step": 2334
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.237263776549911e-06,
+      "loss": 0.0532,
+      "step": 2335
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.229114260399214e-06,
+      "loss": 0.0524,
+      "step": 2336
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.220967662773332e-06,
+      "loss": 0.0489,
+      "step": 2337
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2128239899774275e-06,
+      "loss": 0.0549,
+      "step": 2338
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.204683248314398e-06,
+      "loss": 0.0537,
+      "step": 2339
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.196545444084876e-06,
+      "loss": 0.0482,
+      "step": 2340
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1884105835872165e-06,
+      "loss": 0.0544,
+      "step": 2341
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.180278673117499e-06,
+      "loss": 0.0497,
+      "step": 2342
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.172149718969514e-06,
+      "loss": 0.0602,
+      "step": 2343
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1640237274347735e-06,
+      "loss": 0.0452,
+      "step": 2344
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.15590070480249e-06,
+      "loss": 0.0478,
+      "step": 2345
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.1477806573595835e-06,
+      "loss": 0.0551,
+      "step": 2346
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.139663591390661e-06,
+      "loss": 0.0525,
+      "step": 2347
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.131549513178032e-06,
+      "loss": 0.0457,
+      "step": 2348
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.123438429001689e-06,
+      "loss": 0.0405,
+      "step": 2349
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.11533034513931e-06,
+      "loss": 0.0534,
+      "step": 2350
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.1072252678662504e-06,
+      "loss": 0.0504,
+      "step": 2351
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.099123203455537e-06,
+      "loss": 0.0526,
+      "step": 2352
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.09102415817786e-06,
+      "loss": 0.0512,
+      "step": 2353
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.082928138301583e-06,
+      "loss": 0.054,
+      "step": 2354
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0748351500927235e-06,
+      "loss": 0.0567,
+      "step": 2355
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.066745199814951e-06,
+      "loss": 0.0472,
+      "step": 2356
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0586582937295845e-06,
+      "loss": 0.0499,
+      "step": 2357
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.050574438095591e-06,
+      "loss": 0.0604,
+      "step": 2358
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.042493639169569e-06,
+      "loss": 0.0462,
+      "step": 2359
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0344159032057544e-06,
+      "loss": 0.0462,
+      "step": 2360
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.026341236456017e-06,
+      "loss": 0.0655,
+      "step": 2361
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.018269645169845e-06,
+      "loss": 0.0437,
+      "step": 2362
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.010201135594354e-06,
+      "loss": 0.0542,
+      "step": 2363
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.00213571397426e-06,
+      "loss": 0.0621,
+      "step": 2364
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.994073386551904e-06,
+      "loss": 0.0475,
+      "step": 2365
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.986014159567226e-06,
+      "loss": 0.0429,
+      "step": 2366
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9779580392577675e-06,
+      "loss": 0.0637,
+      "step": 2367
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.969905031858661e-06,
+      "loss": 0.0577,
+      "step": 2368
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.961855143602642e-06,
+      "loss": 0.0486,
+      "step": 2369
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.953808380720016e-06,
+      "loss": 0.0571,
+      "step": 2370
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.945764749438678e-06,
+      "loss": 0.049,
+      "step": 2371
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9377242559841e-06,
+      "loss": 0.0557,
+      "step": 2372
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.929686906579325e-06,
+      "loss": 0.0434,
+      "step": 2373
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.9216527074449626e-06,
+      "loss": 0.045,
+      "step": 2374
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.913621664799175e-06,
+      "loss": 0.0507,
+      "step": 2375
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.905593784857695e-06,
+      "loss": 0.0492,
+      "step": 2376
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.897569073833806e-06,
+      "loss": 0.0448,
+      "step": 2377
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.889547537938327e-06,
+      "loss": 0.0594,
+      "step": 2378
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8815291833796326e-06,
+      "loss": 0.0493,
+      "step": 2379
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.873514016363631e-06,
+      "loss": 0.0434,
+      "step": 2380
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.865502043093758e-06,
+      "loss": 0.0464,
+      "step": 2381
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8574932697709815e-06,
+      "loss": 0.0527,
+      "step": 2382
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.849487702593798e-06,
+      "loss": 0.0393,
+      "step": 2383
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.841485347758213e-06,
+      "loss": 0.0415,
+      "step": 2384
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.833486211457754e-06,
+      "loss": 0.0441,
+      "step": 2385
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.825490299883452e-06,
+      "loss": 0.0479,
+      "step": 2386
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.817497619223845e-06,
+      "loss": 0.0495,
+      "step": 2387
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8095081756649655e-06,
+      "loss": 0.0427,
+      "step": 2388
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.801521975390352e-06,
+      "loss": 0.0515,
+      "step": 2389
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.793539024581021e-06,
+      "loss": 0.0474,
+      "step": 2390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.785559329415478e-06,
+      "loss": 0.0447,
+      "step": 2391
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.777582896069711e-06,
+      "loss": 0.054,
+      "step": 2392
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.769609730717182e-06,
+      "loss": 0.0494,
+      "step": 2393
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7616398395288165e-06,
+      "loss": 0.047,
+      "step": 2394
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.753673228673025e-06,
+      "loss": 0.0551,
+      "step": 2395
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.745709904315657e-06,
+      "loss": 0.0463,
+      "step": 2396
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7377498726200355e-06,
+      "loss": 0.0381,
+      "step": 2397
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.729793139746922e-06,
+      "loss": 0.0426,
+      "step": 2398
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.721839711854533e-06,
+      "loss": 0.0421,
+      "step": 2399
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.713889595098522e-06,
+      "loss": 0.0525,
+      "step": 2400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.705942795631989e-06,
+      "loss": 0.0576,
+      "step": 2401
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.697999319605456e-06,
+      "loss": 0.0568,
+      "step": 2402
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.690059173166877e-06,
+      "loss": 0.0422,
+      "step": 2403
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.682122362461632e-06,
+      "loss": 0.049,
+      "step": 2404
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.674188893632508e-06,
+      "loss": 0.0518,
+      "step": 2405
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.666258772819724e-06,
+      "loss": 0.0406,
+      "step": 2406
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.658332006160894e-06,
+      "loss": 0.0558,
+      "step": 2407
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.650408599791037e-06,
+      "loss": 0.0482,
+      "step": 2408
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.642488559842576e-06,
+      "loss": 0.0504,
+      "step": 2409
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.634571892445326e-06,
+      "loss": 0.047,
+      "step": 2410
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.626658603726489e-06,
+      "loss": 0.0491,
+      "step": 2411
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.61874869981066e-06,
+      "loss": 0.0497,
+      "step": 2412
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6108421868198085e-06,
+      "loss": 0.0505,
+      "step": 2413
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.602939070873279e-06,
+      "loss": 0.0549,
+      "step": 2414
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.595039358087788e-06,
+      "loss": 0.0455,
+      "step": 2415
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5871430545774155e-06,
+      "loss": 0.0544,
+      "step": 2416
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.5792501664536115e-06,
+      "loss": 0.0514,
+      "step": 2417
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.571360699825176e-06,
+      "loss": 0.0631,
+      "step": 2418
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.563474660798256e-06,
+      "loss": 0.0495,
+      "step": 2419
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.555592055476362e-06,
+      "loss": 0.0549,
+      "step": 2420
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.547712889960326e-06,
+      "loss": 0.0528,
+      "step": 2421
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5398371703483286e-06,
+      "loss": 0.0469,
+      "step": 2422
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.53196490273589e-06,
+      "loss": 0.0539,
+      "step": 2423
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.524096093215847e-06,
+      "loss": 0.0459,
+      "step": 2424
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.516230747878365e-06,
+      "loss": 0.0535,
+      "step": 2425
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5083688728109275e-06,
+      "loss": 0.0528,
+      "step": 2426
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.5005104740983284e-06,
+      "loss": 0.042,
+      "step": 2427
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.492655557822684e-06,
+      "loss": 0.0478,
+      "step": 2428
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4848041300633995e-06,
+      "loss": 0.0442,
+      "step": 2429
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.476956196897189e-06,
+      "loss": 0.0563,
+      "step": 2430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.469111764398062e-06,
+      "loss": 0.0618,
+      "step": 2431
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.461270838637313e-06,
+      "loss": 0.0439,
+      "step": 2432
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.453433425683523e-06,
+      "loss": 0.0413,
+      "step": 2433
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.445599531602563e-06,
+      "loss": 0.0581,
+      "step": 2434
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.437769162457576e-06,
+      "loss": 0.0481,
+      "step": 2435
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4299423243089655e-06,
+      "loss": 0.0511,
+      "step": 2436
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.422119023214428e-06,
+      "loss": 0.0503,
+      "step": 2437
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.414299265228893e-06,
+      "loss": 0.0502,
+      "step": 2438
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.406483056404563e-06,
+      "loss": 0.0552,
+      "step": 2439
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.398670402790896e-06,
+      "loss": 0.0466,
+      "step": 2440
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.390861310434594e-06,
+      "loss": 0.0465,
+      "step": 2441
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.383055785379603e-06,
+      "loss": 0.0456,
+      "step": 2442
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.375253833667104e-06,
+      "loss": 0.0479,
+      "step": 2443
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.367455461335518e-06,
+      "loss": 0.0431,
+      "step": 2444
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3596606744204985e-06,
+      "loss": 0.0456,
+      "step": 2445
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.351869478954916e-06,
+      "loss": 0.0493,
+      "step": 2446
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.344081880968861e-06,
+      "loss": 0.0543,
+      "step": 2447
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.336297886489657e-06,
+      "loss": 0.0531,
+      "step": 2448
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.328517501541812e-06,
+      "loss": 0.0482,
+      "step": 2449
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.320740732147055e-06,
+      "loss": 0.0453,
+      "step": 2450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.312967584324321e-06,
+      "loss": 0.049,
+      "step": 2451
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.305198064089734e-06,
+      "loss": 0.0491,
+      "step": 2452
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.297432177456607e-06,
+      "loss": 0.0472,
+      "step": 2453
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.289669930435462e-06,
+      "loss": 0.0511,
+      "step": 2454
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.281911329033968e-06,
+      "loss": 0.0546,
+      "step": 2455
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.274156379257006e-06,
+      "loss": 0.0489,
+      "step": 2456
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2664050871066165e-06,
+      "loss": 0.048,
+      "step": 2457
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.258657458582003e-06,
+      "loss": 0.0475,
+      "step": 2458
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.250913499679554e-06,
+      "loss": 0.0493,
+      "step": 2459
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.243173216392794e-06,
+      "loss": 0.0616,
+      "step": 2460
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2354366147124105e-06,
+      "loss": 0.0444,
+      "step": 2461
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.227703700626256e-06,
+      "loss": 0.0403,
+      "step": 2462
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.219974480119309e-06,
+      "loss": 0.0511,
+      "step": 2463
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.212248959173698e-06,
+      "loss": 0.0405,
+      "step": 2464
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.2045271437686974e-06,
+      "loss": 0.0425,
+      "step": 2465
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.196809039880689e-06,
+      "loss": 0.0537,
+      "step": 2466
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1890946534832086e-06,
+      "loss": 0.0522,
+      "step": 2467
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.1813839905469e-06,
+      "loss": 0.0572,
+      "step": 2468
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.173677057039527e-06,
+      "loss": 0.0499,
+      "step": 2469
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.165973858925965e-06,
+      "loss": 0.0473,
+      "step": 2470
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.158274402168216e-06,
+      "loss": 0.0476,
+      "step": 2471
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.150578692725355e-06,
+      "loss": 0.0482,
+      "step": 2472
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.142886736553583e-06,
+      "loss": 0.0432,
+      "step": 2473
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.135198539606186e-06,
+      "loss": 0.0471,
+      "step": 2474
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.127514107833538e-06,
+      "loss": 0.0576,
+      "step": 2475
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.119833447183112e-06,
+      "loss": 0.0435,
+      "step": 2476
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.112156563599443e-06,
+      "loss": 0.0493,
+      "step": 2477
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.1044834630241504e-06,
+      "loss": 0.0448,
+      "step": 2478
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0968141513959365e-06,
+      "loss": 0.0499,
+      "step": 2479
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.089148634650558e-06,
+      "loss": 0.048,
+      "step": 2480
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.081486918720836e-06,
+      "loss": 0.0488,
+      "step": 2481
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0738290095366616e-06,
+      "loss": 0.0455,
+      "step": 2482
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.066174913024959e-06,
+      "loss": 0.0618,
+      "step": 2483
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.058524635109722e-06,
+      "loss": 0.052,
+      "step": 2484
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.050878181711976e-06,
+      "loss": 0.0523,
+      "step": 2485
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.043235558749788e-06,
+      "loss": 0.0474,
+      "step": 2486
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.035596772138276e-06,
+      "loss": 0.0652,
+      "step": 2487
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.027961827789561e-06,
+      "loss": 0.0433,
+      "step": 2488
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.020330731612806e-06,
+      "loss": 0.0484,
+      "step": 2489
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.012703489514202e-06,
+      "loss": 0.0551,
+      "step": 2490
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.005080107396948e-06,
+      "loss": 0.0455,
+      "step": 2491
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.99746059116125e-06,
+      "loss": 0.0576,
+      "step": 2492
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.989844946704345e-06,
+      "loss": 0.0457,
+      "step": 2493
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.982233179920438e-06,
+      "loss": 0.0511,
+      "step": 2494
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.974625296700766e-06,
+      "loss": 0.0411,
+      "step": 2495
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9670213029335426e-06,
+      "loss": 0.0487,
+      "step": 2496
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.959421204503972e-06,
+      "loss": 0.0445,
+      "step": 2497
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.951825007294251e-06,
+      "loss": 0.0339,
+      "step": 2498
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.944232717183557e-06,
+      "loss": 0.0545,
+      "step": 2499
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.936644340048026e-06,
+      "loss": 0.0469,
+      "step": 2500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.929059881760789e-06,
+      "loss": 0.0473,
+      "step": 2501
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.921479348191932e-06,
+      "loss": 0.056,
+      "step": 2502
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9139027452084995e-06,
+      "loss": 0.054,
+      "step": 2503
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.906330078674512e-06,
+      "loss": 0.0559,
+      "step": 2504
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8987613544509135e-06,
+      "loss": 0.0594,
+      "step": 2505
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.891196578395629e-06,
+      "loss": 0.049,
+      "step": 2506
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.883635756363505e-06,
+      "loss": 0.0499,
+      "step": 2507
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.876078894206339e-06,
+      "loss": 0.0537,
+      "step": 2508
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8685259977728534e-06,
+      "loss": 0.0596,
+      "step": 2509
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.860977072908723e-06,
+      "loss": 0.0479,
+      "step": 2510
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8534321254565156e-06,
+      "loss": 0.0463,
+      "step": 2511
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8458911612557515e-06,
+      "loss": 0.0586,
+      "step": 2512
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.838354186142851e-06,
+      "loss": 0.0446,
+      "step": 2513
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8308212059511495e-06,
+      "loss": 0.049,
+      "step": 2514
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.8232922265108985e-06,
+      "loss": 0.0486,
+      "step": 2515
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.815767253649248e-06,
+      "loss": 0.0397,
+      "step": 2516
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.808246293190233e-06,
+      "loss": 0.0494,
+      "step": 2517
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.800729350954809e-06,
+      "loss": 0.0497,
+      "step": 2518
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.793216432760806e-06,
+      "loss": 0.0451,
+      "step": 2519
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.785707544422937e-06,
+      "loss": 0.0442,
+      "step": 2520
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.778202691752812e-06,
+      "loss": 0.0459,
+      "step": 2521
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7707018805588935e-06,
+      "loss": 0.0497,
+      "step": 2522
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.76320511664654e-06,
+      "loss": 0.0465,
+      "step": 2523
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.755712405817963e-06,
+      "loss": 0.0535,
+      "step": 2524
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.748223753872236e-06,
+      "loss": 0.0544,
+      "step": 2525
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.740739166605306e-06,
+      "loss": 0.0472,
+      "step": 2526
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.733258649809963e-06,
+      "loss": 0.0486,
+      "step": 2527
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.725782209275836e-06,
+      "loss": 0.0569,
+      "step": 2528
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.718309850789423e-06,
+      "loss": 0.0495,
+      "step": 2529
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7108415801340475e-06,
+      "loss": 0.044,
+      "step": 2530
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.703377403089867e-06,
+      "loss": 0.0495,
+      "step": 2531
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6959173254338844e-06,
+      "loss": 0.0435,
+      "step": 2532
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.688461352939916e-06,
+      "loss": 0.0533,
+      "step": 2533
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.681009491378609e-06,
+      "loss": 0.0495,
+      "step": 2534
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.673561746517423e-06,
+      "loss": 0.0419,
+      "step": 2535
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.666118124120633e-06,
+      "loss": 0.0529,
+      "step": 2536
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.658678629949333e-06,
+      "loss": 0.0466,
+      "step": 2537
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.65124326976141e-06,
+      "loss": 0.0599,
+      "step": 2538
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.643812049311548e-06,
+      "loss": 0.0539,
+      "step": 2539
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.636384974351244e-06,
+      "loss": 0.0474,
+      "step": 2540
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.628962050628774e-06,
+      "loss": 0.0525,
+      "step": 2541
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.621543283889198e-06,
+      "loss": 0.0453,
+      "step": 2542
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.6141286798743734e-06,
+      "loss": 0.053,
+      "step": 2543
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.606718244322927e-06,
+      "loss": 0.0461,
+      "step": 2544
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.599311982970248e-06,
+      "loss": 0.0517,
+      "step": 2545
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.591909901548518e-06,
+      "loss": 0.0503,
+      "step": 2546
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.584512005786667e-06,
+      "loss": 0.0472,
+      "step": 2547
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5771183014103874e-06,
+      "loss": 0.0541,
+      "step": 2548
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.569728794142136e-06,
+      "loss": 0.0626,
+      "step": 2549
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.562343489701115e-06,
+      "loss": 0.0432,
+      "step": 2550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.554962393803271e-06,
+      "loss": 0.0503,
+      "step": 2551
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.547585512161299e-06,
+      "loss": 0.0483,
+      "step": 2552
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.540212850484626e-06,
+      "loss": 0.0442,
+      "step": 2553
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.532844414479424e-06,
+      "loss": 0.044,
+      "step": 2554
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.525480209848587e-06,
+      "loss": 0.0506,
+      "step": 2555
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5181202422917244e-06,
+      "loss": 0.0465,
+      "step": 2556
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.510764517505187e-06,
+      "loss": 0.0506,
+      "step": 2557
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5034130411820275e-06,
+      "loss": 0.0411,
+      "step": 2558
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.49606581901201e-06,
+      "loss": 0.0549,
+      "step": 2559
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.488722856681619e-06,
+      "loss": 0.0455,
+      "step": 2560
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4813841598740285e-06,
+      "loss": 0.0527,
+      "step": 2561
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.474049734269116e-06,
+      "loss": 0.0471,
+      "step": 2562
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.466719585543453e-06,
+      "loss": 0.0581,
+      "step": 2563
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.459393719370297e-06,
+      "loss": 0.0528,
+      "step": 2564
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.452072141419605e-06,
+      "loss": 0.0643,
+      "step": 2565
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.444754857357998e-06,
+      "loss": 0.0458,
+      "step": 2566
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.437441872848782e-06,
+      "loss": 0.0574,
+      "step": 2567
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.430133193551934e-06,
+      "loss": 0.039,
+      "step": 2568
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.422828825124099e-06,
+      "loss": 0.051,
+      "step": 2569
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.415528773218582e-06,
+      "loss": 0.0514,
+      "step": 2570
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.408233043485359e-06,
+      "loss": 0.0508,
+      "step": 2571
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.400941641571048e-06,
+      "loss": 0.0506,
+      "step": 2572
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.39365457311892e-06,
+      "loss": 0.0463,
+      "step": 2573
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.386371843768897e-06,
+      "loss": 0.0529,
+      "step": 2574
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.379093459157538e-06,
+      "loss": 0.0454,
+      "step": 2575
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.371819424918039e-06,
+      "loss": 0.0425,
+      "step": 2576
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.364549746680237e-06,
+      "loss": 0.0539,
+      "step": 2577
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.357284430070586e-06,
+      "loss": 0.0524,
+      "step": 2578
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.350023480712173e-06,
+      "loss": 0.0468,
+      "step": 2579
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.342766904224702e-06,
+      "loss": 0.0461,
+      "step": 2580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.335514706224487e-06,
+      "loss": 0.0418,
+      "step": 2581
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.328266892324466e-06,
+      "loss": 0.0547,
+      "step": 2582
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.321023468134172e-06,
+      "loss": 0.0412,
+      "step": 2583
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.313784439259747e-06,
+      "loss": 0.0529,
+      "step": 2584
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.306549811303929e-06,
+      "loss": 0.0432,
+      "step": 2585
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2993195898660474e-06,
+      "loss": 0.0508,
+      "step": 2586
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.292093780542024e-06,
+      "loss": 0.0532,
+      "step": 2587
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.284872388924369e-06,
+      "loss": 0.0442,
+      "step": 2588
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2776554206021695e-06,
+      "loss": 0.0513,
+      "step": 2589
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.270442881161087e-06,
+      "loss": 0.0552,
+      "step": 2590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.263234776183359e-06,
+      "loss": 0.0469,
+      "step": 2591
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.256031111247786e-06,
+      "loss": 0.048,
+      "step": 2592
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.248831891929744e-06,
+      "loss": 0.0477,
+      "step": 2593
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2416371238011545e-06,
+      "loss": 0.0486,
+      "step": 2594
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2344468124305016e-06,
+      "loss": 0.0494,
+      "step": 2595
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.227260963382816e-06,
+      "loss": 0.052,
+      "step": 2596
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2200795822196775e-06,
+      "loss": 0.0492,
+      "step": 2597
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.212902674499203e-06,
+      "loss": 0.0505,
+      "step": 2598
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.20573024577606e-06,
+      "loss": 0.041,
+      "step": 2599
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.198562301601435e-06,
+      "loss": 0.0504,
+      "step": 2600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1913988475230504e-06,
+      "loss": 0.0443,
+      "step": 2601
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.184239889085152e-06,
+      "loss": 0.0502,
+      "step": 2602
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.177085431828503e-06,
+      "loss": 0.046,
+      "step": 2603
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1699354812903925e-06,
+      "loss": 0.0455,
+      "step": 2604
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.162790043004613e-06,
+      "loss": 0.0469,
+      "step": 2605
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.155649122501467e-06,
+      "loss": 0.0606,
+      "step": 2606
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.14851272530776e-06,
+      "loss": 0.0436,
+      "step": 2607
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.141380856946797e-06,
+      "loss": 0.0514,
+      "step": 2608
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.134253522938374e-06,
+      "loss": 0.0566,
+      "step": 2609
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.12713072879879e-06,
+      "loss": 0.0444,
+      "step": 2610
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.120012480040816e-06,
+      "loss": 0.057,
+      "step": 2611
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.112898782173712e-06,
+      "loss": 0.0585,
+      "step": 2612
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.105789640703215e-06,
+      "loss": 0.045,
+      "step": 2613
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.098685061131535e-06,
+      "loss": 0.0623,
+      "step": 2614
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.091585048957349e-06,
+      "loss": 0.0475,
+      "step": 2615
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.084489609675806e-06,
+      "loss": 0.0484,
+      "step": 2616
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.077398748778509e-06,
+      "loss": 0.0532,
+      "step": 2617
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.07031247175352e-06,
+      "loss": 0.0459,
+      "step": 2618
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.063230784085352e-06,
+      "loss": 0.0432,
+      "step": 2619
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.056153691254964e-06,
+      "loss": 0.0372,
+      "step": 2620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0490811987397695e-06,
+      "loss": 0.0516,
+      "step": 2621
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.042013312013609e-06,
+      "loss": 0.0453,
+      "step": 2622
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.034950036546762e-06,
+      "loss": 0.0434,
+      "step": 2623
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.027891377805943e-06,
+      "loss": 0.0429,
+      "step": 2624
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.020837341254288e-06,
+      "loss": 0.0426,
+      "step": 2625
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.0137879323513555e-06,
+      "loss": 0.048,
+      "step": 2626
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.006743156553132e-06,
+      "loss": 0.0435,
+      "step": 2627
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.999703019312007e-06,
+      "loss": 0.05,
+      "step": 2628
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9926675260767844e-06,
+      "loss": 0.049,
+      "step": 2629
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.985636682292673e-06,
+      "loss": 0.0501,
+      "step": 2630
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9786104934012795e-06,
+      "loss": 0.045,
+      "step": 2631
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9715889648406205e-06,
+      "loss": 0.0536,
+      "step": 2632
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.964572102045093e-06,
+      "loss": 0.0473,
+      "step": 2633
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.957559910445484e-06,
+      "loss": 0.0496,
+      "step": 2634
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.950552395468972e-06,
+      "loss": 0.0337,
+      "step": 2635
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.943549562539111e-06,
+      "loss": 0.0462,
+      "step": 2636
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.936551417075826e-06,
+      "loss": 0.0392,
+      "step": 2637
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.929557964495433e-06,
+      "loss": 0.0383,
+      "step": 2638
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.922569210210595e-06,
+      "loss": 0.0482,
+      "step": 2639
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.915585159630347e-06,
+      "loss": 0.0491,
+      "step": 2640
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.908605818160087e-06,
+      "loss": 0.0508,
+      "step": 2641
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.9016311912015585e-06,
+      "loss": 0.0602,
+      "step": 2642
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.89466128415287e-06,
+      "loss": 0.0491,
+      "step": 2643
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.887696102408466e-06,
+      "loss": 0.0456,
+      "step": 2644
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.880735651359137e-06,
+      "loss": 0.0581,
+      "step": 2645
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.873779936392013e-06,
+      "loss": 0.0427,
+      "step": 2646
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.866828962890557e-06,
+      "loss": 0.0404,
+      "step": 2647
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.859882736234559e-06,
+      "loss": 0.0394,
+      "step": 2648
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8529412618001485e-06,
+      "loss": 0.0422,
+      "step": 2649
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.846004544959762e-06,
+      "loss": 0.0475,
+      "step": 2650
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.839072591082159e-06,
+      "loss": 0.0463,
+      "step": 2651
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.832145405532416e-06,
+      "loss": 0.0482,
+      "step": 2652
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.825222993671912e-06,
+      "loss": 0.0505,
+      "step": 2653
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.818305360858335e-06,
+      "loss": 0.0372,
+      "step": 2654
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.811392512445682e-06,
+      "loss": 0.051,
+      "step": 2655
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.804484453784234e-06,
+      "loss": 0.0526,
+      "step": 2656
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.797581190220571e-06,
+      "loss": 0.0397,
+      "step": 2657
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.790682727097562e-06,
+      "loss": 0.0474,
+      "step": 2658
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7837890697543554e-06,
+      "loss": 0.0449,
+      "step": 2659
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.776900223526393e-06,
+      "loss": 0.0477,
+      "step": 2660
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7700161937453815e-06,
+      "loss": 0.0559,
+      "step": 2661
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7631369857392974e-06,
+      "loss": 0.0417,
+      "step": 2662
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7562626048324036e-06,
+      "loss": 0.0507,
+      "step": 2663
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7493930563452018e-06,
+      "loss": 0.0515,
+      "step": 2664
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7425283455944684e-06,
+      "loss": 0.0403,
+      "step": 2665
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7356684778932386e-06,
+      "loss": 0.0591,
+      "step": 2666
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.728813458550791e-06,
+      "loss": 0.0576,
+      "step": 2667
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7219632928726544e-06,
+      "loss": 0.0433,
+      "step": 2668
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7151179861606013e-06,
+      "loss": 0.0409,
+      "step": 2669
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7082775437126407e-06,
+      "loss": 0.0569,
+      "step": 2670
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7014419708230266e-06,
+      "loss": 0.0502,
+      "step": 2671
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6946112727822336e-06,
+      "loss": 0.0486,
+      "step": 2672
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.687785454876964e-06,
+      "loss": 0.0487,
+      "step": 2673
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.680964522390156e-06,
+      "loss": 0.0524,
+      "step": 2674
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6741484806009464e-06,
+      "loss": 0.0524,
+      "step": 2675
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6673373347846965e-06,
+      "loss": 0.0454,
+      "step": 2676
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6605310902129876e-06,
+      "loss": 0.0487,
+      "step": 2677
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.653729752153593e-06,
+      "loss": 0.043,
+      "step": 2678
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.646933325870493e-06,
+      "loss": 0.0437,
+      "step": 2679
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.640141816623879e-06,
+      "loss": 0.0395,
+      "step": 2680
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6333552296701093e-06,
+      "loss": 0.0468,
+      "step": 2681
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.626573570261761e-06,
+      "loss": 0.0618,
+      "step": 2682
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6197968436475805e-06,
+      "loss": 0.0419,
+      "step": 2683
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6130250550725034e-06,
+      "loss": 0.0425,
+      "step": 2684
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.606258209777639e-06,
+      "loss": 0.0502,
+      "step": 2685
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5994963130002745e-06,
+      "loss": 0.0535,
+      "step": 2686
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.592739369973862e-06,
+      "loss": 0.0389,
+      "step": 2687
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.585987385928029e-06,
+      "loss": 0.0344,
+      "step": 2688
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.579240366088558e-06,
+      "loss": 0.0508,
+      "step": 2689
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.572498315677384e-06,
+      "loss": 0.0441,
+      "step": 2690
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5657612399126152e-06,
+      "loss": 0.0505,
+      "step": 2691
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5590291440084847e-06,
+      "loss": 0.0467,
+      "step": 2692
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5523020331753853e-06,
+      "loss": 0.0631,
+      "step": 2693
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5455799126198553e-06,
+      "loss": 0.0513,
+      "step": 2694
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.538862787544561e-06,
+      "loss": 0.0535,
+      "step": 2695
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.532150663148305e-06,
+      "loss": 0.0542,
+      "step": 2696
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5254435446260303e-06,
+      "loss": 0.0367,
+      "step": 2697
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.518741437168782e-06,
+      "loss": 0.0488,
+      "step": 2698
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.512044345963753e-06,
+      "loss": 0.0495,
+      "step": 2699
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.505352276194237e-06,
+      "loss": 0.0586,
+      "step": 2700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4986652330396433e-06,
+      "loss": 0.0486,
+      "step": 2701
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4919832216755035e-06,
+      "loss": 0.0443,
+      "step": 2702
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4853062472734356e-06,
+      "loss": 0.0482,
+      "step": 2703
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.478634315001168e-06,
+      "loss": 0.0492,
+      "step": 2704
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4719674300225347e-06,
+      "loss": 0.0452,
+      "step": 2705
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4653055974974527e-06,
+      "loss": 0.0432,
+      "step": 2706
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.458648822581926e-06,
+      "loss": 0.0584,
+      "step": 2707
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4519971104280625e-06,
+      "loss": 0.0465,
+      "step": 2708
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4453504661840244e-06,
+      "loss": 0.0487,
+      "step": 2709
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.438708894994077e-06,
+      "loss": 0.0474,
+      "step": 2710
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.432072401998545e-06,
+      "loss": 0.0497,
+      "step": 2711
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4254409923338228e-06,
+      "loss": 0.0522,
+      "step": 2712
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.418814671132384e-06,
+      "loss": 0.0503,
+      "step": 2713
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4121934435227444e-06,
+      "loss": 0.0475,
+      "step": 2714
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.405577314629486e-06,
+      "loss": 0.0494,
+      "step": 2715
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3989662895732533e-06,
+      "loss": 0.0442,
+      "step": 2716
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.392360373470729e-06,
+      "loss": 0.0458,
+      "step": 2717
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3857595714346412e-06,
+      "loss": 0.0389,
+      "step": 2718
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3791638885737764e-06,
+      "loss": 0.0419,
+      "step": 2719
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.372573329992933e-06,
+      "loss": 0.0426,
+      "step": 2720
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3659879007929675e-06,
+      "loss": 0.0424,
+      "step": 2721
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3594076060707516e-06,
+      "loss": 0.0461,
+      "step": 2722
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.352832450919189e-06,
+      "loss": 0.0445,
+      "step": 2723
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3462624404272004e-06,
+      "loss": 0.0435,
+      "step": 2724
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3396975796797395e-06,
+      "loss": 0.0484,
+      "step": 2725
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3331378737577493e-06,
+      "loss": 0.0648,
+      "step": 2726
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3265833277382085e-06,
+      "loss": 0.0572,
+      "step": 2727
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.320033946694087e-06,
+      "loss": 0.0484,
+      "step": 2728
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3134897356943572e-06,
+      "loss": 0.0407,
+      "step": 2729
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3069506998040048e-06,
+      "loss": 0.0434,
+      "step": 2730
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.30041684408399e-06,
+      "loss": 0.0506,
+      "step": 2731
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.293888173591273e-06,
+      "loss": 0.0524,
+      "step": 2732
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2873646933788073e-06,
+      "loss": 0.0533,
+      "step": 2733
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2808464084955204e-06,
+      "loss": 0.0465,
+      "step": 2734
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2743333239863174e-06,
+      "loss": 0.0537,
+      "step": 2735
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2678254448920932e-06,
+      "loss": 0.0462,
+      "step": 2736
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.26132277624969e-06,
+      "loss": 0.038,
+      "step": 2737
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.25482532309194e-06,
+      "loss": 0.0569,
+      "step": 2738
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2483330904476275e-06,
+      "loss": 0.0575,
+      "step": 2739
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2418460833414924e-06,
+      "loss": 0.0518,
+      "step": 2740
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2353643067942453e-06,
+      "loss": 0.0583,
+      "step": 2741
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.228887765822538e-06,
+      "loss": 0.0433,
+      "step": 2742
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2224164654389623e-06,
+      "loss": 0.0593,
+      "step": 2743
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2159504106520723e-06,
+      "loss": 0.0479,
+      "step": 2744
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2094896064663494e-06,
+      "loss": 0.0477,
+      "step": 2745
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2030340578822126e-06,
+      "loss": 0.0415,
+      "step": 2746
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1965837698960257e-06,
+      "loss": 0.0462,
+      "step": 2747
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1901387475000567e-06,
+      "loss": 0.0474,
+      "step": 2748
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1836989956825226e-06,
+      "loss": 0.0496,
+      "step": 2749
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1772645194275477e-06,
+      "loss": 0.0546,
+      "step": 2750
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1708353237151725e-06,
+      "loss": 0.0479,
+      "step": 2751
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1644114135213623e-06,
+      "loss": 0.0462,
+      "step": 2752
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.157992793817983e-06,
+      "loss": 0.0427,
+      "step": 2753
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1515794695727975e-06,
+      "loss": 0.0508,
+      "step": 2754
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1451714457494887e-06,
+      "loss": 0.0511,
+      "step": 2755
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.138768727307624e-06,
+      "loss": 0.0446,
+      "step": 2756
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1323713192026663e-06,
+      "loss": 0.0573,
+      "step": 2757
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.125979226385977e-06,
+      "loss": 0.0471,
+      "step": 2758
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.119592453804797e-06,
+      "loss": 0.0443,
+      "step": 2759
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1132110064022403e-06,
+      "loss": 0.048,
+      "step": 2760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.106834889117318e-06,
+      "loss": 0.0483,
+      "step": 2761
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.100464106884905e-06,
+      "loss": 0.0438,
+      "step": 2762
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.094098664635745e-06,
+      "loss": 0.0413,
+      "step": 2763
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.087738567296463e-06,
+      "loss": 0.053,
+      "step": 2764
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.081383819789524e-06,
+      "loss": 0.0375,
+      "step": 2765
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0750344270332734e-06,
+      "loss": 0.0495,
+      "step": 2766
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.068690393941902e-06,
+      "loss": 0.0453,
+      "step": 2767
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.062351725425453e-06,
+      "loss": 0.0487,
+      "step": 2768
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.056018426389823e-06,
+      "loss": 0.0443,
+      "step": 2769
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0496905017367518e-06,
+      "loss": 0.0426,
+      "step": 2770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0433679563638042e-06,
+      "loss": 0.0488,
+      "step": 2771
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0370507951644057e-06,
+      "loss": 0.0504,
+      "step": 2772
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.030739023027801e-06,
+      "loss": 0.0439,
+      "step": 2773
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0244326448390616e-06,
+      "loss": 0.0587,
+      "step": 2774
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.018131665479095e-06,
+      "loss": 0.0425,
+      "step": 2775
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0118360898246214e-06,
+      "loss": 0.0472,
+      "step": 2776
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.005545922748181e-06,
+      "loss": 0.0466,
+      "step": 2777
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9992611691181296e-06,
+      "loss": 0.0555,
+      "step": 2778
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.992981833798627e-06,
+      "loss": 0.0431,
+      "step": 2779
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.986707921649651e-06,
+      "loss": 0.0472,
+      "step": 2780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9804394375269774e-06,
+      "loss": 0.045,
+      "step": 2781
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.974176386282167e-06,
+      "loss": 0.0408,
+      "step": 2782
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.967918772762598e-06,
+      "loss": 0.0511,
+      "step": 2783
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.961666601811427e-06,
+      "loss": 0.0452,
+      "step": 2784
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.955419878267598e-06,
+      "loss": 0.0505,
+      "step": 2785
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.949178606965847e-06,
+      "loss": 0.048,
+      "step": 2786
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.942942792736683e-06,
+      "loss": 0.0491,
+      "step": 2787
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.936712440406393e-06,
+      "loss": 0.0491,
+      "step": 2788
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.930487554797037e-06,
+      "loss": 0.0385,
+      "step": 2789
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.924268140726444e-06,
+      "loss": 0.0533,
+      "step": 2790
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.918054203008206e-06,
+      "loss": 0.0425,
+      "step": 2791
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.911845746451685e-06,
+      "loss": 0.0528,
+      "step": 2792
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.90564277586199e-06,
+      "loss": 0.0349,
+      "step": 2793
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.899445296039991e-06,
+      "loss": 0.0412,
+      "step": 2794
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.893253311782305e-06,
+      "loss": 0.0436,
+      "step": 2795
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.887066827881292e-06,
+      "loss": 0.0442,
+      "step": 2796
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.880885849125068e-06,
+      "loss": 0.041,
+      "step": 2797
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.87471038029748e-06,
+      "loss": 0.0354,
+      "step": 2798
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8685404261780993e-06,
+      "loss": 0.0346,
+      "step": 2799
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.862375991542251e-06,
+      "loss": 0.0613,
+      "step": 2800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8562170811609734e-06,
+      "loss": 0.0385,
+      "step": 2801
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.85006369980103e-06,
+      "loss": 0.0418,
+      "step": 2802
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8439158522249155e-06,
+      "loss": 0.0407,
+      "step": 2803
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8377735431908315e-06,
+      "loss": 0.0442,
+      "step": 2804
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8316367774526942e-06,
+      "loss": 0.0414,
+      "step": 2805
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8255055597601323e-06,
+      "loss": 0.047,
+      "step": 2806
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8193798948584774e-06,
+      "loss": 0.0472,
+      "step": 2807
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.813259787488769e-06,
+      "loss": 0.0502,
+      "step": 2808
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.807145242387741e-06,
+      "loss": 0.0484,
+      "step": 2809
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.801036264287822e-06,
+      "loss": 0.0362,
+      "step": 2810
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.794932857917134e-06,
+      "loss": 0.0479,
+      "step": 2811
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.788835027999485e-06,
+      "loss": 0.0441,
+      "step": 2812
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7827427792543648e-06,
+      "loss": 0.0413,
+      "step": 2813
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.776656116396952e-06,
+      "loss": 0.05,
+      "step": 2814
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.770575044138095e-06,
+      "loss": 0.0447,
+      "step": 2815
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7644995671843154e-06,
+      "loss": 0.0411,
+      "step": 2816
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7584296902378062e-06,
+      "loss": 0.0559,
+      "step": 2817
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.752365417996421e-06,
+      "loss": 0.0366,
+      "step": 2818
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7463067551536894e-06,
+      "loss": 0.0516,
+      "step": 2819
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.740253706398784e-06,
+      "loss": 0.0472,
+      "step": 2820
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.734206276416541e-06,
+      "loss": 0.0387,
+      "step": 2821
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7281644698874455e-06,
+      "loss": 0.0459,
+      "step": 2822
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.722128291487628e-06,
+      "loss": 0.045,
+      "step": 2823
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7160977458888636e-06,
+      "loss": 0.0445,
+      "step": 2824
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.710072837758575e-06,
+      "loss": 0.0369,
+      "step": 2825
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.7040535717598147e-06,
+      "loss": 0.044,
+      "step": 2826
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.698039952551269e-06,
+      "loss": 0.0566,
+      "step": 2827
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6920319847872547e-06,
+      "loss": 0.0564,
+      "step": 2828
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6860296731177136e-06,
+      "loss": 0.049,
+      "step": 2829
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6800330221882098e-06,
+      "loss": 0.0429,
+      "step": 2830
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.674042036639931e-06,
+      "loss": 0.0437,
+      "step": 2831
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6680567211096763e-06,
+      "loss": 0.0412,
+      "step": 2832
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6620770802298536e-06,
+      "loss": 0.0427,
+      "step": 2833
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6561031186284836e-06,
+      "loss": 0.0402,
+      "step": 2834
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.650134840929187e-06,
+      "loss": 0.0543,
+      "step": 2835
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6441722517511914e-06,
+      "loss": 0.0424,
+      "step": 2836
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.638215355709318e-06,
+      "loss": 0.0503,
+      "step": 2837
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6322641574139807e-06,
+      "loss": 0.0413,
+      "step": 2838
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6263186614711854e-06,
+      "loss": 0.0486,
+      "step": 2839
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.620378872482524e-06,
+      "loss": 0.0652,
+      "step": 2840
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6144447950451666e-06,
+      "loss": 0.0479,
+      "step": 2841
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6085164337518776e-06,
+      "loss": 0.0392,
+      "step": 2842
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6025937931909806e-06,
+      "loss": 0.0426,
+      "step": 2843
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5966768779463793e-06,
+      "loss": 0.062,
+      "step": 2844
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5907656925975465e-06,
+      "loss": 0.0457,
+      "step": 2845
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5848602417195144e-06,
+      "loss": 0.0593,
+      "step": 2846
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.57896052988289e-06,
+      "loss": 0.0459,
+      "step": 2847
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.573066561653824e-06,
+      "loss": 0.0528,
+      "step": 2848
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.56717834159403e-06,
+      "loss": 0.0441,
+      "step": 2849
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5612958742607697e-06,
+      "loss": 0.0614,
+      "step": 2850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.555419164206853e-06,
+      "loss": 0.0454,
+      "step": 2851
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5495482159806316e-06,
+      "loss": 0.033,
+      "step": 2852
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5436830341260067e-06,
+      "loss": 0.0451,
+      "step": 2853
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.537823623182406e-06,
+      "loss": 0.0464,
+      "step": 2854
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.531969987684796e-06,
+      "loss": 0.0467,
+      "step": 2855
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.526122132163671e-06,
+      "loss": 0.0426,
+      "step": 2856
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.520280061145051e-06,
+      "loss": 0.0455,
+      "step": 2857
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.514443779150486e-06,
+      "loss": 0.0459,
+      "step": 2858
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.508613290697037e-06,
+      "loss": 0.0436,
+      "step": 2859
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5027886002972867e-06,
+      "loss": 0.0486,
+      "step": 2860
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4969697124593263e-06,
+      "loss": 0.0456,
+      "step": 2861
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4911566316867577e-06,
+      "loss": 0.0408,
+      "step": 2862
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.485349362478686e-06,
+      "loss": 0.0442,
+      "step": 2863
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.479547909329727e-06,
+      "loss": 0.0501,
+      "step": 2864
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.473752276729986e-06,
+      "loss": 0.0498,
+      "step": 2865
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.467962469165065e-06,
+      "loss": 0.0466,
+      "step": 2866
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4621784911160595e-06,
+      "loss": 0.0636,
+      "step": 2867
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4564003470595545e-06,
+      "loss": 0.0513,
+      "step": 2868
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.450628041467612e-06,
+      "loss": 0.0523,
+      "step": 2869
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.444861578807789e-06,
+      "loss": 0.0453,
+      "step": 2870
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4391009635431085e-06,
+      "loss": 0.0457,
+      "step": 2871
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.433346200132072e-06,
+      "loss": 0.0486,
+      "step": 2872
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4275972930286506e-06,
+      "loss": 0.0496,
+      "step": 2873
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4218542466822837e-06,
+      "loss": 0.0403,
+      "step": 2874
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4161170655378774e-06,
+      "loss": 0.0437,
+      "step": 2875
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4103857540357944e-06,
+      "loss": 0.0478,
+      "step": 2876
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4046603166118556e-06,
+      "loss": 0.0488,
+      "step": 2877
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3989407576973368e-06,
+      "loss": 0.0421,
+      "step": 2878
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.393227081718961e-06,
+      "loss": 0.0557,
+      "step": 2879
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3875192930988987e-06,
+      "loss": 0.0488,
+      "step": 2880
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.381817396254772e-06,
+      "loss": 0.0553,
+      "step": 2881
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.376121395599632e-06,
+      "loss": 0.0429,
+      "step": 2882
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3704312955419693e-06,
+      "loss": 0.0465,
+      "step": 2883
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3647471004857126e-06,
+      "loss": 0.045,
+      "step": 2884
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3590688148302098e-06,
+      "loss": 0.0475,
+      "step": 2885
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3533964429702506e-06,
+      "loss": 0.0462,
+      "step": 2886
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3477299892960357e-06,
+      "loss": 0.0452,
+      "step": 2887
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3420694581931857e-06,
+      "loss": 0.0451,
+      "step": 2888
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3364148540427477e-06,
+      "loss": 0.0388,
+      "step": 2889
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.330766181221167e-06,
+      "loss": 0.0474,
+      "step": 2890
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3251234441003055e-06,
+      "loss": 0.0432,
+      "step": 2891
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3194866470474376e-06,
+      "loss": 0.0371,
+      "step": 2892
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3138557944252295e-06,
+      "loss": 0.0463,
+      "step": 2893
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3082308905917506e-06,
+      "loss": 0.0411,
+      "step": 2894
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3026119399004676e-06,
+      "loss": 0.043,
+      "step": 2895
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.296998946700236e-06,
+      "loss": 0.0547,
+      "step": 2896
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.291391915335308e-06,
+      "loss": 0.0387,
+      "step": 2897
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2857908501453143e-06,
+      "loss": 0.0535,
+      "step": 2898
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.28019575546527e-06,
+      "loss": 0.0441,
+      "step": 2899
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2746066356255692e-06,
+      "loss": 0.0473,
+      "step": 2900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2690234949519817e-06,
+      "loss": 0.0428,
+      "step": 2901
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2634463377656492e-06,
+      "loss": 0.0493,
+      "step": 2902
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.257875168383087e-06,
+      "loss": 0.0425,
+      "step": 2903
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2523099911161705e-06,
+      "loss": 0.0404,
+      "step": 2904
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2467508102721346e-06,
+      "loss": 0.0503,
+      "step": 2905
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2411976301535886e-06,
+      "loss": 0.0539,
+      "step": 2906
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2356504550584766e-06,
+      "loss": 0.0549,
+      "step": 2907
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.230109289280106e-06,
+      "loss": 0.055,
+      "step": 2908
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.224574137107136e-06,
+      "loss": 0.047,
+      "step": 2909
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.219045002823568e-06,
+      "loss": 0.0397,
+      "step": 2910
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.213521890708743e-06,
+      "loss": 0.0516,
+      "step": 2911
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2080048050373447e-06,
+      "loss": 0.0529,
+      "step": 2912
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2024937500793887e-06,
+      "loss": 0.0487,
+      "step": 2913
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1969887301002314e-06,
+      "loss": 0.0491,
+      "step": 2914
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.191489749360549e-06,
+      "loss": 0.0453,
+      "step": 2915
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1859968121163465e-06,
+      "loss": 0.0501,
+      "step": 2916
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1805099226189595e-06,
+      "loss": 0.0444,
+      "step": 2917
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.175029085115027e-06,
+      "loss": 0.0443,
+      "step": 2918
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1695543038465128e-06,
+      "loss": 0.0481,
+      "step": 2919
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1640855830507e-06,
+      "loss": 0.0476,
+      "step": 2920
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1586229269601698e-06,
+      "loss": 0.0457,
+      "step": 2921
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1531663398028136e-06,
+      "loss": 0.0538,
+      "step": 2922
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.147715825801832e-06,
+      "loss": 0.0377,
+      "step": 2923
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1422713891757095e-06,
+      "loss": 0.0412,
+      "step": 2924
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.136833034138246e-06,
+      "loss": 0.0487,
+      "step": 2925
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.13140076489852e-06,
+      "loss": 0.045,
+      "step": 2926
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.125974585660906e-06,
+      "loss": 0.0481,
+      "step": 2927
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1205545006250694e-06,
+      "loss": 0.0491,
+      "step": 2928
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.115140513985946e-06,
+      "loss": 0.0496,
+      "step": 2929
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1097326299337606e-06,
+      "loss": 0.0593,
+      "step": 2930
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1043308526540183e-06,
+      "loss": 0.0473,
+      "step": 2931
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0989351863274908e-06,
+      "loss": 0.043,
+      "step": 2932
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.09354563513022e-06,
+      "loss": 0.0435,
+      "step": 2933
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0881622032335236e-06,
+      "loss": 0.05,
+      "step": 2934
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.082784894803971e-06,
+      "loss": 0.0495,
+      "step": 2935
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.077413714003397e-06,
+      "loss": 0.0462,
+      "step": 2936
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0720486649889005e-06,
+      "loss": 0.0372,
+      "step": 2937
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.066689751912827e-06,
+      "loss": 0.0469,
+      "step": 2938
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0613369789227756e-06,
+      "loss": 0.0502,
+      "step": 2939
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0559903501615917e-06,
+      "loss": 0.0469,
+      "step": 2940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.050649869767364e-06,
+      "loss": 0.0381,
+      "step": 2941
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0453155418734317e-06,
+      "loss": 0.0464,
+      "step": 2942
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0399873706083607e-06,
+      "loss": 0.0512,
+      "step": 2943
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0346653600959543e-06,
+      "loss": 0.0494,
+      "step": 2944
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.02934951445526e-06,
+      "loss": 0.0521,
+      "step": 2945
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0240398378005342e-06,
+      "loss": 0.0439,
+      "step": 2946
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.018736334241268e-06,
+      "loss": 0.0441,
+      "step": 2947
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.013439007882183e-06,
+      "loss": 0.0425,
+      "step": 2948
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0081478628232064e-06,
+      "loss": 0.0473,
+      "step": 2949
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.002862903159486e-06,
+      "loss": 0.0372,
+      "step": 2950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9975841329813916e-06,
+      "loss": 0.0467,
+      "step": 2951
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9923115563744834e-06,
+      "loss": 0.0398,
+      "step": 2952
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9870451774195452e-06,
+      "loss": 0.0404,
+      "step": 2953
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9817850001925577e-06,
+      "loss": 0.053,
+      "step": 2954
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9765310287646966e-06,
+      "loss": 0.0444,
+      "step": 2955
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9712832672023494e-06,
+      "loss": 0.0473,
+      "step": 2956
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.96604171956708e-06,
+      "loss": 0.0488,
+      "step": 2957
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9608063899156483e-06,
+      "loss": 0.0529,
+      "step": 2958
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.955577282300012e-06,
+      "loss": 0.0501,
+      "step": 2959
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9503544007673015e-06,
+      "loss": 0.0435,
+      "step": 2960
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9451377493598288e-06,
+      "loss": 0.0496,
+      "step": 2961
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9399273321150956e-06,
+      "loss": 0.0422,
+      "step": 2962
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9347231530657606e-06,
+      "loss": 0.0435,
+      "step": 2963
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9295252162396717e-06,
+      "loss": 0.0493,
+      "step": 2964
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.924333525659835e-06,
+      "loss": 0.0565,
+      "step": 2965
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9191480853444254e-06,
+      "loss": 0.0542,
+      "step": 2966
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9139688993067786e-06,
+      "loss": 0.044,
+      "step": 2967
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.908795971555397e-06,
+      "loss": 0.0525,
+      "step": 2968
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9036293060939248e-06,
+      "loss": 0.0497,
+      "step": 2969
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8984689069211758e-06,
+      "loss": 0.0594,
+      "step": 2970
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8933147780311034e-06,
+      "loss": 0.0397,
+      "step": 2971
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8881669234128074e-06,
+      "loss": 0.0568,
+      "step": 2972
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8830253470505456e-06,
+      "loss": 0.0419,
+      "step": 2973
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8778900529236953e-06,
+      "loss": 0.0429,
+      "step": 2974
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8727610450067857e-06,
+      "loss": 0.0436,
+      "step": 2975
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8676383272694799e-06,
+      "loss": 0.0485,
+      "step": 2976
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8625219036765685e-06,
+      "loss": 0.0543,
+      "step": 2977
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8574117781879697e-06,
+      "loss": 0.0498,
+      "step": 2978
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8523079547587385e-06,
+      "loss": 0.0405,
+      "step": 2979
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8472104373390331e-06,
+      "loss": 0.0349,
+      "step": 2980
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8421192298741487e-06,
+      "loss": 0.0486,
+      "step": 2981
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8370343363044884e-06,
+      "loss": 0.0526,
+      "step": 2982
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8319557605655658e-06,
+      "loss": 0.0463,
+      "step": 2983
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8268835065880163e-06,
+      "loss": 0.0507,
+      "step": 2984
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8218175782975734e-06,
+      "loss": 0.0531,
+      "step": 2985
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8167579796150692e-06,
+      "loss": 0.0461,
+      "step": 2986
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8117047144564526e-06,
+      "loss": 0.0549,
+      "step": 2987
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8066577867327595e-06,
+      "loss": 0.0543,
+      "step": 2988
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8016172003501198e-06,
+      "loss": 0.0439,
+      "step": 2989
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7965829592097683e-06,
+      "loss": 0.0498,
+      "step": 2990
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7915550672080085e-06,
+      "loss": 0.0496,
+      "step": 2991
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7865335282362494e-06,
+      "loss": 0.049,
+      "step": 2992
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7815183461809727e-06,
+      "loss": 0.0344,
+      "step": 2993
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7765095249237384e-06,
+      "loss": 0.048,
+      "step": 2994
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7715070683411939e-06,
+      "loss": 0.049,
+      "step": 2995
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7665109803050517e-06,
+      "loss": 0.049,
+      "step": 2996
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7615212646820911e-06,
+      "loss": 0.0443,
+      "step": 2997
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.756537925334174e-06,
+      "loss": 0.0487,
+      "step": 2998
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7515609661182153e-06,
+      "loss": 0.0337,
+      "step": 2999
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7465903908861936e-06,
+      "loss": 0.0448,
+      "step": 3000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7416262034851538e-06,
+      "loss": 0.0435,
+      "step": 3001
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7366684077571893e-06,
+      "loss": 0.0523,
+      "step": 3002
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7317170075394484e-06,
+      "loss": 0.0445,
+      "step": 3003
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7267720066641302e-06,
+      "loss": 0.0517,
+      "step": 3004
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7218334089584799e-06,
+      "loss": 0.0512,
+      "step": 3005
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7169012182447874e-06,
+      "loss": 0.0538,
+      "step": 3006
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.711975438340391e-06,
+      "loss": 0.0504,
+      "step": 3007
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.707056073057649e-06,
+      "loss": 0.0568,
+      "step": 3008
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7021431262039756e-06,
+      "loss": 0.0516,
+      "step": 3009
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6972366015818055e-06,
+      "loss": 0.0461,
+      "step": 3010
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.692336502988603e-06,
+      "loss": 0.0464,
+      "step": 3011
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.687442834216867e-06,
+      "loss": 0.0505,
+      "step": 3012
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.682555599054112e-06,
+      "loss": 0.0433,
+      "step": 3013
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6776748012828713e-06,
+      "loss": 0.0477,
+      "step": 3014
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6728004446807066e-06,
+      "loss": 0.0412,
+      "step": 3015
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6679325330201855e-06,
+      "loss": 0.0396,
+      "step": 3016
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6630710700688867e-06,
+      "loss": 0.042,
+      "step": 3017
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.658216059589407e-06,
+      "loss": 0.0485,
+      "step": 3018
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6533675053393395e-06,
+      "loss": 0.0485,
+      "step": 3019
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6485254110712856e-06,
+      "loss": 0.0488,
+      "step": 3020
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6436897805328456e-06,
+      "loss": 0.043,
+      "step": 3021
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6388606174666134e-06,
+      "loss": 0.0455,
+      "step": 3022
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6340379256101868e-06,
+      "loss": 0.0502,
+      "step": 3023
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6292217086961482e-06,
+      "loss": 0.0445,
+      "step": 3024
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6244119704520645e-06,
+      "loss": 0.0442,
+      "step": 3025
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.619608714600498e-06,
+      "loss": 0.0424,
+      "step": 3026
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6148119448589893e-06,
+      "loss": 0.0419,
+      "step": 3027
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.610021664940057e-06,
+      "loss": 0.0516,
+      "step": 3028
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6052378785512023e-06,
+      "loss": 0.0556,
+      "step": 3029
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6004605893948944e-06,
+      "loss": 0.0518,
+      "step": 3030
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5956898011685784e-06,
+      "loss": 0.044,
+      "step": 3031
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5909255175646665e-06,
+      "loss": 0.0417,
+      "step": 3032
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5861677422705324e-06,
+      "loss": 0.0422,
+      "step": 3033
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5814164789685227e-06,
+      "loss": 0.0404,
+      "step": 3034
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5766717313359347e-06,
+      "loss": 0.0493,
+      "step": 3035
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5719335030450255e-06,
+      "loss": 0.0409,
+      "step": 3036
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5672017977630082e-06,
+      "loss": 0.0454,
+      "step": 3037
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5624766191520457e-06,
+      "loss": 0.039,
+      "step": 3038
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.557757970869247e-06,
+      "loss": 0.0514,
+      "step": 3039
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5530458565666762e-06,
+      "loss": 0.0495,
+      "step": 3040
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5483402798913283e-06,
+      "loss": 0.0456,
+      "step": 3041
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5436412444851468e-06,
+      "loss": 0.0518,
+      "step": 3042
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.538948753985009e-06,
+      "loss": 0.0554,
+      "step": 3043
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5342628120227266e-06,
+      "loss": 0.0479,
+      "step": 3044
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5295834222250428e-06,
+      "loss": 0.0388,
+      "step": 3045
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5249105882136339e-06,
+      "loss": 0.0469,
+      "step": 3046
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.520244313605097e-06,
+      "loss": 0.0448,
+      "step": 3047
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5155846020109555e-06,
+      "loss": 0.0521,
+      "step": 3048
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.510931457037651e-06,
+      "loss": 0.039,
+      "step": 3049
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5062848822865429e-06,
+      "loss": 0.0418,
+      "step": 3050
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5016448813539097e-06,
+      "loss": 0.04,
+      "step": 3051
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4970114578309404e-06,
+      "loss": 0.052,
+      "step": 3052
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4923846153037247e-06,
+      "loss": 0.042,
+      "step": 3053
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4877643573532718e-06,
+      "loss": 0.043,
+      "step": 3054
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.483150687555488e-06,
+      "loss": 0.0441,
+      "step": 3055
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.478543609481179e-06,
+      "loss": 0.0429,
+      "step": 3056
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4739431266960548e-06,
+      "loss": 0.0425,
+      "step": 3057
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4693492427607158e-06,
+      "loss": 0.0381,
+      "step": 3058
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4647619612306563e-06,
+      "loss": 0.0434,
+      "step": 3059
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4601812856562625e-06,
+      "loss": 0.0397,
+      "step": 3060
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.455607219582802e-06,
+      "loss": 0.0596,
+      "step": 3061
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4510397665504372e-06,
+      "loss": 0.0384,
+      "step": 3062
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4464789300942028e-06,
+      "loss": 0.0597,
+      "step": 3063
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4419247137440173e-06,
+      "loss": 0.0468,
+      "step": 3064
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4373771210246733e-06,
+      "loss": 0.0493,
+      "step": 3065
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4328361554558378e-06,
+      "loss": 0.0424,
+      "step": 3066
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4283018205520471e-06,
+      "loss": 0.0484,
+      "step": 3067
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4237741198227117e-06,
+      "loss": 0.0466,
+      "step": 3068
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4192530567720996e-06,
+      "loss": 0.0433,
+      "step": 3069
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4147386348993452e-06,
+      "loss": 0.0553,
+      "step": 3070
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4102308576984425e-06,
+      "loss": 0.0536,
+      "step": 3071
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4057297286582416e-06,
+      "loss": 0.0429,
+      "step": 3072
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4012352512624516e-06,
+      "loss": 0.0497,
+      "step": 3073
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3967474289896299e-06,
+      "loss": 0.0424,
+      "step": 3074
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3922662653131802e-06,
+      "loss": 0.0456,
+      "step": 3075
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3877917637013593e-06,
+      "loss": 0.0567,
+      "step": 3076
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3833239276172617e-06,
+      "loss": 0.038,
+      "step": 3077
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3788627605188253e-06,
+      "loss": 0.0457,
+      "step": 3078
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.374408265858831e-06,
+      "loss": 0.0408,
+      "step": 3079
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3699604470848892e-06,
+      "loss": 0.0642,
+      "step": 3080
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3655193076394456e-06,
+      "loss": 0.0509,
+      "step": 3081
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3610848509597752e-06,
+      "loss": 0.052,
+      "step": 3082
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3566570804779843e-06,
+      "loss": 0.0441,
+      "step": 3083
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3522359996209977e-06,
+      "loss": 0.0484,
+      "step": 3084
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3478216118105713e-06,
+      "loss": 0.0414,
+      "step": 3085
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3434139204632769e-06,
+      "loss": 0.0465,
+      "step": 3086
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3390129289904997e-06,
+      "loss": 0.0455,
+      "step": 3087
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3346186407984452e-06,
+      "loss": 0.055,
+      "step": 3088
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3302310592881251e-06,
+      "loss": 0.0431,
+      "step": 3089
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3258501878553687e-06,
+      "loss": 0.0515,
+      "step": 3090
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3214760298908048e-06,
+      "loss": 0.0454,
+      "step": 3091
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3171085887798684e-06,
+      "loss": 0.0539,
+      "step": 3092
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3127478679027949e-06,
+      "loss": 0.0377,
+      "step": 3093
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.308393870634621e-06,
+      "loss": 0.0552,
+      "step": 3094
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3040466003451746e-06,
+      "loss": 0.0498,
+      "step": 3095
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2997060603990853e-06,
+      "loss": 0.0543,
+      "step": 3096
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.295372254155768e-06,
+      "loss": 0.0427,
+      "step": 3097
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.291045184969425e-06,
+      "loss": 0.0489,
+      "step": 3098
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.286724856189049e-06,
+      "loss": 0.0439,
+      "step": 3099
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.282411271158408e-06,
+      "loss": 0.0415,
+      "step": 3100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.278104433216062e-06,
+      "loss": 0.0382,
+      "step": 3101
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2738043456953408e-06,
+      "loss": 0.0447,
+      "step": 3102
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2695110119243504e-06,
+      "loss": 0.048,
+      "step": 3103
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2652244352259712e-06,
+      "loss": 0.048,
+      "step": 3104
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2609446189178554e-06,
+      "loss": 0.041,
+      "step": 3105
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2566715663124162e-06,
+      "loss": 0.039,
+      "step": 3106
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2524052807168429e-06,
+      "loss": 0.0476,
+      "step": 3107
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2481457654330787e-06,
+      "loss": 0.0521,
+      "step": 3108
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2438930237578294e-06,
+      "loss": 0.0518,
+      "step": 3109
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2396470589825582e-06,
+      "loss": 0.0513,
+      "step": 3110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2354078743934816e-06,
+      "loss": 0.0469,
+      "step": 3111
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2311754732715743e-06,
+      "loss": 0.0446,
+      "step": 3112
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2269498588925543e-06,
+      "loss": 0.0396,
+      "step": 3113
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2227310345268895e-06,
+      "loss": 0.0491,
+      "step": 3114
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2185190034397931e-06,
+      "loss": 0.0406,
+      "step": 3115
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2143137688912198e-06,
+      "loss": 0.0557,
+      "step": 3116
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2101153341358618e-06,
+      "loss": 0.0494,
+      "step": 3117
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2059237024231552e-06,
+      "loss": 0.0404,
+      "step": 3118
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2017388769972638e-06,
+      "loss": 0.0423,
+      "step": 3119
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.197560861097087e-06,
+      "loss": 0.0469,
+      "step": 3120
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1933896579562531e-06,
+      "loss": 0.0452,
+      "step": 3121
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1892252708031172e-06,
+      "loss": 0.048,
+      "step": 3122
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1850677028607583e-06,
+      "loss": 0.0499,
+      "step": 3123
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1809169573469826e-06,
+      "loss": 0.0529,
+      "step": 3124
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1767730374743093e-06,
+      "loss": 0.041,
+      "step": 3125
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1726359464499792e-06,
+      "loss": 0.0408,
+      "step": 3126
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1685056874759447e-06,
+      "loss": 0.0487,
+      "step": 3127
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.164382263748871e-06,
+      "loss": 0.0504,
+      "step": 3128
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1602656784601385e-06,
+      "loss": 0.0443,
+      "step": 3129
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1561559347958273e-06,
+      "loss": 0.0427,
+      "step": 3130
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1520530359367254e-06,
+      "loss": 0.0562,
+      "step": 3131
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.147956985058327e-06,
+      "loss": 0.0487,
+      "step": 3132
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.14386778533082e-06,
+      "loss": 0.0429,
+      "step": 3133
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1397854399190889e-06,
+      "loss": 0.0425,
+      "step": 3134
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1357099519827219e-06,
+      "loss": 0.0481,
+      "step": 3135
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1316413246759938e-06,
+      "loss": 0.0446,
+      "step": 3136
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.127579561147869e-06,
+      "loss": 0.046,
+      "step": 3137
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1235246645420005e-06,
+      "loss": 0.0578,
+      "step": 3138
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.119476637996726e-06,
+      "loss": 0.0401,
+      "step": 3139
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.11543548464507e-06,
+      "loss": 0.0406,
+      "step": 3140
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.111401207614733e-06,
+      "loss": 0.0589,
+      "step": 3141
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.107373810028095e-06,
+      "loss": 0.0418,
+      "step": 3142
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1033532950022107e-06,
+      "loss": 0.0376,
+      "step": 3143
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0993396656488099e-06,
+      "loss": 0.0512,
+      "step": 3144
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0953329250742884e-06,
+      "loss": 0.0468,
+      "step": 3145
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.091333076379718e-06,
+      "loss": 0.0477,
+      "step": 3146
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0873401226608316e-06,
+      "loss": 0.0472,
+      "step": 3147
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0833540670080233e-06,
+      "loss": 0.0394,
+      "step": 3148
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.079374912506358e-06,
+      "loss": 0.0519,
+      "step": 3149
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0754026622355463e-06,
+      "loss": 0.045,
+      "step": 3150
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0714373192699612e-06,
+      "loss": 0.0508,
+      "step": 3151
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0674788866786356e-06,
+      "loss": 0.0417,
+      "step": 3152
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0635273675252444e-06,
+      "loss": 0.0435,
+      "step": 3153
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0595827648681179e-06,
+      "loss": 0.0391,
+      "step": 3154
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0556450817602316e-06,
+      "loss": 0.0406,
+      "step": 3155
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.051714321249201e-06,
+      "loss": 0.0416,
+      "step": 3156
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0477904863772936e-06,
+      "loss": 0.0465,
+      "step": 3157
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0438735801814092e-06,
+      "loss": 0.0391,
+      "step": 3158
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0399636056930839e-06,
+      "loss": 0.0467,
+      "step": 3159
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0360605659384993e-06,
+      "loss": 0.0473,
+      "step": 3160
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0321644639384564e-06,
+      "loss": 0.0509,
+      "step": 3161
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0282753027083914e-06,
+      "loss": 0.0503,
+      "step": 3162
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.024393085258375e-06,
+      "loss": 0.0417,
+      "step": 3163
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0205178145930972e-06,
+      "loss": 0.0411,
+      "step": 3164
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0166494937118708e-06,
+      "loss": 0.0525,
+      "step": 3165
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0127881256086337e-06,
+      "loss": 0.0535,
+      "step": 3166
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0089337132719378e-06,
+      "loss": 0.0534,
+      "step": 3167
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0050862596849597e-06,
+      "loss": 0.0481,
+      "step": 3168
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0012457678254806e-06,
+      "loss": 0.0432,
+      "step": 3169
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.97412240665897e-07,
+      "loss": 0.0434,
+      "step": 3170
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.935856811732224e-07,
+      "loss": 0.0412,
+      "step": 3171
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.897660923090646e-07,
+      "loss": 0.0386,
+      "step": 3172
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.85953477029642e-07,
+      "loss": 0.0489,
+      "step": 3173
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.82147838285782e-07,
+      "loss": 0.0392,
+      "step": 3174
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.783491790229027e-07,
+      "loss": 0.0453,
+      "step": 3175
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.745575021810261e-07,
+      "loss": 0.0478,
+      "step": 3176
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.70772810694771e-07,
+      "loss": 0.0417,
+      "step": 3177
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.669951074933414e-07,
+      "loss": 0.0469,
+      "step": 3178
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.632243955005438e-07,
+      "loss": 0.0538,
+      "step": 3179
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.59460677634767e-07,
+      "loss": 0.0418,
+      "step": 3180
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.557039568089876e-07,
+      "loss": 0.05,
+      "step": 3181
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.519542359307676e-07,
+      "loss": 0.0615,
+      "step": 3182
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.482115179022511e-07,
+      "loss": 0.0398,
+      "step": 3183
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.444758056201608e-07,
+      "loss": 0.0536,
+      "step": 3184
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.407471019758007e-07,
+      "loss": 0.0413,
+      "step": 3185
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.370254098550479e-07,
+      "loss": 0.0511,
+      "step": 3186
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.333107321383505e-07,
+      "loss": 0.0449,
+      "step": 3187
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.296030717007376e-07,
+      "loss": 0.0456,
+      "step": 3188
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.259024314117914e-07,
+      "loss": 0.038,
+      "step": 3189
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.222088141356722e-07,
+      "loss": 0.0438,
+      "step": 3190
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.185222227311041e-07,
+      "loss": 0.0517,
+      "step": 3191
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.148426600513693e-07,
+      "loss": 0.0494,
+      "step": 3192
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.111701289443109e-07,
+      "loss": 0.049,
+      "step": 3193
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.075046322523362e-07,
+      "loss": 0.0541,
+      "step": 3194
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.03846172812396e-07,
+      "loss": 0.0499,
+      "step": 3195
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.001947534560052e-07,
+      "loss": 0.0546,
+      "step": 3196
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.965503770092265e-07,
+      "loss": 0.0397,
+      "step": 3197
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.929130462926672e-07,
+      "loss": 0.0485,
+      "step": 3198
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.89282764121493e-07,
+      "loss": 0.0449,
+      "step": 3199
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.856595333053996e-07,
+      "loss": 0.0438,
+      "step": 3200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.820433566486331e-07,
+      "loss": 0.0576,
+      "step": 3201
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.784342369499832e-07,
+      "loss": 0.0517,
+      "step": 3202
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.748321770027712e-07,
+      "loss": 0.044,
+      "step": 3203
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.71237179594857e-07,
+      "loss": 0.0434,
+      "step": 3204
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.676492475086384e-07,
+      "loss": 0.0606,
+      "step": 3205
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.640683835210328e-07,
+      "loss": 0.0483,
+      "step": 3206
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.604945904035033e-07,
+      "loss": 0.0471,
+      "step": 3207
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.569278709220286e-07,
+      "loss": 0.0476,
+      "step": 3208
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.533682278371147e-07,
+      "loss": 0.046,
+      "step": 3209
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.498156639037969e-07,
+      "loss": 0.0411,
+      "step": 3210
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.462701818716267e-07,
+      "loss": 0.0538,
+      "step": 3211
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.427317844846694e-07,
+      "loss": 0.0448,
+      "step": 3212
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.392004744815185e-07,
+      "loss": 0.0358,
+      "step": 3213
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.356762545952734e-07,
+      "loss": 0.0417,
+      "step": 3214
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.321591275535479e-07,
+      "loss": 0.0374,
+      "step": 3215
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.28649096078471e-07,
+      "loss": 0.0534,
+      "step": 3216
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.251461628866708e-07,
+      "loss": 0.0506,
+      "step": 3217
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.216503306892909e-07,
+      "loss": 0.0438,
+      "step": 3218
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.181616021919725e-07,
+      "loss": 0.0508,
+      "step": 3219
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.146799800948635e-07,
+      "loss": 0.0436,
+      "step": 3220
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.112054670926062e-07,
+      "loss": 0.0491,
+      "step": 3221
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.077380658743495e-07,
+      "loss": 0.0402,
+      "step": 3222
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.042777791237245e-07,
+      "loss": 0.0502,
+      "step": 3223
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.008246095188709e-07,
+      "loss": 0.0394,
+      "step": 3224
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.973785597324113e-07,
+      "loss": 0.039,
+      "step": 3225
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.939396324314564e-07,
+      "loss": 0.0441,
+      "step": 3226
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.905078302776115e-07,
+      "loss": 0.0436,
+      "step": 3227
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.870831559269643e-07,
+      "loss": 0.0424,
+      "step": 3228
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.83665612030079e-07,
+      "loss": 0.049,
+      "step": 3229
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.80255201232013e-07,
+      "loss": 0.0515,
+      "step": 3230
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.76851926172294e-07,
+      "loss": 0.0405,
+      "step": 3231
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.734557894849292e-07,
+      "loss": 0.036,
+      "step": 3232
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.700667937984052e-07,
+      "loss": 0.0517,
+      "step": 3233
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.666849417356736e-07,
+      "loss": 0.0487,
+      "step": 3234
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.633102359141653e-07,
+      "loss": 0.0527,
+      "step": 3235
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.599426789457753e-07,
+      "loss": 0.0473,
+      "step": 3236
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.565822734368656e-07,
+      "loss": 0.0443,
+      "step": 3237
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.532290219882676e-07,
+      "loss": 0.0508,
+      "step": 3238
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.498829271952723e-07,
+      "loss": 0.0475,
+      "step": 3239
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.46543991647628e-07,
+      "loss": 0.0451,
+      "step": 3240
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.432122179295509e-07,
+      "loss": 0.0393,
+      "step": 3241
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.398876086197082e-07,
+      "loss": 0.0475,
+      "step": 3242
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.365701662912206e-07,
+      "loss": 0.0394,
+      "step": 3243
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.332598935116698e-07,
+      "loss": 0.0578,
+      "step": 3244
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.299567928430807e-07,
+      "loss": 0.0488,
+      "step": 3245
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.266608668419295e-07,
+      "loss": 0.0476,
+      "step": 3246
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.233721180591402e-07,
+      "loss": 0.052,
+      "step": 3247
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.2009054904008e-07,
+      "loss": 0.0539,
+      "step": 3248
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.16816162324564e-07,
+      "loss": 0.0462,
+      "step": 3249
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.13548960446846e-07,
+      "loss": 0.0504,
+      "step": 3250
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.102889459356122e-07,
+      "loss": 0.0477,
+      "step": 3251
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.070361213139986e-07,
+      "loss": 0.0417,
+      "step": 3252
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.037904890995673e-07,
+      "loss": 0.0397,
+      "step": 3253
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.005520518043152e-07,
+      "loss": 0.0385,
+      "step": 3254
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.973208119346742e-07,
+      "loss": 0.0406,
+      "step": 3255
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.940967719915037e-07,
+      "loss": 0.0367,
+      "step": 3256
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.90879934470089e-07,
+      "loss": 0.0532,
+      "step": 3257
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.876703018601416e-07,
+      "loss": 0.0539,
+      "step": 3258
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.844678766457991e-07,
+      "loss": 0.0515,
+      "step": 3259
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.812726613056165e-07,
+      "loss": 0.0375,
+      "step": 3260
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.780846583125744e-07,
+      "loss": 0.0364,
+      "step": 3261
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.74903870134066e-07,
+      "loss": 0.0504,
+      "step": 3262
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.717302992319041e-07,
+      "loss": 0.0455,
+      "step": 3263
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.68563948062313e-07,
+      "loss": 0.0435,
+      "step": 3264
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.654048190759288e-07,
+      "loss": 0.0446,
+      "step": 3265
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.622529147178036e-07,
+      "loss": 0.0387,
+      "step": 3266
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.591082374273916e-07,
+      "loss": 0.0445,
+      "step": 3267
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.559707896385526e-07,
+      "loss": 0.0473,
+      "step": 3268
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.528405737795596e-07,
+      "loss": 0.0461,
+      "step": 3269
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.497175922730792e-07,
+      "loss": 0.0446,
+      "step": 3270
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.466018475361813e-07,
+      "loss": 0.0477,
+      "step": 3271
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.434933419803402e-07,
+      "loss": 0.0536,
+      "step": 3272
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.403920780114215e-07,
+      "loss": 0.0527,
+      "step": 3273
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.372980580296872e-07,
+      "loss": 0.0348,
+      "step": 3274
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.342112844297932e-07,
+      "loss": 0.0391,
+      "step": 3275
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.31131759600786e-07,
+      "loss": 0.0609,
+      "step": 3276
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.280594859261058e-07,
+      "loss": 0.0475,
+      "step": 3277
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.249944657835761e-07,
+      "loss": 0.0416,
+      "step": 3278
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.219367015454081e-07,
+      "loss": 0.0443,
+      "step": 3279
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.188861955781967e-07,
+      "loss": 0.044,
+      "step": 3280
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.158429502429208e-07,
+      "loss": 0.049,
+      "step": 3281
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.128069678949366e-07,
+      "loss": 0.0407,
+      "step": 3282
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.097782508839845e-07,
+      "loss": 0.0564,
+      "step": 3283
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.067568015541758e-07,
+      "loss": 0.0479,
+      "step": 3284
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.037426222440002e-07,
+      "loss": 0.0442,
+      "step": 3285
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.007357152863202e-07,
+      "loss": 0.0393,
+      "step": 3286
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.977360830083679e-07,
+      "loss": 0.0424,
+      "step": 3287
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.947437277317503e-07,
+      "loss": 0.0424,
+      "step": 3288
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.917586517724361e-07,
+      "loss": 0.0581,
+      "step": 3289
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.887808574407627e-07,
+      "loss": 0.0412,
+      "step": 3290
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.858103470414322e-07,
+      "loss": 0.0451,
+      "step": 3291
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.828471228735099e-07,
+      "loss": 0.0495,
+      "step": 3292
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.79891187230417e-07,
+      "loss": 0.0324,
+      "step": 3293
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.769425423999419e-07,
+      "loss": 0.0377,
+      "step": 3294
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.740011906642251e-07,
+      "loss": 0.0479,
+      "step": 3295
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.710671342997587e-07,
+      "loss": 0.0452,
+      "step": 3296
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.681403755773973e-07,
+      "loss": 0.0505,
+      "step": 3297
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.652209167623413e-07,
+      "loss": 0.0416,
+      "step": 3298
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.623087601141419e-07,
+      "loss": 0.0409,
+      "step": 3299
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.594039078867019e-07,
+      "loss": 0.0492,
+      "step": 3300
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.565063623282674e-07,
+      "loss": 0.0415,
+      "step": 3301
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.536161256814299e-07,
+      "loss": 0.0366,
+      "step": 3302
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.507332001831256e-07,
+      "loss": 0.0452,
+      "step": 3303
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.478575880646297e-07,
+      "loss": 0.0396,
+      "step": 3304
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.449892915515608e-07,
+      "loss": 0.0429,
+      "step": 3305
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.421283128638721e-07,
+      "loss": 0.0598,
+      "step": 3306
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.392746542158544e-07,
+      "loss": 0.0344,
+      "step": 3307
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.364283178161334e-07,
+      "loss": 0.0556,
+      "step": 3308
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.335893058676667e-07,
+      "loss": 0.0465,
+      "step": 3309
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.307576205677434e-07,
+      "loss": 0.0567,
+      "step": 3310
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.279332641079826e-07,
+      "loss": 0.0456,
+      "step": 3311
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.251162386743314e-07,
+      "loss": 0.0447,
+      "step": 3312
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.223065464470623e-07,
+      "loss": 0.0427,
+      "step": 3313
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.195041896007713e-07,
+      "loss": 0.048,
+      "step": 3314
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.167091703043769e-07,
+      "loss": 0.0438,
+      "step": 3315
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.139214907211232e-07,
+      "loss": 0.0435,
+      "step": 3316
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.111411530085675e-07,
+      "loss": 0.0554,
+      "step": 3317
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.083681593185886e-07,
+      "loss": 0.0424,
+      "step": 3318
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.056025117973784e-07,
+      "loss": 0.0409,
+      "step": 3319
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.028442125854471e-07,
+      "loss": 0.0483,
+      "step": 3320
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.000932638176125e-07,
+      "loss": 0.0499,
+      "step": 3321
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.973496676230094e-07,
+      "loss": 0.039,
+      "step": 3322
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.946134261250779e-07,
+      "loss": 0.0517,
+      "step": 3323
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.918845414415652e-07,
+      "loss": 0.0486,
+      "step": 3324
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.891630156845284e-07,
+      "loss": 0.0449,
+      "step": 3325
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.86448850960326e-07,
+      "loss": 0.0422,
+      "step": 3326
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.837420493696176e-07,
+      "loss": 0.0392,
+      "step": 3327
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.81042613007372e-07,
+      "loss": 0.0517,
+      "step": 3328
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.783505439628478e-07,
+      "loss": 0.0462,
+      "step": 3329
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7566584431960717e-07,
+      "loss": 0.0377,
+      "step": 3330
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.729885161555081e-07,
+      "loss": 0.0475,
+      "step": 3331
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.7031856154270085e-07,
+      "loss": 0.0513,
+      "step": 3332
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6765598254763346e-07,
+      "loss": 0.0553,
+      "step": 3333
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6500078123104197e-07,
+      "loss": 0.0444,
+      "step": 3334
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6235295964795145e-07,
+      "loss": 0.0431,
+      "step": 3335
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5971251984767707e-07,
+      "loss": 0.0468,
+      "step": 3336
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5707946387382187e-07,
+      "loss": 0.046,
+      "step": 3337
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5445379376427237e-07,
+      "loss": 0.0365,
+      "step": 3338
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.518355115511996e-07,
+      "loss": 0.042,
+      "step": 3339
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.492246192610561e-07,
+      "loss": 0.0458,
+      "step": 3340
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.466211189145742e-07,
+      "loss": 0.0441,
+      "step": 3341
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.4402501252676756e-07,
+      "loss": 0.0422,
+      "step": 3342
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.4143630210692345e-07,
+      "loss": 0.0419,
+      "step": 3343
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.388549896586103e-07,
+      "loss": 0.0453,
+      "step": 3344
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.362810771796644e-07,
+      "loss": 0.0446,
+      "step": 3345
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.337145666622011e-07,
+      "loss": 0.0412,
+      "step": 3346
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3115546009260045e-07,
+      "loss": 0.0526,
+      "step": 3347
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.28603759451518e-07,
+      "loss": 0.0635,
+      "step": 3348
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2605946671387176e-07,
+      "loss": 0.0451,
+      "step": 3349
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2352258384885213e-07,
+      "loss": 0.0445,
+      "step": 3350
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.209931128199118e-07,
+      "loss": 0.0415,
+      "step": 3351
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1847105558476595e-07,
+      "loss": 0.048,
+      "step": 3352
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.15956414095392e-07,
+      "loss": 0.0431,
+      "step": 3353
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1344919029802866e-07,
+      "loss": 0.0471,
+      "step": 3354
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.109493861331759e-07,
+      "loss": 0.0382,
+      "step": 3355
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0845700353558727e-07,
+      "loss": 0.0532,
+      "step": 3356
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.059720444342741e-07,
+      "loss": 0.0384,
+      "step": 3357
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0349451075250124e-07,
+      "loss": 0.043,
+      "step": 3358
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0102440440778825e-07,
+      "loss": 0.0511,
+      "step": 3359
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.985617273119036e-07,
+      "loss": 0.0443,
+      "step": 3360
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9610648137086926e-07,
+      "loss": 0.0467,
+      "step": 3361
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9365866848495413e-07,
+      "loss": 0.0468,
+      "step": 3362
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9121829054867276e-07,
+      "loss": 0.0529,
+      "step": 3363
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.887853494507876e-07,
+      "loss": 0.0454,
+      "step": 3364
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8635984707430353e-07,
+      "loss": 0.051,
+      "step": 3365
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.839417852964677e-07,
+      "loss": 0.0466,
+      "step": 3366
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8153116598877215e-07,
+      "loss": 0.0521,
+      "step": 3367
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.791279910169454e-07,
+      "loss": 0.0471,
+      "step": 3368
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.7673226224095526e-07,
+      "loss": 0.0369,
+      "step": 3369
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.743439815150052e-07,
+      "loss": 0.0491,
+      "step": 3370
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.719631506875354e-07,
+      "loss": 0.055,
+      "step": 3371
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.69589771601222e-07,
+      "loss": 0.0599,
+      "step": 3372
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6722384609297e-07,
+      "loss": 0.0457,
+      "step": 3373
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.64865375993918e-07,
+      "loss": 0.0354,
+      "step": 3374
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.625143631294359e-07,
+      "loss": 0.0522,
+      "step": 3375
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6017080931911585e-07,
+      "loss": 0.0404,
+      "step": 3376
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.578347163767837e-07,
+      "loss": 0.0448,
+      "step": 3377
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5550608611048753e-07,
+      "loss": 0.0481,
+      "step": 3378
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.5318492032250017e-07,
+      "loss": 0.0625,
+      "step": 3379
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.508712208093179e-07,
+      "loss": 0.0439,
+      "step": 3380
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4856498936165717e-07,
+      "loss": 0.0491,
+      "step": 3381
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.462662277644546e-07,
+      "loss": 0.047,
+      "step": 3382
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.439749377968671e-07,
+      "loss": 0.0444,
+      "step": 3383
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.416911212322671e-07,
+      "loss": 0.0423,
+      "step": 3384
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.39414779838243e-07,
+      "loss": 0.0437,
+      "step": 3385
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.371459153765988e-07,
+      "loss": 0.0373,
+      "step": 3386
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3488452960335095e-07,
+      "loss": 0.0519,
+      "step": 3387
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3263062426872493e-07,
+      "loss": 0.0433,
+      "step": 3388
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.303842011171632e-07,
+      "loss": 0.0473,
+      "step": 3389
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2814526188731153e-07,
+      "loss": 0.0437,
+      "step": 3390
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.259138083120239e-07,
+      "loss": 0.0486,
+      "step": 3391
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.236898421183665e-07,
+      "loss": 0.0419,
+      "step": 3392
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.214733650276014e-07,
+      "loss": 0.0437,
+      "step": 3393
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.192643787552019e-07,
+      "loss": 0.0561,
+      "step": 3394
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1706288501083926e-07,
+      "loss": 0.0399,
+      "step": 3395
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1486888549838947e-07,
+      "loss": 0.0518,
+      "step": 3396
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1268238191592523e-07,
+      "loss": 0.0534,
+      "step": 3397
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1050337595571743e-07,
+      "loss": 0.0382,
+      "step": 3398
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.08331869304237e-07,
+      "loss": 0.038,
+      "step": 3399
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0616786364214835e-07,
+      "loss": 0.0441,
+      "step": 3400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.040113606443118e-07,
+      "loss": 0.0449,
+      "step": 3401
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.018623619797789e-07,
+      "loss": 0.0441,
+      "step": 3402
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.99720869311797e-07,
+      "loss": 0.0416,
+      "step": 3403
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9758688429779913e-07,
+      "loss": 0.0449,
+      "step": 3404
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.954604085894086e-07,
+      "loss": 0.0409,
+      "step": 3405
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.933414438324422e-07,
+      "loss": 0.0376,
+      "step": 3406
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.91229991666897e-07,
+      "loss": 0.0461,
+      "step": 3407
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.891260537269591e-07,
+      "loss": 0.0477,
+      "step": 3408
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.870296316409971e-07,
+      "loss": 0.0435,
+      "step": 3409
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8494072703156405e-07,
+      "loss": 0.0419,
+      "step": 3410
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8285934151539353e-07,
+      "loss": 0.0476,
+      "step": 3411
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.8078547670340127e-07,
+      "loss": 0.0509,
+      "step": 3412
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.787191342006801e-07,
+      "loss": 0.0473,
+      "step": 3413
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.766603156065051e-07,
+      "loss": 0.0446,
+      "step": 3414
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.746090225143205e-07,
+      "loss": 0.0438,
+      "step": 3415
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.72565256511752e-07,
+      "loss": 0.0423,
+      "step": 3416
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7052901918059983e-07,
+      "loss": 0.0503,
+      "step": 3417
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6850031209683436e-07,
+      "loss": 0.0526,
+      "step": 3418
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.664791368305986e-07,
+      "loss": 0.0465,
+      "step": 3419
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6446549494620997e-07,
+      "loss": 0.0491,
+      "step": 3420
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6245938800214753e-07,
+      "loss": 0.0437,
+      "step": 3421
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.60460817551067e-07,
+      "loss": 0.0439,
+      "step": 3422
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.584697851397844e-07,
+      "loss": 0.0489,
+      "step": 3423
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.564862923092859e-07,
+      "loss": 0.0366,
+      "step": 3424
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5451034059472246e-07,
+      "loss": 0.0464,
+      "step": 3425
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.525419315254041e-07,
+      "loss": 0.0471,
+      "step": 3426
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.505810666248065e-07,
+      "loss": 0.0599,
+      "step": 3427
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4862774741056804e-07,
+      "loss": 0.0521,
+      "step": 3428
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4668197539448267e-07,
+      "loss": 0.0397,
+      "step": 3429
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.447437520825058e-07,
+      "loss": 0.057,
+      "step": 3430
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4281307897475184e-07,
+      "loss": 0.0424,
+      "step": 3431
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.408899575654866e-07,
+      "loss": 0.0402,
+      "step": 3432
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3897438934313624e-07,
+      "loss": 0.0387,
+      "step": 3433
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3706637579027914e-07,
+      "loss": 0.0411,
+      "step": 3434
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3516591838364523e-07,
+      "loss": 0.0429,
+      "step": 3435
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.332730185941179e-07,
+      "loss": 0.0488,
+      "step": 3436
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.313876778867341e-07,
+      "loss": 0.0562,
+      "step": 3437
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.295098977206711e-07,
+      "loss": 0.0417,
+      "step": 3438
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2763967954926637e-07,
+      "loss": 0.049,
+      "step": 3439
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2577702481999532e-07,
+      "loss": 0.0528,
+      "step": 3440
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.239219349744848e-07,
+      "loss": 0.0425,
+      "step": 3441
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2207441144850626e-07,
+      "loss": 0.0433,
+      "step": 3442
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2023445567197156e-07,
+      "loss": 0.0483,
+      "step": 3443
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1840206906893812e-07,
+      "loss": 0.0435,
+      "step": 3444
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.16577253057606e-07,
+      "loss": 0.0414,
+      "step": 3445
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1476000905031434e-07,
+      "loss": 0.0511,
+      "step": 3446
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.129503384535414e-07,
+      "loss": 0.0393,
+      "step": 3447
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1114824266790569e-07,
+      "loss": 0.0451,
+      "step": 3448
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0935372308816148e-07,
+      "loss": 0.0468,
+      "step": 3449
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0756678110320005e-07,
+      "loss": 0.0415,
+      "step": 3450
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.057874180960484e-07,
+      "loss": 0.0489,
+      "step": 3451
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0401563544386492e-07,
+      "loss": 0.0448,
+      "step": 3452
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0225143451794605e-07,
+      "loss": 0.0498,
+      "step": 3453
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0049481668371728e-07,
+      "loss": 0.0557,
+      "step": 3454
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9874578330073225e-07,
+      "loss": 0.0483,
+      "step": 3455
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9700433572268028e-07,
+      "loss": 0.039,
+      "step": 3456
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9527047529737554e-07,
+      "loss": 0.0404,
+      "step": 3457
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9354420336676227e-07,
+      "loss": 0.0429,
+      "step": 3458
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9182552126691067e-07,
+      "loss": 0.0361,
+      "step": 3459
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9011443032801446e-07,
+      "loss": 0.0407,
+      "step": 3460
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8841093187439653e-07,
+      "loss": 0.0435,
+      "step": 3461
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8671502722449997e-07,
+      "loss": 0.0441,
+      "step": 3462
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8502671769089043e-07,
+      "loss": 0.0439,
+      "step": 3463
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.833460045802593e-07,
+      "loss": 0.0486,
+      "step": 3464
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8167288919341498e-07,
+      "loss": 0.0404,
+      "step": 3465
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.800073728252838e-07,
+      "loss": 0.0434,
+      "step": 3466
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7834945676491688e-07,
+      "loss": 0.038,
+      "step": 3467
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.766991422954767e-07,
+      "loss": 0.0419,
+      "step": 3468
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7505643069424594e-07,
+      "loss": 0.0396,
+      "step": 3469
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7342132323262318e-07,
+      "loss": 0.0498,
+      "step": 3470
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7179382117611943e-07,
+      "loss": 0.0576,
+      "step": 3471
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7017392578435933e-07,
+      "loss": 0.0536,
+      "step": 3472
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.685616383110844e-07,
+      "loss": 0.0532,
+      "step": 3473
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.669569600041421e-07,
+      "loss": 0.0508,
+      "step": 3474
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6535989210549552e-07,
+      "loss": 0.0579,
+      "step": 3475
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.637704358512149e-07,
+      "loss": 0.0489,
+      "step": 3476
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6218859247147945e-07,
+      "loss": 0.0438,
+      "step": 3477
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6061436319057765e-07,
+      "loss": 0.0526,
+      "step": 3478
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.590477492269038e-07,
+      "loss": 0.0479,
+      "step": 3479
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.574887517929591e-07,
+      "loss": 0.0461,
+      "step": 3480
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5593737209534942e-07,
+      "loss": 0.0441,
+      "step": 3481
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5439361133478436e-07,
+      "loss": 0.0418,
+      "step": 3482
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5285747070607704e-07,
+      "loss": 0.0426,
+      "step": 3483
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5132895139814307e-07,
+      "loss": 0.0474,
+      "step": 3484
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.498080545939995e-07,
+      "loss": 0.0594,
+      "step": 3485
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.482947814707636e-07,
+      "loss": 0.0466,
+      "step": 3486
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4678913319965293e-07,
+      "loss": 0.0438,
+      "step": 3487
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4529111094598202e-07,
+      "loss": 0.0436,
+      "step": 3488
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4380071586916455e-07,
+      "loss": 0.0446,
+      "step": 3489
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4231794912271002e-07,
+      "loss": 0.0415,
+      "step": 3490
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.40842811854226e-07,
+      "loss": 0.0543,
+      "step": 3491
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3937530520541254e-07,
+      "loss": 0.0407,
+      "step": 3492
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3791543031206666e-07,
+      "loss": 0.0432,
+      "step": 3493
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3646318830407346e-07,
+      "loss": 0.053,
+      "step": 3494
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3501858030541825e-07,
+      "loss": 0.0402,
+      "step": 3495
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3358160743417116e-07,
+      "loss": 0.0445,
+      "step": 3496
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3215227080249583e-07,
+      "loss": 0.0456,
+      "step": 3497
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3073057151664625e-07,
+      "loss": 0.0516,
+      "step": 3498
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.293165106769645e-07,
+      "loss": 0.051,
+      "step": 3499
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2791008937787952e-07,
+      "loss": 0.0399,
+      "step": 3500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2651130870791172e-07,
+      "loss": 0.0329,
+      "step": 3501
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.251201697496618e-07,
+      "loss": 0.0468,
+      "step": 3502
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2373667357982068e-07,
+      "loss": 0.045,
+      "step": 3503
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2236082126916405e-07,
+      "loss": 0.0479,
+      "step": 3504
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2099261388254901e-07,
+      "loss": 0.0492,
+      "step": 3505
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.196320524789174e-07,
+      "loss": 0.0387,
+      "step": 3506
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1827913811129355e-07,
+      "loss": 0.0478,
+      "step": 3507
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1693387182677985e-07,
+      "loss": 0.0492,
+      "step": 3508
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1559625466656565e-07,
+      "loss": 0.0415,
+      "step": 3509
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1426628766591619e-07,
+      "loss": 0.0511,
+      "step": 3510
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.129439718541736e-07,
+      "loss": 0.041,
+      "step": 3511
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.116293082547637e-07,
+      "loss": 0.0411,
+      "step": 3512
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.103222978851859e-07,
+      "loss": 0.0471,
+      "step": 3513
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0902294175701766e-07,
+      "loss": 0.0403,
+      "step": 3514
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0773124087591236e-07,
+      "loss": 0.0509,
+      "step": 3515
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0644719624159805e-07,
+      "loss": 0.0419,
+      "step": 3516
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0517080884787756e-07,
+      "loss": 0.0509,
+      "step": 3517
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0390207968262622e-07,
+      "loss": 0.0446,
+      "step": 3518
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0264100972779411e-07,
+      "loss": 0.0447,
+      "step": 3519
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0138759995940162e-07,
+      "loss": 0.0437,
+      "step": 3520
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0014185134754273e-07,
+      "loss": 0.0413,
+      "step": 3521
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.890376485637731e-08,
+      "loss": 0.0497,
+      "step": 3522
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.767334144413998e-08,
+      "loss": 0.0596,
+      "step": 3523
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.64505820631334e-08,
+      "loss": 0.0448,
+      "step": 3524
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.523548765972502e-08,
+      "loss": 0.0415,
+      "step": 3525
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.402805917435587e-08,
+      "loss": 0.0366,
+      "step": 3526
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.282829754152845e-08,
+      "loss": 0.0457,
+      "step": 3527
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.16362036898144e-08,
+      "loss": 0.0454,
+      "step": 3528
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.045177854184794e-08,
+      "loss": 0.0368,
+      "step": 3529
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.927502301433243e-08,
+      "loss": 0.0485,
+      "step": 3530
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.810593801803158e-08,
+      "loss": 0.0461,
+      "step": 3531
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.694452445777268e-08,
+      "loss": 0.0307,
+      "step": 3532
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.579078323244893e-08,
+      "loss": 0.0381,
+      "step": 3533
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.46447152350105e-08,
+      "loss": 0.0459,
+      "step": 3534
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.350632135247228e-08,
+      "loss": 0.0431,
+      "step": 3535
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.237560246590836e-08,
+      "loss": 0.0381,
+      "step": 3536
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.12525594504543e-08,
+      "loss": 0.0427,
+      "step": 3537
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.013719317530145e-08,
+      "loss": 0.0557,
+      "step": 3538
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.902950450370261e-08,
+      "loss": 0.0584,
+      "step": 3539
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.792949429296759e-08,
+      "loss": 0.0474,
+      "step": 3540
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.683716339446312e-08,
+      "loss": 0.0485,
+      "step": 3541
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.575251265361183e-08,
+      "loss": 0.0414,
+      "step": 3542
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.467554290989443e-08,
+      "loss": 0.046,
+      "step": 3543
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.360625499684416e-08,
+      "loss": 0.0442,
+      "step": 3544
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.254464974205011e-08,
+      "loss": 0.0498,
+      "step": 3545
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.149072796715506e-08,
+      "loss": 0.0432,
+      "step": 3546
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.04444904878554e-08,
+      "loss": 0.0555,
+      "step": 3547
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.940593811390006e-08,
+      "loss": 0.0573,
+      "step": 3548
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.83750716490894e-08,
+      "loss": 0.0488,
+      "step": 3549
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.735189189127522e-08,
+      "loss": 0.0526,
+      "step": 3550
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.633639963236183e-08,
+      "loss": 0.0497,
+      "step": 3551
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.532859565830163e-08,
+      "loss": 0.0514,
+      "step": 3552
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.432848074909626e-08,
+      "loss": 0.0394,
+      "step": 3553
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.333605567879875e-08,
+      "loss": 0.0505,
+      "step": 3554
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.235132121550803e-08,
+      "loss": 0.0469,
+      "step": 3555
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.13742781213733e-08,
+      "loss": 0.0402,
+      "step": 3556
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.04049271525875e-08,
+      "loss": 0.046,
+      "step": 3557
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.944326905939268e-08,
+      "loss": 0.0442,
+      "step": 3558
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.848930458607793e-08,
+      "loss": 0.0428,
+      "step": 3559
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.7543034470973756e-08,
+      "loss": 0.0502,
+      "step": 3560
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.660445944645987e-08,
+      "loss": 0.0487,
+      "step": 3561
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.5673580238957416e-08,
+      "loss": 0.0493,
+      "step": 3562
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.47503975689323e-08,
+      "loss": 0.0403,
+      "step": 3563
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.383491215089298e-08,
+      "loss": 0.0436,
+      "step": 3564
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.292712469339156e-08,
+      "loss": 0.05,
+      "step": 3565
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.202703589902158e-08,
+      "loss": 0.0432,
+      "step": 3566
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.1134646464418014e-08,
+      "loss": 0.0544,
+      "step": 3567
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.024995708025726e-08,
+      "loss": 0.0543,
+      "step": 3568
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.937296843125605e-08,
+      "loss": 0.0447,
+      "step": 3569
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.8503681196169214e-08,
+      "loss": 0.0445,
+      "step": 3570
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.764209604779524e-08,
+      "loss": 0.0561,
+      "step": 3571
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6788213652966264e-08,
+      "loss": 0.0501,
+      "step": 3572
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.594203467255809e-08,
+      "loss": 0.0469,
+      "step": 3573
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.5103559761479066e-08,
+      "loss": 0.0505,
+      "step": 3574
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.427278956867897e-08,
+      "loss": 0.0405,
+      "step": 3575
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3449724737142373e-08,
+      "loss": 0.044,
+      "step": 3576
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.2634365903890805e-08,
+      "loss": 0.0414,
+      "step": 3577
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.182671369998281e-08,
+      "loss": 0.0546,
+      "step": 3578
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.102676875050948e-08,
+      "loss": 0.0427,
+      "step": 3579
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.02345316745989e-08,
+      "loss": 0.0388,
+      "step": 3580
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.945000308541169e-08,
+      "loss": 0.0419,
+      "step": 3581
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.867318359014549e-08,
+      "loss": 0.0473,
+      "step": 3582
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.7904073790029364e-08,
+      "loss": 0.0471,
+      "step": 3583
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.714267428032603e-08,
+      "loss": 0.0379,
+      "step": 3584
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.638898565032856e-08,
+      "loss": 0.0459,
+      "step": 3585
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.564300848336477e-08,
+      "loss": 0.0537,
+      "step": 3586
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.490474335679284e-08,
+      "loss": 0.053,
+      "step": 3587
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4174190842003464e-08,
+      "loss": 0.0508,
+      "step": 3588
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3451351504416586e-08,
+      "loss": 0.0476,
+      "step": 3589
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2736225903482464e-08,
+      "loss": 0.0394,
+      "step": 3590
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.202881459268059e-08,
+      "loss": 0.0509,
+      "step": 3591
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.132911811952299e-08,
+      "loss": 0.0419,
+      "step": 3592
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0637137025547605e-08,
+      "loss": 0.0421,
+      "step": 3593
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.9952871846321564e-08,
+      "loss": 0.0432,
+      "step": 3594
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.927632311144124e-08,
+      "loss": 0.0449,
+      "step": 3595
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8607491344530003e-08,
+      "loss": 0.0453,
+      "step": 3596
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.7946377063239326e-08,
+      "loss": 0.0458,
+      "step": 3597
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.729298077924547e-08,
+      "loss": 0.0427,
+      "step": 3598
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.664730299825391e-08,
+      "loss": 0.048,
+      "step": 3599
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.6009344219993795e-08,
+      "loss": 0.044,
+      "step": 3600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5379104938223487e-08,
+      "loss": 0.0424,
+      "step": 3601
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.475658564072392e-08,
+      "loss": 0.0449,
+      "step": 3602
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4141786809300793e-08,
+      "loss": 0.0437,
+      "step": 3603
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.353470891978571e-08,
+      "loss": 0.0513,
+      "step": 3604
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2935352442035042e-08,
+      "loss": 0.0444,
+      "step": 3605
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2343717839928836e-08,
+      "loss": 0.0496,
+      "step": 3606
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1759805571368588e-08,
+      "loss": 0.0467,
+      "step": 3607
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1183616088281678e-08,
+      "loss": 0.0422,
+      "step": 3608
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0615149836616944e-08,
+      "loss": 0.0472,
+      "step": 3609
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.0054407256345776e-08,
+      "loss": 0.0483,
+      "step": 3610
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9501388781462127e-08,
+      "loss": 0.0461,
+      "step": 3611
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8956094839982508e-08,
+      "loss": 0.0533,
+      "step": 3612
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.841852585394266e-08,
+      "loss": 0.0504,
+      "step": 3613
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7888682239401988e-08,
+      "loss": 0.032,
+      "step": 3614
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7366564406439134e-08,
+      "loss": 0.0398,
+      "step": 3615
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.6852172759154185e-08,
+      "loss": 0.0385,
+      "step": 3616
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.634550769566645e-08,
+      "loss": 0.0418,
+      "step": 3617
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.58465696081167e-08,
+      "loss": 0.0472,
+      "step": 3618
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.5355358882663817e-08,
+      "loss": 0.0384,
+      "step": 3619
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.487187589948702e-08,
+      "loss": 0.0424,
+      "step": 3620
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.439612103278365e-08,
+      "loss": 0.0421,
+      "step": 3621
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3928094650769164e-08,
+      "loss": 0.0458,
+      "step": 3622
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.346779711568047e-08,
+      "loss": 0.0396,
+      "step": 3623
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3015228783770373e-08,
+      "loss": 0.0576,
+      "step": 3624
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2570390005308686e-08,
+      "loss": 0.0367,
+      "step": 3625
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.213328112458445e-08,
+      "loss": 0.0373,
+      "step": 3626
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1703902479904827e-08,
+      "loss": 0.0384,
+      "step": 3627
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1282254403590653e-08,
+      "loss": 0.0456,
+      "step": 3628
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0868337221984216e-08,
+      "loss": 0.0412,
+      "step": 3629
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0462151255440368e-08,
+      "loss": 0.0441,
+      "step": 3630
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.006369681833208e-08,
+      "loss": 0.045,
+      "step": 3631
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.672974219048226e-09,
+      "loss": 0.0357,
+      "step": 3632
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.28998375999468e-09,
+      "loss": 0.0439,
+      "step": 3633
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.91472573758989e-09,
+      "loss": 0.0485,
+      "step": 3634
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.54720044227153e-09,
+      "loss": 0.0446,
+      "step": 3635
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.18740815848762e-09,
+      "loss": 0.0473,
+      "step": 3636
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.835349164707629e-09,
+      "loss": 0.0461,
+      "step": 3637
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.491023733409153e-09,
+      "loss": 0.0418,
+      "step": 3638
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.154432131087907e-09,
+      "loss": 0.0423,
+      "step": 3639
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.825574618254393e-09,
+      "loss": 0.0434,
+      "step": 3640
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.504451449431681e-09,
+      "loss": 0.0455,
+      "step": 3641
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.191062873155407e-09,
+      "loss": 0.0589,
+      "step": 3642
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.885409131979325e-09,
+      "loss": 0.0434,
+      "step": 3643
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.587490462465317e-09,
+      "loss": 0.0386,
+      "step": 3644
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.297307095193382e-09,
+      "loss": 0.051,
+      "step": 3645
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.014859254753868e-09,
+      "loss": 0.0365,
+      "step": 3646
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.740147159750796e-09,
+      "loss": 0.041,
+      "step": 3647
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.473171022800759e-09,
+      "loss": 0.0435,
+      "step": 3648
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.213931050534026e-09,
+      "loss": 0.0499,
+      "step": 3649
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.9624274435912105e-09,
+      "loss": 0.041,
+      "step": 3650
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.718660396628826e-09,
+      "loss": 0.0483,
+      "step": 3651
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4826300983126227e-09,
+      "loss": 0.0359,
+      "step": 3652
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.254336731320917e-09,
+      "loss": 0.0412,
+      "step": 3653
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.033780472343484e-09,
+      "loss": 0.0387,
+      "step": 3654
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.820961492084884e-09,
+      "loss": 0.0482,
+      "step": 3655
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.615879955256695e-09,
+      "loss": 0.0361,
+      "step": 3656
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4185360205863927e-09,
+      "loss": 0.0482,
+      "step": 3657
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2289298408095792e-09,
+      "loss": 0.0507,
+      "step": 3658
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.047061562674424e-09,
+      "loss": 0.0503,
+      "step": 3659
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8729313269405526e-09,
+      "loss": 0.039,
+      "step": 3660
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.706539268377938e-09,
+      "loss": 0.0571,
+      "step": 3661
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5478855157669003e-09,
+      "loss": 0.0528,
+      "step": 3662
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3969701919025468e-09,
+      "loss": 0.0454,
+      "step": 3663
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2537934135847807e-09,
+      "loss": 0.0631,
+      "step": 3664
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1183552916282924e-09,
+      "loss": 0.0432,
+      "step": 3665
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.906559308570096e-10,
+      "loss": 0.043,
+      "step": 3666
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.706954301052062e-10,
+      "loss": 0.0447,
+      "step": 3667
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.584738822175031e-10,
+      "loss": 0.0526,
+      "step": 3668
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.539913740499782e-10,
+      "loss": 0.05,
+      "step": 3669
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.572479864679459e-10,
+      "loss": 0.0499,
+      "step": 3670
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.682437943470674e-10,
+      "loss": 0.0417,
+      "step": 3671
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.8697886657446116e-10,
+      "loss": 0.0503,
+      "step": 3672
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.1345326604315106e-10,
+      "loss": 0.0537,
+      "step": 3673
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.476670496620592e-10,
+      "loss": 0.0513,
+      "step": 3674
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8962026834712377e-10,
+      "loss": 0.0465,
+      "step": 3675
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3931296702240915e-10,
+      "loss": 0.0506,
+      "step": 3676
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.674518462454707e-11,
+      "loss": 0.0417,
+      "step": 3677
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.191695410162624e-11,
+      "loss": 0.0472,
+      "step": 3678
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.482830240653101e-11,
+      "loss": 0.0449,
+      "step": 3679
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5479250505823217e-11,
+      "loss": 0.0359,
+      "step": 3680
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.869813375301234e-12,
+      "loss": 0.0466,
+      "step": 3681
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.0503,
+      "step": 3682
+    },
+    {
+      "epoch": 2.0,
+      "step": 3682,
+      "total_flos": 228160431292416.0,
+      "train_loss": 0.06622124443489336,
+      "train_runtime": 9277.2004,
+      "train_samples_per_second": 50.791,
+      "train_steps_per_second": 0.397
+    }
+  ],
+  "max_steps": 3682,
+  "num_train_epochs": 2,
+  "total_flos": 228160431292416.0,
+  "trial_name": null,
+  "trial_params": null
+}