diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14707 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 2447,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-07,
+      "loss": 0.7816,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-07,
+      "loss": 0.7654,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.108108108108109e-07,
+      "loss": 0.7799,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 0.7273,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3513513513513515e-06,
+      "loss": 0.7163,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6216216216216219e-06,
+      "loss": 0.7276,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8918918918918922e-06,
+      "loss": 0.7067,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 0.6702,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.432432432432433e-06,
+      "loss": 0.6692,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 0.6684,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9729729729729736e-06,
+      "loss": 0.6678,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 0.6631,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.513513513513514e-06,
+      "loss": 0.6289,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 0.6567,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0540540540540545e-06,
+      "loss": 0.6347,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 0.6207,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.594594594594596e-06,
+      "loss": 0.6058,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.864864864864866e-06,
+      "loss": 0.6407,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.135135135135135e-06,
+      "loss": 0.6021,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 0.5953,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.675675675675676e-06,
+      "loss": 0.5985,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.945945945945947e-06,
+      "loss": 0.6051,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2162162162162164e-06,
+      "loss": 0.5991,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 0.5756,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.7567567567567575e-06,
+      "loss": 0.6037,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.027027027027028e-06,
+      "loss": 0.5853,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.297297297297298e-06,
+      "loss": 0.578,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 0.5873,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.837837837837838e-06,
+      "loss": 0.5981,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 0.5729,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.378378378378378e-06,
+      "loss": 0.6011,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 0.6065,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.91891891891892e-06,
+      "loss": 0.6008,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 0.5759,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.45945945945946e-06,
+      "loss": 0.5737,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 0.5966,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.5795,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.027027027027027e-05,
+      "loss": 0.5761,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0540540540540541e-05,
+      "loss": 0.6732,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 0.5787,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1081081081081081e-05,
+      "loss": 0.5901,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1351351351351352e-05,
+      "loss": 0.5744,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1621621621621622e-05,
+      "loss": 0.5573,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 0.5831,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2162162162162164e-05,
+      "loss": 0.5977,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2432432432432433e-05,
+      "loss": 0.5716,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2702702702702702e-05,
+      "loss": 0.5718,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 0.5873,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3243243243243244e-05,
+      "loss": 0.5908,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 0.5786,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3783783783783784e-05,
+      "loss": 0.5724,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 0.6004,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4324324324324326e-05,
+      "loss": 0.5834,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 0.5703,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4864864864864865e-05,
+      "loss": 0.565,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 0.5846,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.540540540540541e-05,
+      "loss": 0.5681,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5675675675675676e-05,
+      "loss": 0.5643,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5945945945945947e-05,
+      "loss": 0.5611,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 0.5715,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.648648648648649e-05,
+      "loss": 0.5786,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6756756756756757e-05,
+      "loss": 0.5664,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7027027027027028e-05,
+      "loss": 0.5659,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 0.569,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.756756756756757e-05,
+      "loss": 0.5768,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.783783783783784e-05,
+      "loss": 0.563,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8108108108108108e-05,
+      "loss": 0.5677,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 0.5486,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.864864864864865e-05,
+      "loss": 0.5492,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 0.5634,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.918918918918919e-05,
+      "loss": 0.5609,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 0.589,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.972972972972973e-05,
+      "loss": 0.557,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.5767,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999123656823e-05,
+      "loss": 0.552,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999964946288268e-05,
+      "loss": 0.5796,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999921129206198e-05,
+      "loss": 0.5736,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999985978539882e-05,
+      "loss": 0.5504,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999780914973646e-05,
+      "loss": 0.5629,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999684518068916e-05,
+      "loss": 0.5666,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999570594853575e-05,
+      "loss": 0.5689,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999439145527305e-05,
+      "loss": 0.5692,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999290170320487e-05,
+      "loss": 0.5434,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999912366949423e-05,
+      "loss": 0.5358,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998939643340364e-05,
+      "loss": 0.539,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998738092181423e-05,
+      "loss": 0.5488,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999851901637066e-05,
+      "loss": 0.5828,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998282416292057e-05,
+      "loss": 0.5657,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998028292360287e-05,
+      "loss": 0.5596,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999775664502076e-05,
+      "loss": 0.548,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997467474749583e-05,
+      "loss": 0.5552,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999716078205358e-05,
+      "loss": 0.5485,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999683656747029e-05,
+      "loss": 0.5509,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999649483156796e-05,
+      "loss": 0.5309,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996135574945543e-05,
+      "loss": 0.5628,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995758798232704e-05,
+      "loss": 0.5402,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995364502089815e-05,
+      "loss": 0.5427,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994952687207957e-05,
+      "loss": 0.545,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994523354308904e-05,
+      "loss": 0.5559,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994076504145147e-05,
+      "loss": 0.5336,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993612137499875e-05,
+      "loss": 0.5656,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999313025518698e-05,
+      "loss": 0.5477,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999263085805104e-05,
+      "loss": 0.578,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999211394696735e-05,
+      "loss": 0.5381,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991579522841894e-05,
+      "loss": 0.536,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991027586611346e-05,
+      "loss": 0.5329,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990458139243075e-05,
+      "loss": 0.5403,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998987118173515e-05,
+      "loss": 0.5514,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989266715116316e-05,
+      "loss": 0.5509,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988644740446022e-05,
+      "loss": 0.5536,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998800525881439e-05,
+      "loss": 0.5464,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998734827134223e-05,
+      "loss": 0.5779,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986673779181033e-05,
+      "loss": 0.5465,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985981783512977e-05,
+      "loss": 0.5351,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998527228555091e-05,
+      "loss": 0.5478,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984545286538362e-05,
+      "loss": 0.5561,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998380078774953e-05,
+      "loss": 0.5528,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983038790489296e-05,
+      "loss": 0.5524,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982259296093192e-05,
+      "loss": 0.5505,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998146230592743e-05,
+      "loss": 0.5812,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980647821388888e-05,
+      "loss": 0.5507,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979815843905098e-05,
+      "loss": 0.5406,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978966374934255e-05,
+      "loss": 0.5517,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978099415965214e-05,
+      "loss": 0.5582,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997721496851748e-05,
+      "loss": 0.5287,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976313034141214e-05,
+      "loss": 0.5221,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975393614417225e-05,
+      "loss": 0.5449,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974456710956965e-05,
+      "loss": 0.5422,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973502325402533e-05,
+      "loss": 0.5302,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972530459426663e-05,
+      "loss": 0.5219,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971541114732743e-05,
+      "loss": 0.5281,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997053429305477e-05,
+      "loss": 0.5408,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969509996157398e-05,
+      "loss": 0.5406,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996846822583589e-05,
+      "loss": 0.534,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967408983916145e-05,
+      "loss": 0.5223,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966332272254685e-05,
+      "loss": 0.5291,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965238092738643e-05,
+      "loss": 0.536,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996412644728578e-05,
+      "loss": 0.5304,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996299733784445e-05,
+      "loss": 0.5467,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996185076639364e-05,
+      "loss": 0.5206,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996068673494292e-05,
+      "loss": 0.5217,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995950524553248e-05,
+      "loss": 0.5163,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99583063002331e-05,
+      "loss": 0.5261,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995708990114615e-05,
+      "loss": 0.5416,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955856050403596e-05,
+      "loss": 0.526,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954604750167995e-05,
+      "loss": 0.5409,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953336002632483e-05,
+      "loss": 0.5202,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952049810020774e-05,
+      "loss": 0.5397,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950746174587163e-05,
+      "loss": 0.5481,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949425098616513e-05,
+      "loss": 0.5137,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948086584424258e-05,
+      "loss": 0.5415,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994673063435639e-05,
+      "loss": 0.5408,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945357250789467e-05,
+      "loss": 0.5133,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943966436130598e-05,
+      "loss": 0.5346,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994255819281744e-05,
+      "loss": 0.5441,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941132523318217e-05,
+      "loss": 0.5348,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939689430131667e-05,
+      "loss": 0.5315,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993822891578708e-05,
+      "loss": 0.5376,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936750982844284e-05,
+      "loss": 0.5355,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935255633893632e-05,
+      "loss": 0.5385,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933742871556e-05,
+      "loss": 0.5275,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932212698482787e-05,
+      "loss": 0.5372,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930665117355906e-05,
+      "loss": 0.538,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929100130887782e-05,
+      "loss": 0.5182,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927517741821343e-05,
+      "loss": 0.5317,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925917952930024e-05,
+      "loss": 0.5376,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992430076701775e-05,
+      "loss": 0.5241,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922666186918943e-05,
+      "loss": 0.5413,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921014215498507e-05,
+      "loss": 0.5316,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919344855651834e-05,
+      "loss": 0.5149,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917658110304783e-05,
+      "loss": 0.5261,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915953982413693e-05,
+      "loss": 0.5354,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914232474965364e-05,
+      "loss": 0.5228,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991249359097706e-05,
+      "loss": 0.5273,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99107373334965e-05,
+      "loss": 0.5279,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908963705601848e-05,
+      "loss": 0.5253,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907172710401723e-05,
+      "loss": 0.5237,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905364351035176e-05,
+      "loss": 0.5357,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903538630671687e-05,
+      "loss": 0.5316,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901695552511185e-05,
+      "loss": 0.515,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899835119783996e-05,
+      "loss": 0.5165,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989795733575088e-05,
+      "loss": 0.5251,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896062203703004e-05,
+      "loss": 0.5087,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894149726961938e-05,
+      "loss": 0.5145,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892219908879652e-05,
+      "loss": 0.5161,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989027275283852e-05,
+      "loss": 0.5201,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888308262251286e-05,
+      "loss": 0.5141,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988632644056109e-05,
+      "loss": 0.5217,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884327291241446e-05,
+      "loss": 0.5156,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882310817796235e-05,
+      "loss": 0.5095,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98802770237597e-05,
+      "loss": 0.5114,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878225912696446e-05,
+      "loss": 0.5185,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876157488201426e-05,
+      "loss": 0.5199,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987407175389994e-05,
+      "loss": 0.4925,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871968713447626e-05,
+      "loss": 0.5128,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869848370530455e-05,
+      "loss": 0.5016,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867710728864723e-05,
+      "loss": 0.5162,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865555792197042e-05,
+      "loss": 0.5263,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863383564304344e-05,
+      "loss": 0.5146,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861194048993865e-05,
+      "loss": 0.5164,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858987250103132e-05,
+      "loss": 0.4977,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985676317149998e-05,
+      "loss": 0.5154,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985452181708251e-05,
+      "loss": 0.5176,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852263190779123e-05,
+      "loss": 0.5014,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849987296548477e-05,
+      "loss": 0.5209,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847694138379508e-05,
+      "loss": 0.5131,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845383720291392e-05,
+      "loss": 0.5046,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984305604633358e-05,
+      "loss": 0.5165,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984071112058574e-05,
+      "loss": 0.51,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838348947157802e-05,
+      "loss": 0.5115,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835969530189912e-05,
+      "loss": 0.5124,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833572873852446e-05,
+      "loss": 0.5018,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831158982345982e-05,
+      "loss": 0.5237,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982872785990132e-05,
+      "loss": 0.5036,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826279510779454e-05,
+      "loss": 0.5101,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823813939271573e-05,
+      "loss": 0.5103,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982133114969905e-05,
+      "loss": 0.4904,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818831146413435e-05,
+      "loss": 0.499,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981631393379645e-05,
+      "loss": 0.5145,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813779516259987e-05,
+      "loss": 0.5294,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811227898246072e-05,
+      "loss": 0.5209,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98086590842269e-05,
+      "loss": 0.4896,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806073078704793e-05,
+      "loss": 0.5183,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980346988621221e-05,
+      "loss": 0.5248,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800849511311726e-05,
+      "loss": 0.5069,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979821195859604e-05,
+      "loss": 0.5,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795557232687956e-05,
+      "loss": 0.5094,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792885338240375e-05,
+      "loss": 0.5116,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790196279936286e-05,
+      "loss": 0.5134,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978749006248877e-05,
+      "loss": 0.5182,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784766690640976e-05,
+      "loss": 0.5199,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782026169166116e-05,
+      "loss": 0.5007,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779268502867474e-05,
+      "loss": 0.4903,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776493696578365e-05,
+      "loss": 0.5153,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773701755162157e-05,
+      "loss": 0.5021,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977089268351225e-05,
+      "loss": 0.5059,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768066486552064e-05,
+      "loss": 0.5039,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765223169235036e-05,
+      "loss": 0.5058,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762362736544606e-05,
+      "loss": 0.5217,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975948519349422e-05,
+      "loss": 0.5058,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975659054512731e-05,
+      "loss": 0.4991,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975367879651728e-05,
+      "loss": 0.5012,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975074995276752e-05,
+      "loss": 0.4968,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974780401901137e-05,
+      "loss": 0.5016,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744841000412122e-05,
+      "loss": 0.5085,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741860902163027e-05,
+      "loss": 0.5248,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973886372948726e-05,
+      "loss": 0.4966,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735849487637927e-05,
+      "loss": 0.4934,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732818181898046e-05,
+      "loss": 0.5097,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972976981758054e-05,
+      "loss": 0.4918,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972670440002825e-05,
+      "loss": 0.5091,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723621934613873e-05,
+      "loss": 0.506,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720522426740022e-05,
+      "loss": 0.4891,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717405881839146e-05,
+      "loss": 0.4989,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714272305373584e-05,
+      "loss": 0.5003,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711121702835504e-05,
+      "loss": 0.5051,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9707954079746928e-05,
+      "loss": 0.5057,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704769441659703e-05,
+      "loss": 0.4968,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701567794155504e-05,
+      "loss": 0.5111,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698349142845813e-05,
+      "loss": 0.4962,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695113493371917e-05,
+      "loss": 0.5023,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691860851404895e-05,
+      "loss": 0.516,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688591222645607e-05,
+      "loss": 0.5027,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968530461282469e-05,
+      "loss": 0.4938,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682001027702533e-05,
+      "loss": 0.5186,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678680473069294e-05,
+      "loss": 0.4822,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675342954744855e-05,
+      "loss": 0.5038,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671988478578844e-05,
+      "loss": 0.4968,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668617050450604e-05,
+      "loss": 0.4982,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966522867626919e-05,
+      "loss": 0.4979,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661823361973362e-05,
+      "loss": 0.5255,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658401113531567e-05,
+      "loss": 0.4923,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965496193694193e-05,
+      "loss": 0.5155,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651505838232255e-05,
+      "loss": 0.5131,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648032823459996e-05,
+      "loss": 0.4863,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644542898712253e-05,
+      "loss": 0.4944,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964103607010578e-05,
+      "loss": 0.4933,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963751234378694e-05,
+      "loss": 0.5005,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963397172593172e-05,
+      "loss": 0.5091,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963041422274572e-05,
+      "loss": 0.4884,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962683984046412e-05,
+      "loss": 0.4938,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962324858535169e-05,
+      "loss": 0.5114,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961964046370278e-05,
+      "loss": 0.4864,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616015481841293e-05,
+      "loss": 0.5102,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612373646120684e-05,
+      "loss": 0.497,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960871496292395e-05,
+      "loss": 0.4896,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605039438663614e-05,
+      "loss": 0.5045,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601347079781718e-05,
+      "loss": 0.4992,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959763789274981e-05,
+      "loss": 0.4896,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959391188406893e-05,
+      "loss": 0.5041,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590169060269602e-05,
+      "loss": 0.4935,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586409427911827e-05,
+      "loss": 0.4872,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582632993585055e-05,
+      "loss": 0.4945,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578839763908193e-05,
+      "loss": 0.5141,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575029745529585e-05,
+      "loss": 0.5178,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571202945126995e-05,
+      "loss": 0.5078,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567359369407605e-05,
+      "loss": 0.4977,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563499025108e-05,
+      "loss": 0.4713,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955962191899415e-05,
+      "loss": 0.4674,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955572805786141e-05,
+      "loss": 0.5106,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955181744853449e-05,
+      "loss": 0.4968,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547890097867466e-05,
+      "loss": 0.5002,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543946012743757e-05,
+      "loss": 0.4868,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539985200076096e-05,
+      "loss": 0.5002,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536007666806555e-05,
+      "loss": 0.4921,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9532013419906498e-05,
+      "loss": 0.4835,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528002466376587e-05,
+      "loss": 0.5058,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523974813246765e-05,
+      "loss": 0.5007,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519930467576246e-05,
+      "loss": 0.4819,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515869436453502e-05,
+      "loss": 0.4828,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511791726996243e-05,
+      "loss": 0.4928,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507697346351414e-05,
+      "loss": 0.4933,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503586301695185e-05,
+      "loss": 0.5051,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949945860023292e-05,
+      "loss": 0.4858,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949531424919919e-05,
+      "loss": 0.4873,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949115325585775e-05,
+      "loss": 0.4832,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486975627501503e-05,
+      "loss": 0.5024,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482781371452527e-05,
+      "loss": 0.4912,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478570495062038e-05,
+      "loss": 0.4869,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947434300571038e-05,
+      "loss": 0.4962,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9470098910807016e-05,
+      "loss": 0.4941,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465838217790517e-05,
+      "loss": 0.4942,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461560934128535e-05,
+      "loss": 0.4933,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945726706731781e-05,
+      "loss": 0.505,
+      "step": 324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452956624884145e-05,
+      "loss": 0.4913,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448629614382394e-05,
+      "loss": 0.4784,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444286043396447e-05,
+      "loss": 0.4833,
+      "step": 327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943992591953922e-05,
+      "loss": 0.4908,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435549250452644e-05,
+      "loss": 0.5054,
+      "step": 329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431156043807654e-05,
+      "loss": 0.4755,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426746307304154e-05,
+      "loss": 0.5066,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942232004867103e-05,
+      "loss": 0.5016,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941787727566613e-05,
+      "loss": 0.4952,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413417996076244e-05,
+      "loss": 0.4836,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940894221771708e-05,
+      "loss": 0.4828,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404449948433278e-05,
+      "loss": 0.4993,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399941196098382e-05,
+      "loss": 0.4988,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395415968614815e-05,
+      "loss": 0.494,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390874273913884e-05,
+      "loss": 0.4941,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386316119955757e-05,
+      "loss": 0.5065,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381741514729443e-05,
+      "loss": 0.498,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377150466252797e-05,
+      "loss": 0.4879,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937254298257248e-05,
+      "loss": 0.4909,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936791907176397e-05,
+      "loss": 0.4906,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363278741931532e-05,
+      "loss": 0.4897,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358622001208206e-05,
+      "loss": 0.5006,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93539488577558e-05,
+      "loss": 0.4988,
+      "step": 347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349259319764874e-05,
+      "loss": 0.4834,
+      "step": 348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934455339545471e-05,
+      "loss": 0.4821,
+      "step": 349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339831093073317e-05,
+      "loss": 0.4847,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335092420897416e-05,
+      "loss": 0.4958,
+      "step": 351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330337387232406e-05,
+      "loss": 0.4972,
+      "step": 352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325566000412377e-05,
+      "loss": 0.5034,
+      "step": 353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320778268800068e-05,
+      "loss": 0.5027,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931597420078687e-05,
+      "loss": 0.4804,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311153804792814e-05,
+      "loss": 0.5015,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306317089266534e-05,
+      "loss": 0.4709,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301464062685286e-05,
+      "loss": 0.48,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929659473355489e-05,
+      "loss": 0.4793,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291709110409764e-05,
+      "loss": 0.4867,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286807201812866e-05,
+      "loss": 0.4892,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928188901635571e-05,
+      "loss": 0.4753,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276954562658328e-05,
+      "loss": 0.4694,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272003849369274e-05,
+      "loss": 0.4747,
+      "step": 364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926703688516559e-05,
+      "loss": 0.4786,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926205367875281e-05,
+      "loss": 0.4784,
+      "step": 366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925705423886493e-05,
+      "loss": 0.494,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252038574264403e-05,
+      "loss": 0.4885,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247006693742115e-05,
+      "loss": 0.4868,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924195860611737e-05,
+      "loss": 0.4698,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236894320237894e-05,
+      "loss": 0.4923,
+      "step": 371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231813844979777e-05,
+      "loss": 0.4976,
+      "step": 372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226717189247503e-05,
+      "loss": 0.5065,
+      "step": 373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922160436197392e-05,
+      "loss": 0.5122,
+      "step": 374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216475372120198e-05,
+      "loss": 0.4824,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211330228675855e-05,
+      "loss": 0.501,
+      "step": 376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206168940658713e-05,
+      "loss": 0.4773,
+      "step": 377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920099151711489e-05,
+      "loss": 0.4768,
+      "step": 378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195797967118785e-05,
+      "loss": 0.4871,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190588299773064e-05,
+      "loss": 0.4904,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918536252420864e-05,
+      "loss": 0.4689,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180120649584655e-05,
+      "loss": 0.4784,
+      "step": 382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174862685088472e-05,
+      "loss": 0.4787,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916958863993566e-05,
+      "loss": 0.4936,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916429852336996e-05,
+      "loss": 0.478,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158992344663282e-05,
+      "loss": 0.481,
+      "step": 386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153670113115704e-05,
+      "loss": 0.487,
+      "step": 387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148331838055424e-05,
+      "loss": 0.4773,
+      "step": 388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142977528838763e-05,
+      "loss": 0.4755,
+      "step": 389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137607194850145e-05,
+      "loss": 0.4912,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132220845502085e-05,
+      "loss": 0.478,
+      "step": 391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912681849023516e-05,
+      "loss": 0.4838,
+      "step": 392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9121400138518008e-05,
+      "loss": 0.4981,
+      "step": 393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115965799847295e-05,
+      "loss": 0.4782,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110515483747716e-05,
+      "loss": 0.4887,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105049199771963e-05,
+      "loss": 0.4792,
+      "step": 396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909956695750072e-05,
+      "loss": 0.5021,
+      "step": 397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094068766542642e-05,
+      "loss": 0.496,
+      "step": 398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9088554636534323e-05,
+      "loss": 0.4892,
+      "step": 399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908302457714031e-05,
+      "loss": 0.5035,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077478598053062e-05,
+      "loss": 0.4773,
+      "step": 401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071916708992943e-05,
+      "loss": 0.4928,
+      "step": 402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90663389197082e-05,
+      "loss": 0.498,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.906074523997494e-05,
+      "loss": 0.472,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055135679597138e-05,
+      "loss": 0.4764,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049510248406586e-05,
+      "loss": 0.4755,
+      "step": 406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9043868956262913e-05,
+      "loss": 0.4679,
+      "step": 407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903821181305352e-05,
+      "loss": 0.487,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032538828693615e-05,
+      "loss": 0.4675,
+      "step": 409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902685001312616e-05,
+      "loss": 0.4695,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021145376321857e-05,
+      "loss": 0.4733,
+      "step": 411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901542492827915e-05,
+      "loss": 0.4844,
+      "step": 412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.4782,
+      "step": 413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003936638610826e-05,
+      "loss": 0.4907,
+      "step": 414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998168817120578e-05,
+      "loss": 0.4867,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992385224662625e-05,
+      "loss": 0.4781,
+      "step": 416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986585871373792e-05,
+      "loss": 0.4559,
+      "step": 417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898077076741853e-05,
+      "loss": 0.4821,
+      "step": 418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974939922988886e-05,
+      "loss": 0.4977,
+      "step": 419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969093348304507e-05,
+      "loss": 0.4747,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89632310536126e-05,
+      "loss": 0.5069,
+      "step": 421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957353049187936e-05,
+      "loss": 0.47,
+      "step": 422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895145934533281e-05,
+      "loss": 0.4788,
+      "step": 423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894554995237703e-05,
+      "loss": 0.4795,
+      "step": 424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893962488067792e-05,
+      "loss": 0.4944,
+      "step": 425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893368414062026e-05,
+      "loss": 0.4849,
+      "step": 426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892772774261631e-05,
+      "loss": 0.4799,
+      "step": 427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892175569710577e-05,
+      "loss": 0.4956,
+      "step": 428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891576801455576e-05,
+      "loss": 0.4854,
+      "step": 429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890976470546081e-05,
+      "loss": 0.4802,
+      "step": 430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903745780342838e-05,
+      "loss": 0.5011,
+      "step": 431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889771124975113e-05,
+      "loss": 0.4687,
+      "step": 432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889166112426233e-05,
+      "loss": 0.4776,
+      "step": 433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885595414480408e-05,
+      "loss": 0.4779,
+      "step": 434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879514131036643e-05,
+      "loss": 0.4803,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887341728458963e-05,
+      "loss": 0.4629,
+      "step": 436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867304885825222e-05,
+      "loss": 0.4892,
+      "step": 437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8861176945456542e-05,
+      "loss": 0.4708,
+      "step": 438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855033474223937e-05,
+      "loss": 0.4873,
+      "step": 439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848874482894995e-05,
+      "loss": 0.4936,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884269998226449e-05,
+      "loss": 0.4819,
+      "step": 441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883650998315439e-05,
+      "loss": 0.4699,
+      "step": 442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830304496413822e-05,
+      "loss": 0.4809,
+      "step": 443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824083532919054e-05,
+      "loss": 0.4687,
+      "step": 444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817847103573484e-05,
+      "loss": 0.4838,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811595219307622e-05,
+      "loss": 0.4682,
+      "step": 446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805327891079055e-05,
+      "loss": 0.4844,
+      "step": 447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799045129872442e-05,
+      "loss": 0.4718,
+      "step": 448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87927469466995e-05,
+      "loss": 0.4831,
+      "step": 449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786433352598963e-05,
+      "loss": 0.4767,
+      "step": 450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8780104358636583e-05,
+      "loss": 0.4598,
+      "step": 451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773759975905098e-05,
+      "loss": 0.4695,
+      "step": 452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876740021552423e-05,
+      "loss": 0.4832,
+      "step": 453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761025088640633e-05,
+      "loss": 0.469,
+      "step": 454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754634606427914e-05,
+      "loss": 0.476,
+      "step": 455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748228780086578e-05,
+      "loss": 0.4969,
+      "step": 456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741807620844037e-05,
+      "loss": 0.4858,
+      "step": 457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873537113995456e-05,
+      "loss": 0.4853,
+      "step": 458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728919348699285e-05,
+      "loss": 0.4871,
+      "step": 459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722452258386172e-05,
+      "loss": 0.4911,
+      "step": 460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715969880350013e-05,
+      "loss": 0.4711,
+      "step": 461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8709472225952372e-05,
+      "loss": 0.4822,
+      "step": 462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702959306581605e-05,
+      "loss": 0.4674,
+      "step": 463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696431133652818e-05,
+      "loss": 0.4596,
+      "step": 464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868988771860785e-05,
+      "loss": 0.4545,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683329072915253e-05,
+      "loss": 0.4564,
+      "step": 466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676755208070276e-05,
+      "loss": 0.4764,
+      "step": 467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670166135594843e-05,
+      "loss": 0.4607,
+      "step": 468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663561867037533e-05,
+      "loss": 0.4681,
+      "step": 469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656942413973556e-05,
+      "loss": 0.4706,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650307788004735e-05,
+      "loss": 0.4757,
+      "step": 471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643658000759495e-05,
+      "loss": 0.4862,
+      "step": 472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636993063892822e-05,
+      "loss": 0.4806,
+      "step": 473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630312989086257e-05,
+      "loss": 0.4589,
+      "step": 474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623617788047882e-05,
+      "loss": 0.4712,
+      "step": 475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616907472512282e-05,
+      "loss": 0.4593,
+      "step": 476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861018205424053e-05,
+      "loss": 0.4738,
+      "step": 477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860344154502019e-05,
+      "loss": 0.4778,
+      "step": 478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8596685956665245e-05,
+      "loss": 0.463,
+      "step": 479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858991530101613e-05,
+      "loss": 0.487,
+      "step": 480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583129589939682e-05,
+      "loss": 0.4801,
+      "step": 481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576328835329117e-05,
+      "loss": 0.4655,
+      "step": 482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569513049104033e-05,
+      "loss": 0.4578,
+      "step": 483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856268224321036e-05,
+      "loss": 0.464,
+      "step": 484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855583642962036e-05,
+      "loss": 0.4655,
+      "step": 485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85489756203326e-05,
+      "loss": 0.4722,
+      "step": 486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854209982737192e-05,
+      "loss": 0.465,
+      "step": 487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535209062789434e-05,
+      "loss": 0.4594,
+      "step": 488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852830333866249e-05,
+      "loss": 0.4713,
+      "step": 489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521382667094658e-05,
+      "loss": 0.4693,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85144470602157e-05,
+      "loss": 0.4858,
+      "step": 491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850749653018156e-05,
+      "loss": 0.4606,
+      "step": 492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500531089174343e-05,
+      "loss": 0.4819,
+      "step": 493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493550749402278e-05,
+      "loss": 0.4722,
+      "step": 494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486555523099712e-05,
+      "loss": 0.4686,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8479545422527083e-05,
+      "loss": 0.4683,
+      "step": 496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8472520459970896e-05,
+      "loss": 0.4656,
+      "step": 497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465480647743713e-05,
+      "loss": 0.4712,
+      "step": 498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458425998184114e-05,
+      "loss": 0.4701,
+      "step": 499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845135652365668e-05,
+      "loss": 0.4599,
+      "step": 500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844427223655199e-05,
+      "loss": 0.4672,
+      "step": 501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8437173149286582e-05,
+      "loss": 0.4826,
+      "step": 502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843005927430292e-05,
+      "loss": 0.468,
+      "step": 503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8422930624069398e-05,
+      "loss": 0.4929,
+      "step": 504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8415787211080304e-05,
+      "loss": 0.4998,
+      "step": 505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408629047855804e-05,
+      "loss": 0.4825,
+      "step": 506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840145614694191e-05,
+      "loss": 0.4866,
+      "step": 507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394268520910467e-05,
+      "loss": 0.5094,
+      "step": 508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387066182359135e-05,
+      "loss": 0.4804,
+      "step": 509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379849143911342e-05,
+      "loss": 0.4984,
+      "step": 510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8372617418216308e-05,
+      "loss": 0.4976,
+      "step": 511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365371017948966e-05,
+      "loss": 0.4903,
+      "step": 512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358109955809993e-05,
+      "loss": 0.488,
+      "step": 513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8350834244525747e-05,
+      "loss": 0.5153,
+      "step": 514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343543896848275e-05,
+      "loss": 0.5003,
+      "step": 515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8336238925555263e-05,
+      "loss": 0.4941,
+      "step": 516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328919343450036e-05,
+      "loss": 0.4919,
+      "step": 517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832158516336153e-05,
+      "loss": 0.5047,
+      "step": 518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314236398144256e-05,
+      "loss": 0.5078,
+      "step": 519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8306873060678295e-05,
+      "loss": 0.5111,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8299495163869277e-05,
+      "loss": 0.5015,
+      "step": 521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8292102720648333e-05,
+      "loss": 0.4974,
+      "step": 522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284695743972095e-05,
+      "loss": 0.4987,
+      "step": 523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827727424682268e-05,
+      "loss": 0.5077,
+      "step": 524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269838242207634e-05,
+      "loss": 0.4841,
+      "step": 525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826238774315995e-05,
+      "loss": 0.4964,
+      "step": 526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825492276273801e-05,
+      "loss": 0.5264,
+      "step": 527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247443314025585e-05,
+      "loss": 0.5069,
+      "step": 528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8239949410131803e-05,
+      "loss": 0.5267,
+      "step": 529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232441064191125e-05,
+      "loss": 0.4958,
+      "step": 530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8224918289363333e-05,
+      "loss": 0.5146,
+      "step": 531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217381098833482e-05,
+      "loss": 0.5141,
+      "step": 532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820982950581191e-05,
+      "loss": 0.5085,
+      "step": 533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820226352353419e-05,
+      "loss": 0.5064,
+      "step": 534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194683165261116e-05,
+      "loss": 0.5163,
+      "step": 535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187088444278675e-05,
+      "loss": 0.5159,
+      "step": 536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179479373898036e-05,
+      "loss": 0.505,
+      "step": 537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171855967455508e-05,
+      "loss": 0.5051,
+      "step": 538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8164218238312534e-05,
+      "loss": 0.5154,
+      "step": 539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8156566199855657e-05,
+      "loss": 0.5,
+      "step": 540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148899865496503e-05,
+      "loss": 0.5209,
+      "step": 541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8141219248671747e-05,
+      "loss": 0.5142,
+      "step": 542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133524362843105e-05,
+      "loss": 0.5225,
+      "step": 543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125815221497296e-05,
+      "loss": 0.5006,
+      "step": 544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811809183814603e-05,
+      "loss": 0.495,
+      "step": 545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811035422632597e-05,
+      "loss": 0.5136,
+      "step": 546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810260239959873e-05,
+      "loss": 0.512,
+      "step": 547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094836371550823e-05,
+      "loss": 0.4905,
+      "step": 548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808705615579367e-05,
+      "loss": 0.4971,
+      "step": 549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8079261765963537e-05,
+      "loss": 0.505,
+      "step": 550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8071453215721554e-05,
+      "loss": 0.4935,
+      "step": 551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806363051875366e-05,
+      "loss": 0.4876,
+      "step": 552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8055793688770586e-05,
+      "loss": 0.4881,
+      "step": 553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047942739507836e-05,
+      "loss": 0.4986,
+      "step": 554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040077684725667e-05,
+      "loss": 0.5062,
+      "step": 555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8032198538209042e-05,
+      "loss": 0.5159,
+      "step": 556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024305313767648e-05,
+      "loss": 0.4897,
+      "step": 557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016398025235822e-05,
+      "loss": 0.5107,
+      "step": 558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008476686472563e-05,
+      "loss": 0.5049,
+      "step": 559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80005413113615e-05,
+      "loss": 0.5133,
+      "step": 560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992591913810846e-05,
+      "loss": 0.4958,
+      "step": 561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984628507753406e-05,
+      "loss": 0.5078,
+      "step": 562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7976651107146533e-05,
+      "loss": 0.4842,
+      "step": 563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7968659725972113e-05,
+      "loss": 0.5005,
+      "step": 564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796065437823652e-05,
+      "loss": 0.4892,
+      "step": 565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795263507797063e-05,
+      "loss": 0.5065,
+      "step": 566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944601839229755e-05,
+      "loss": 0.5102,
+      "step": 567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793655467609364e-05,
+      "loss": 0.5007,
+      "step": 568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928493602666446e-05,
+      "loss": 0.4887,
+      "step": 569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79204186330767e-05,
+      "loss": 0.4867,
+      "step": 570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912329781477287e-05,
+      "loss": 0.4881,
+      "step": 571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790422706204544e-05,
+      "loss": 0.4992,
+      "step": 572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789611048898267e-05,
+      "loss": 0.4891,
+      "step": 573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7887980076514798e-05,
+      "loss": 0.4857,
+      "step": 574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879835838891877e-05,
+      "loss": 0.4871,
+      "step": 575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871677790388206e-05,
+      "loss": 0.4987,
+      "step": 576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786350594530228e-05,
+      "loss": 0.5017,
+      "step": 577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7855320317956785e-05,
+      "loss": 0.4908,
+      "step": 578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784712092269856e-05,
+      "loss": 0.4795,
+      "step": 579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783890777389857e-05,
+      "loss": 0.4932,
+      "step": 580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783068088595189e-05,
+      "loss": 0.4821,
+      "step": 581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7822440273277674e-05,
+      "loss": 0.504,
+      "step": 582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7814185950319127e-05,
+      "loss": 0.4856,
+      "step": 583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805917931543493e-05,
+      "loss": 0.495,
+      "step": 584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7797636231442018e-05,
+      "loss": 0.481,
+      "step": 585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778934086452992e-05,
+      "loss": 0.494,
+      "step": 586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7781031845346375e-05,
+      "loss": 0.4904,
+      "step": 587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777270918845449e-05,
+      "loss": 0.5166,
+      "step": 588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7764372908441276e-05,
+      "loss": 0.4718,
+      "step": 589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7756023019917607e-05,
+      "loss": 0.5031,
+      "step": 590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774765953751823e-05,
+      "loss": 0.4923,
+      "step": 591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77392824759017e-05,
+      "loss": 0.4918,
+      "step": 592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7730891849750377e-05,
+      "loss": 0.496,
+      "step": 593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722487673770403e-05,
+      "loss": 0.4894,
+      "step": 594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7714069962691656e-05,
+      "loss": 0.4847,
+      "step": 595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705638731267747e-05,
+      "loss": 0.4839,
+      "step": 596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7697193994275983e-05,
+      "loss": 0.4854,
+      "step": 597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7688735766517334e-05,
+      "loss": 0.4852,
+      "step": 598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768026406281642e-05,
+      "loss": 0.4914,
+      "step": 599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671778898021487e-05,
+      "loss": 0.4956,
+      "step": 600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663280287004364e-05,
+      "loss": 0.4841,
+      "step": 601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765476824466045e-05,
+      "loss": 0.4601,
+      "step": 602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7646242785908684e-05,
+      "loss": 0.4761,
+      "step": 603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763770392569152e-05,
+      "loss": 0.4624,
+      "step": 604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762915167897491e-05,
+      "loss": 0.4758,
+      "step": 605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762058606074825e-05,
+      "loss": 0.4768,
+      "step": 606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761200708602439e-05,
+      "loss": 0.5059,
+      "step": 607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760341476983958e-05,
+      "loss": 0.4665,
+      "step": 608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759480912725345e-05,
+      "loss": 0.476,
+      "step": 609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7586190173349e-05,
+      "loss": 0.4744,
+      "step": 610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577557923232548e-05,
+      "loss": 0.5007,
+      "step": 611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7568912392033723e-05,
+      "loss": 0.4793,
+      "step": 612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560253594905425e-05,
+      "loss": 0.4923,
+      "step": 613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755158154702382e-05,
+      "loss": 0.4787,
+      "step": 614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542896263588274e-05,
+      "loss": 0.4664,
+      "step": 615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753419775982138e-05,
+      "loss": 0.481,
+      "step": 616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525486050968875e-05,
+      "loss": 0.4558,
+      "step": 617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7516761152299656e-05,
+      "loss": 0.4692,
+      "step": 618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7508023079105737e-05,
+      "loss": 0.4898,
+      "step": 619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499271846702216e-05,
+      "loss": 0.4725,
+      "step": 620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749050747042726e-05,
+      "loss": 0.474,
+      "step": 621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7481729965642068e-05,
+      "loss": 0.488,
+      "step": 622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7472939347730857e-05,
+      "loss": 0.4752,
+      "step": 623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7464135632100824e-05,
+      "loss": 0.4772,
+      "step": 624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7455318834182117e-05,
+      "loss": 0.4875,
+      "step": 625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7446488969427823e-05,
+      "loss": 0.4623,
+      "step": 626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743764605331392e-05,
+      "loss": 0.4698,
+      "step": 627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742879010133927e-05,
+      "loss": 0.4782,
+      "step": 628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7419921129025578e-05,
+      "loss": 0.472,
+      "step": 629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7411039151917368e-05,
+      "loss": 0.4653,
+      "step": 630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402144185581965e-05,
+      "loss": 0.4847,
+      "step": 631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739323624560945e-05,
+      "loss": 0.474,
+      "step": 632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384315347612655e-05,
+      "loss": 0.4813,
+      "step": 633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7375381507227108e-05,
+      "loss": 0.4798,
+      "step": 634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7366434740111036e-05,
+      "loss": 0.4735,
+      "step": 635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735747506194531e-05,
+      "loss": 0.4848,
+      "step": 636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7348502488433442e-05,
+      "loss": 0.4827,
+      "step": 637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339517035301533e-05,
+      "loss": 0.4835,
+      "step": 638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7330518718298263e-05,
+      "loss": 0.469,
+      "step": 639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7321507553194866e-05,
+      "loss": 0.4808,
+      "step": 640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7312483555785087e-05,
+      "loss": 0.4597,
+      "step": 641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730344674188516e-05,
+      "loss": 0.4709,
+      "step": 642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294397127333785e-05,
+      "loss": 0.4743,
+      "step": 643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7285334727992104e-05,
+      "loss": 0.4668,
+      "step": 644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727625955974365e-05,
+      "loss": 0.4747,
+      "step": 645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7267171638494358e-05,
+      "loss": 0.4955,
+      "step": 646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7258070980172494e-05,
+      "loss": 0.4724,
+      "step": 647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7248957600728664e-05,
+      "loss": 0.5047,
+      "step": 648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723983151613576e-05,
+      "loss": 0.4701,
+      "step": 649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723069274238895e-05,
+      "loss": 0.4809,
+      "step": 650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7221541295505638e-05,
+      "loss": 0.4644,
+      "step": 651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212377191525434e-05,
+      "loss": 0.4582,
+      "step": 652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7203200446510143e-05,
+      "loss": 0.4637,
+      "step": 653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7194011076543716e-05,
+      "loss": 0.4676,
+      "step": 654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7184809097732247e-05,
+      "loss": 0.4913,
+      "step": 655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175594526203906e-05,
+      "loss": 0.4746,
+      "step": 656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7166367378108954e-05,
+      "loss": 0.4765,
+      "step": 657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715712766961969e-05,
+      "loss": 0.4588,
+      "step": 658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7147875416930414e-05,
+      "loss": 0.47,
+      "step": 659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7138610636257434e-05,
+      "loss": 0.4826,
+      "step": 660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129333343839004e-05,
+      "loss": 0.4597,
+      "step": 661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71200435559353e-05,
+      "loss": 0.4692,
+      "step": 662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711074128882841e-05,
+      "loss": 0.4618,
+      "step": 663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7101426558822292e-05,
+      "loss": 0.4804,
+      "step": 664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709209938224275e-05,
+      "loss": 0.474,
+      "step": 665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082759775437392e-05,
+      "loss": 0.4746,
+      "step": 666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707340775477562e-05,
+      "loss": 0.4397,
+      "step": 667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70640433366486e-05,
+      "loss": 0.47,
+      "step": 668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7054666537469213e-05,
+      "loss": 0.4829,
+      "step": 669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704527737367205e-05,
+      "loss": 0.465,
+      "step": 670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703587586171337e-05,
+      "loss": 0.4425,
+      "step": 671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702646201807107e-05,
+      "loss": 0.4705,
+      "step": 672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7017035859244674e-05,
+      "loss": 0.467,
+      "step": 673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7007597401755277e-05,
+      "loss": 0.4667,
+      "step": 674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699814666214553e-05,
+      "loss": 0.4552,
+      "step": 675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988683656979624e-05,
+      "loss": 0.4717,
+      "step": 676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979208402843235e-05,
+      "loss": 0.4666,
+      "step": 677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6969720916343515e-05,
+      "loss": 0.464,
+      "step": 678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960221214109046e-05,
+      "loss": 0.4759,
+      "step": 679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695070931278983e-05,
+      "loss": 0.4769,
+      "step": 680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694118522905725e-05,
+      "loss": 0.4682,
+      "step": 681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693164897960403e-05,
+      "loss": 0.4671,
+      "step": 682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6922100581144228e-05,
+      "loss": 0.4815,
+      "step": 683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6912540050413196e-05,
+      "loss": 0.4722,
+      "step": 684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6902967404167538e-05,
+      "loss": 0.471,
+      "step": 685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6893382659185106e-05,
+      "loss": 0.4625,
+      "step": 686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688378583226495e-05,
+      "loss": 0.4861,
+      "step": 687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874176940227296e-05,
+      "loss": 0.4679,
+      "step": 688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686455599991352e-05,
+      "loss": 0.4476,
+      "step": 689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6854923028186112e-05,
+      "loss": 0.4627,
+      "step": 690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684527804192865e-05,
+      "loss": 0.473,
+      "step": 691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683562105804577e-05,
+      "loss": 0.465,
+      "step": 692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6825952093463137e-05,
+      "loss": 0.4854,
+      "step": 693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681627116512741e-05,
+      "loss": 0.4525,
+      "step": 694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6806578290006225e-05,
+      "loss": 0.4478,
+      "step": 695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796873485088147e-05,
+      "loss": 0.4596,
+      "step": 696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678715676738266e-05,
+      "loss": 0.4701,
+      "step": 697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677742815392012e-05,
+      "loss": 0.4715,
+      "step": 698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6767687661751734e-05,
+      "loss": 0.4774,
+      "step": 699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675793530794953e-05,
+      "loss": 0.4654,
+      "step": 700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674817110960633e-05,
+      "loss": 0.4646,
+      "step": 701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67383950838357e-05,
+      "loss": 0.4502,
+      "step": 702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6728607247771957e-05,
+      "loss": 0.4586,
+      "step": 703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671880761857011e-05,
+      "loss": 0.4438,
+      "step": 704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6708996213405826e-05,
+      "loss": 0.4471,
+      "step": 705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6699173049475425e-05,
+      "loss": 0.4589,
+      "step": 706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6689338143995835e-05,
+      "loss": 0.4815,
+      "step": 707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679491514204555e-05,
+      "loss": 0.4739,
+      "step": 708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6669633177359647e-05,
+      "loss": 0.4626,
+      "step": 709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6659763150739675e-05,
+      "loss": 0.4565,
+      "step": 710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6649881451643706e-05,
+      "loss": 0.4726,
+      "step": 711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6639988097391252e-05,
+      "loss": 0.4472,
+      "step": 712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6630083105322267e-05,
+      "loss": 0.4574,
+      "step": 713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6620166492797095e-05,
+      "loss": 0.4668,
+      "step": 714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6610238277196448e-05,
+      "loss": 0.4531,
+      "step": 715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6600298475921367e-05,
+      "loss": 0.4514,
+      "step": 716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659034710639321e-05,
+      "loss": 0.4648,
+      "step": 717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658038418605361e-05,
+      "loss": 0.4621,
+      "step": 718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657040973236444e-05,
+      "loss": 0.453,
+      "step": 719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6560423762807783e-05,
+      "loss": 0.4446,
+      "step": 720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655042629488592e-05,
+      "loss": 0.4648,
+      "step": 721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6540417346121272e-05,
+      "loss": 0.4659,
+      "step": 722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6530396934056386e-05,
+      "loss": 0.486,
+      "step": 723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6520365076253907e-05,
+      "loss": 0.4692,
+      "step": 724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6510321790296527e-05,
+      "loss": 0.4676,
+      "step": 725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6500267093786983e-05,
+      "loss": 0.4789,
+      "step": 726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6490201004348003e-05,
+      "loss": 0.4647,
+      "step": 727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6480123539622283e-05,
+      "loss": 0.4613,
+      "step": 728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6470034717272456e-05,
+      "loss": 0.4729,
+      "step": 729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6459934554981067e-05,
+      "loss": 0.4898,
+      "step": 730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644982307045053e-05,
+      "loss": 0.4618,
+      "step": 731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439700281403113e-05,
+      "loss": 0.4481,
+      "step": 732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6429566205580886e-05,
+      "loss": 0.4568,
+      "step": 733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6419420860745698e-05,
+      "loss": 0.4547,
+      "step": 734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6409264264679165e-05,
+      "loss": 0.4616,
+      "step": 735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6399096435182614e-05,
+      "loss": 0.4458,
+      "step": 736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6388917390077054e-05,
+      "loss": 0.4601,
+      "step": 737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6378727147203166e-05,
+      "loss": 0.4579,
+      "step": 738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6368525724421248e-05,
+      "loss": 0.4561,
+      "step": 739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6358313139611194e-05,
+      "loss": 0.4589,
+      "step": 740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634808941067246e-05,
+      "loss": 0.4759,
+      "step": 741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633785455552404e-05,
+      "loss": 0.4466,
+      "step": 742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632760859210442e-05,
+      "loss": 0.4711,
+      "step": 743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6317351538371563e-05,
+      "loss": 0.4506,
+      "step": 744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630708341230287e-05,
+      "loss": 0.4545,
+      "step": 745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629680423189514e-05,
+      "loss": 0.4469,
+      "step": 746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628651401516456e-05,
+      "loss": 0.4547,
+      "step": 747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627621278014664e-05,
+      "loss": 0.4648,
+      "step": 748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6265900544896223e-05,
+      "loss": 0.4727,
+      "step": 749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6255577327487425e-05,
+      "loss": 0.4587,
+      "step": 750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6245243146013602e-05,
+      "loss": 0.4491,
+      "step": 751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.4612,
+      "step": 752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6224541963340392e-05,
+      "loss": 0.4592,
+      "step": 753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621417499842368e-05,
+      "loss": 0.4407,
+      "step": 754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620379714200725e-05,
+      "loss": 0.4613,
+      "step": 755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6193408412280217e-05,
+      "loss": 0.4712,
+      "step": 756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618300882745077e-05,
+      "loss": 0.4578,
+      "step": 757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6172598405746125e-05,
+      "loss": 0.4561,
+      "step": 758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616217716541248e-05,
+      "loss": 0.4668,
+      "step": 759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6151745124715003e-05,
+      "loss": 0.4644,
+      "step": 760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141302301937785e-05,
+      "loss": 0.4696,
+      "step": 761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6130848715383827e-05,
+      "loss": 0.4581,
+      "step": 762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612038438337498e-05,
+      "loss": 0.4394,
+      "step": 763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610990932425194e-05,
+      "loss": 0.4502,
+      "step": 764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6099423556374198e-05,
+      "loss": 0.4598,
+      "step": 765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608892709812002e-05,
+      "loss": 0.454,
+      "step": 766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6078419967886402e-05,
+      "loss": 0.4723,
+      "step": 767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606790218408905e-05,
+      "loss": 0.4469,
+      "step": 768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057373765162333e-05,
+      "loss": 0.4587,
+      "step": 769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604683472955928e-05,
+      "loss": 0.4743,
+      "step": 770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60362850957515e-05,
+      "loss": 0.4547,
+      "step": 771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6025724882229206e-05,
+      "loss": 0.4532,
+      "step": 772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015154107501132e-05,
+      "loss": 0.4397,
+      "step": 773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6004572790094535e-05,
+      "loss": 0.4582,
+      "step": 774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599398094855514e-05,
+      "loss": 0.4433,
+      "step": 775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5983378601447128e-05,
+      "loss": 0.4469,
+      "step": 776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5972765767353088e-05,
+      "loss": 0.46,
+      "step": 777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5962142464873985e-05,
+      "loss": 0.4472,
+      "step": 778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595150871262914e-05,
+      "loss": 0.4534,
+      "step": 779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940864529256187e-05,
+      "loss": 0.469,
+      "step": 780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5930209933411036e-05,
+      "loss": 0.4459,
+      "step": 781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5919544943767856e-05,
+      "loss": 0.4687,
+      "step": 782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5908869579019025e-05,
+      "loss": 0.4493,
+      "step": 783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5898183857875115e-05,
+      "loss": 0.4464,
+      "step": 784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588748779906484e-05,
+      "loss": 0.4519,
+      "step": 785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5876781421335033e-05,
+      "loss": 0.4636,
+      "step": 786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5866064743450618e-05,
+      "loss": 0.4473,
+      "step": 787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5855337784194576e-05,
+      "loss": 0.4522,
+      "step": 788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58446005623679e-05,
+      "loss": 0.466,
+      "step": 789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5833853096789566e-05,
+      "loss": 0.4587,
+      "step": 790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823095406296515e-05,
+      "loss": 0.4672,
+      "step": 791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5812327509743603e-05,
+      "loss": 0.4483,
+      "step": 792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5801549426003578e-05,
+      "loss": 0.4576,
+      "step": 793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5790761173967036e-05,
+      "loss": 0.4563,
+      "step": 794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5779962772542404e-05,
+      "loss": 0.4708,
+      "step": 795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5769154240655886e-05,
+      "loss": 0.4536,
+      "step": 796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5758335597251456e-05,
+      "loss": 0.4689,
+      "step": 797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747506861290796e-05,
+      "loss": 0.4792,
+      "step": 798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573666805175329e-05,
+      "loss": 0.4486,
+      "step": 799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5725819187635968e-05,
+      "loss": 0.4469,
+      "step": 800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5714960287953487e-05,
+      "loss": 0.4336,
+      "step": 801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570409137173809e-05,
+      "loss": 0.4695,
+      "step": 802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693212458039585e-05,
+      "loss": 0.4608,
+      "step": 803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682323565925287e-05,
+      "loss": 0.4766,
+      "step": 804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5671424714480012e-05,
+      "loss": 0.4645,
+      "step": 805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566051592280603e-05,
+      "loss": 0.4559,
+      "step": 806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5649597210023027e-05,
+      "loss": 0.4561,
+      "step": 807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5638668595268086e-05,
+      "loss": 0.4374,
+      "step": 808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562773009769564e-05,
+      "loss": 0.4457,
+      "step": 809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5616781736477445e-05,
+      "loss": 0.4557,
+      "step": 810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5605823530802543e-05,
+      "loss": 0.4366,
+      "step": 811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559485549987723e-05,
+      "loss": 0.4714,
+      "step": 812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5583877662925032e-05,
+      "loss": 0.459,
+      "step": 813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557289003918664e-05,
+      "loss": 0.4627,
+      "step": 814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5561892647919922e-05,
+      "loss": 0.4583,
+      "step": 815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5550885508399857e-05,
+      "loss": 0.4599,
+      "step": 816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5539868639918507e-05,
+      "loss": 0.4486,
+      "step": 817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552884206178498e-05,
+      "loss": 0.443,
+      "step": 818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551780579332542e-05,
+      "loss": 0.4501,
+      "step": 819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5506759853882936e-05,
+      "loss": 0.4553,
+      "step": 820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5495704262817595e-05,
+      "loss": 0.4455,
+      "step": 821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548463903950639e-05,
+      "loss": 0.4385,
+      "step": 822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5473564203343173e-05,
+      "loss": 0.4604,
+      "step": 823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546247977373867e-05,
+      "loss": 0.4361,
+      "step": 824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5451385770120402e-05,
+      "loss": 0.4331,
+      "step": 825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544028221193268e-05,
+      "loss": 0.4377,
+      "step": 826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5429169118636566e-05,
+      "loss": 0.4435,
+      "step": 827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5418046509709817e-05,
+      "loss": 0.4321,
+      "step": 828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540691440464688e-05,
+      "loss": 0.4623,
+      "step": 829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5395772822958844e-05,
+      "loss": 0.4197,
+      "step": 830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5384621784173415e-05,
+      "loss": 0.4325,
+      "step": 831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5373461307834854e-05,
+      "loss": 0.4573,
+      "step": 832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362291413503984e-05,
+      "loss": 0.4486,
+      "step": 833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5351112120758122e-05,
+      "loss": 0.4512,
+      "step": 834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5339923449191067e-05,
+      "loss": 0.4435,
+      "step": 835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328725418413045e-05,
+      "loss": 0.4255,
+      "step": 836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5317518048050698e-05,
+      "loss": 0.4336,
+      "step": 837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530630135774702e-05,
+      "loss": 0.4449,
+      "step": 838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5295075367161366e-05,
+      "loss": 0.4451,
+      "step": 839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5283840095969365e-05,
+      "loss": 0.4373,
+      "step": 840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5272595563862935e-05,
+      "loss": 0.45,
+      "step": 841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5261341790550196e-05,
+      "loss": 0.4528,
+      "step": 842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5250078795755498e-05,
+      "loss": 0.4409,
+      "step": 843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5238806599219337e-05,
+      "loss": 0.4372,
+      "step": 844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5227525220698332e-05,
+      "loss": 0.4446,
+      "step": 845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5216234679965205e-05,
+      "loss": 0.4599,
+      "step": 846
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5204934996808729e-05,
+      "loss": 0.4552,
+      "step": 847
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5193626191033713e-05,
+      "loss": 0.4396,
+      "step": 848
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5182308282460936e-05,
+      "loss": 0.4738,
+      "step": 849
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5170981290927148e-05,
+      "loss": 0.4494,
+      "step": 850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515964523628501e-05,
+      "loss": 0.4406,
+      "step": 851
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5148300138403076e-05,
+      "loss": 0.4571,
+      "step": 852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136946017165734e-05,
+      "loss": 0.4525,
+      "step": 853
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5125582892473206e-05,
+      "loss": 0.4284,
+      "step": 854
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114210784241483e-05,
+      "loss": 0.4499,
+      "step": 855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5102829712402303e-05,
+      "loss": 0.4525,
+      "step": 856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5091439696903116e-05,
+      "loss": 0.4389,
+      "step": 857
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080040757707045e-05,
+      "loss": 0.4437,
+      "step": 858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068632914792862e-05,
+      "loss": 0.4525,
+      "step": 859
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5057216188154928e-05,
+      "loss": 0.4521,
+      "step": 860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045790597803191e-05,
+      "loss": 0.448,
+      "step": 861
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5034356163763125e-05,
+      "loss": 0.4311,
+      "step": 862
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5022912906075703e-05,
+      "loss": 0.4183,
+      "step": 863
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5011460844797372e-05,
+      "loss": 0.4525,
+      "step": 864
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.4457,
+      "step": 865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4988530391770857e-05,
+      "loss": 0.4608,
+      "step": 866
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4977052040212567e-05,
+      "loss": 0.4458,
+      "step": 867
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4965564965443077e-05,
+      "loss": 0.4626,
+      "step": 868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4954069187595633e-05,
+      "loss": 0.4296,
+      "step": 869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4942564726818721e-05,
+      "loss": 0.4506,
+      "step": 870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4931051603276055e-05,
+      "loss": 0.4327,
+      "step": 871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4919529837146529e-05,
+      "loss": 0.4453,
+      "step": 872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4907999448624188e-05,
+      "loss": 0.4361,
+      "step": 873
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4896460457918184e-05,
+      "loss": 0.4316,
+      "step": 874
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4884912885252751e-05,
+      "loss": 0.4209,
+      "step": 875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.487335675086716e-05,
+      "loss": 0.429,
+      "step": 876
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861792075015688e-05,
+      "loss": 0.4571,
+      "step": 877
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485021887796759e-05,
+      "loss": 0.4437,
+      "step": 878
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4838637180007048e-05,
+      "loss": 0.4495,
+      "step": 879
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4827047001433142e-05,
+      "loss": 0.4319,
+      "step": 880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4815448362559826e-05,
+      "loss": 0.4399,
+      "step": 881
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4803841283715872e-05,
+      "loss": 0.4343,
+      "step": 882
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4792225785244849e-05,
+      "loss": 0.437,
+      "step": 883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4780601887505088e-05,
+      "loss": 0.4444,
+      "step": 884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4768969610869628e-05,
+      "loss": 0.4479,
+      "step": 885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4757328975726207e-05,
+      "loss": 0.4427,
+      "step": 886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4745680002477206e-05,
+      "loss": 0.4317,
+      "step": 887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4734022711539619e-05,
+      "loss": 0.4406,
+      "step": 888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722357123345023e-05,
+      "loss": 0.4444,
+      "step": 889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4710683258339536e-05,
+      "loss": 0.4472,
+      "step": 890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4699001136983783e-05,
+      "loss": 0.4521,
+      "step": 891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4687310779752854e-05,
+      "loss": 0.4407,
+      "step": 892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4675612207136283e-05,
+      "loss": 0.4516,
+      "step": 893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4663905439637995e-05,
+      "loss": 0.4589,
+      "step": 894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4652190497776287e-05,
+      "loss": 0.441,
+      "step": 895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4640467402083772e-05,
+      "loss": 0.4344,
+      "step": 896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4628736173107362e-05,
+      "loss": 0.457,
+      "step": 897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4616996831408222e-05,
+      "loss": 0.4594,
+      "step": 898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4605249397561735e-05,
+      "loss": 0.4272,
+      "step": 899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593493892157473e-05,
+      "loss": 0.4437,
+      "step": 900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4581730335799143e-05,
+      "loss": 0.4711,
+      "step": 901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4569958749104576e-05,
+      "loss": 0.4558,
+      "step": 902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4558179152705663e-05,
+      "loss": 0.4402,
+      "step": 903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4546391567248353e-05,
+      "loss": 0.4382,
+      "step": 904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534596013392574e-05,
+      "loss": 0.4354,
+      "step": 905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522792511812242e-05,
+      "loss": 0.4506,
+      "step": 906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510981083195188e-05,
+      "loss": 0.4411,
+      "step": 907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4499161748243147e-05,
+      "loss": 0.4403,
+      "step": 908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44873345276717e-05,
+      "loss": 0.4505,
+      "step": 909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447549944221026e-05,
+      "loss": 0.433,
+      "step": 910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4463656512602012e-05,
+      "loss": 0.4509,
+      "step": 911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4451805759603907e-05,
+      "loss": 0.4446,
+      "step": 912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443994720398659e-05,
+      "loss": 0.4471,
+      "step": 913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428080866534397e-05,
+      "loss": 0.4278,
+      "step": 914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441620676804529e-05,
+      "loss": 0.4401,
+      "step": 915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4404324929330842e-05,
+      "loss": 0.4483,
+      "step": 916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392435371216188e-05,
+      "loss": 0.4466,
+      "step": 917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4380538114539998e-05,
+      "loss": 0.4487,
+      "step": 918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368633180154423e-05,
+      "loss": 0.4453,
+      "step": 919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435672058892509e-05,
+      "loss": 0.4368,
+      "step": 920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4344800361731028e-05,
+      "loss": 0.4521,
+      "step": 921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433287251946466e-05,
+      "loss": 0.4436,
+      "step": 922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4320937083031748e-05,
+      "loss": 0.4509,
+      "step": 923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4308994073351373e-05,
+      "loss": 0.4474,
+      "step": 924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4297043511355882e-05,
+      "loss": 0.4327,
+      "step": 925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285085417990863e-05,
+      "loss": 0.4459,
+      "step": 926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4273119814215102e-05,
+      "loss": 0.4453,
+      "step": 927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4261146721000554e-05,
+      "loss": 0.4317,
+      "step": 928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424916615933229e-05,
+      "loss": 0.4347,
+      "step": 929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237178150208486e-05,
+      "loss": 0.4434,
+      "step": 930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422518271464035e-05,
+      "loss": 0.4561,
+      "step": 931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4213179873652127e-05,
+      "loss": 0.4231,
+      "step": 932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4201169648281027e-05,
+      "loss": 0.4278,
+      "step": 933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4189152059577214e-05,
+      "loss": 0.4509,
+      "step": 934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4177127128603748e-05,
+      "loss": 0.4403,
+      "step": 935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4165094876436562e-05,
+      "loss": 0.4341,
+      "step": 936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4153055324164417e-05,
+      "loss": 0.435,
+      "step": 937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141008492888881e-05,
+      "loss": 0.4381,
+      "step": 938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412895440372426e-05,
+      "loss": 0.4586,
+      "step": 939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41168930777976e-05,
+      "loss": 0.4318,
+      "step": 940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4104824536248615e-05,
+      "loss": 0.4561,
+      "step": 941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092748800229684e-05,
+      "loss": 0.4469,
+      "step": 942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408066589090577e-05,
+      "loss": 0.4658,
+      "step": 943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4068575829454436e-05,
+      "loss": 0.4384,
+      "step": 944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4056478637065763e-05,
+      "loss": 0.4503,
+      "step": 945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4044374334942333e-05,
+      "loss": 0.4295,
+      "step": 946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4032262944299193e-05,
+      "loss": 0.4217,
+      "step": 947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4020144486363813e-05,
+      "loss": 0.4235,
+      "step": 948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4008018982376045e-05,
+      "loss": 0.4279,
+      "step": 949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3995886453588103e-05,
+      "loss": 0.4313,
+      "step": 950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3983746921264496e-05,
+      "loss": 0.4277,
+      "step": 951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397160040668202e-05,
+      "loss": 0.4551,
+      "step": 952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3959446931129703e-05,
+      "loss": 0.4298,
+      "step": 953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394728651590878e-05,
+      "loss": 0.4367,
+      "step": 954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3935119182332641e-05,
+      "loss": 0.4307,
+      "step": 955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3922944951726811e-05,
+      "loss": 0.4453,
+      "step": 956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391076384542889e-05,
+      "loss": 0.4337,
+      "step": 957
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3898575884788544e-05,
+      "loss": 0.4381,
+      "step": 958
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388638109116744e-05,
+      "loss": 0.4425,
+      "step": 959
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.387417948593923e-05,
+      "loss": 0.457,
+      "step": 960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861971090489498e-05,
+      "loss": 0.4484,
+      "step": 961
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3849755926215736e-05,
+      "loss": 0.4469,
+      "step": 962
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3837534014527292e-05,
+      "loss": 0.424,
+      "step": 963
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3825305376845346e-05,
+      "loss": 0.4352,
+      "step": 964
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3813070034602862e-05,
+      "loss": 0.4466,
+      "step": 965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.380082800924456e-05,
+      "loss": 0.4292,
+      "step": 966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3788579322226869e-05,
+      "loss": 0.4486,
+      "step": 967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3776323995017897e-05,
+      "loss": 0.4343,
+      "step": 968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376406204909739e-05,
+      "loss": 0.4446,
+      "step": 969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375179350595669e-05,
+      "loss": 0.4419,
+      "step": 970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3739518387098704e-05,
+      "loss": 0.4333,
+      "step": 971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3727236714037873e-05,
+      "loss": 0.4507,
+      "step": 972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.371494850830011e-05,
+      "loss": 0.4312,
+      "step": 973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702653791422788e-05,
+      "loss": 0.4314,
+      "step": 974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369035258495469e-05,
+      "loss": 0.4333,
+      "step": 975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678044910455975e-05,
+      "loss": 0.4303,
+      "step": 976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366573078949813e-05,
+      "loss": 0.4412,
+      "step": 977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3653410243663953e-05,
+      "loss": 0.4354,
+      "step": 978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3641083294547492e-05,
+      "loss": 0.4319,
+      "step": 979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3628749963754026e-05,
+      "loss": 0.4545,
+      "step": 980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3616410272900014e-05,
+      "loss": 0.4235,
+      "step": 981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3604064243613063e-05,
+      "loss": 0.4455,
+      "step": 982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359171189753189e-05,
+      "loss": 0.4345,
+      "step": 983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3579353256306287e-05,
+      "loss": 0.4333,
+      "step": 984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3566988341597069e-05,
+      "loss": 0.4346,
+      "step": 985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3554617175076064e-05,
+      "loss": 0.4433,
+      "step": 986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3542239778426034e-05,
+      "loss": 0.439,
+      "step": 987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529856173340683e-05,
+      "loss": 0.4313,
+      "step": 988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517466381524582e-05,
+      "loss": 0.4508,
+      "step": 989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3505070424693152e-05,
+      "loss": 0.4358,
+      "step": 990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3492668324572615e-05,
+      "loss": 0.4379,
+      "step": 991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3480260102899967e-05,
+      "loss": 0.4375,
+      "step": 992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467845781422924e-05,
+      "loss": 0.4421,
+      "step": 993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34554253818999e-05,
+      "loss": 0.4204,
+      "step": 994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442998926099959e-05,
+      "loss": 0.4382,
+      "step": 995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3430566435802783e-05,
+      "loss": 0.4318,
+      "step": 996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3418127932798623e-05,
+      "loss": 0.4343,
+      "step": 997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3405683438888281e-05,
+      "loss": 0.4302,
+      "step": 998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393232975883044e-05,
+      "loss": 0.4272,
+      "step": 999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3380776565604677e-05,
+      "loss": 0.4407,
+      "step": 1000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3368314229885349e-05,
+      "loss": 0.4479,
+      "step": 1001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3355845990567635e-05,
+      "loss": 0.4196,
+      "step": 1002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343371869504444e-05,
+      "loss": 0.4343,
+      "step": 1003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3330891888559002e-05,
+      "loss": 0.4529,
+      "step": 1004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3318406069604794e-05,
+      "loss": 0.4363,
+      "step": 1005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3305914434525551e-05,
+      "loss": 0.4369,
+      "step": 1006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3293417005215187e-05,
+      "loss": 0.4372,
+      "step": 1007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3280913803577781e-05,
+      "loss": 0.4501,
+      "step": 1008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3268404851527518e-05,
+      "loss": 0.4279,
+      "step": 1009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325589017098867e-05,
+      "loss": 0.4512,
+      "step": 1010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3243369783895548e-05,
+      "loss": 0.4295,
+      "step": 1011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3230843712192463e-05,
+      "loss": 0.4176,
+      "step": 1012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321831197783369e-05,
+      "loss": 0.4546,
+      "step": 1013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3205774602783428e-05,
+      "loss": 0.4347,
+      "step": 1014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3193231609015762e-05,
+      "loss": 0.423,
+      "step": 1015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.318068301851463e-05,
+      "loss": 0.4277,
+      "step": 1016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3168128853273771e-05,
+      "loss": 0.4424,
+      "step": 1017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3155569135296705e-05,
+      "loss": 0.4511,
+      "step": 1018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314300388659667e-05,
+      "loss": 0.4275,
+      "step": 1019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3130433129196613e-05,
+      "loss": 0.4404,
+      "step": 1020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3117856885129125e-05,
+      "loss": 0.4303,
+      "step": 1021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310527517643642e-05,
+      "loss": 0.4537,
+      "step": 1022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3092688025170283e-05,
+      "loss": 0.4249,
+      "step": 1023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308009545339205e-05,
+      "loss": 0.441,
+      "step": 1024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306749748317254e-05,
+      "loss": 0.4365,
+      "step": 1025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3054894136592052e-05,
+      "loss": 0.4252,
+      "step": 1026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042285435740292e-05,
+      "loss": 0.4229,
+      "step": 1027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3029671402716366e-05,
+      "loss": 0.4297,
+      "step": 1028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301705205962871e-05,
+      "loss": 0.4264,
+      "step": 1029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300442742859508e-05,
+      "loss": 0.4419,
+      "step": 1030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2991797531742492e-05,
+      "loss": 0.4368,
+      "step": 1031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2979162391207193e-05,
+      "loss": 0.4274,
+      "step": 1032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2966522029134622e-05,
+      "loss": 0.4248,
+      "step": 1033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2953876467679373e-05,
+      "loss": 0.4178,
+      "step": 1034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2941225729005144e-05,
+      "loss": 0.4326,
+      "step": 1035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2928569835284714e-05,
+      "loss": 0.4195,
+      "step": 1036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2915908808699893e-05,
+      "loss": 0.4371,
+      "step": 1037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2903242671441495e-05,
+      "loss": 0.435,
+      "step": 1038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289057144570928e-05,
+      "loss": 0.434,
+      "step": 1039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2877895153711935e-05,
+      "loss": 0.4387,
+      "step": 1040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2865213817667024e-05,
+      "loss": 0.4396,
+      "step": 1041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2852527459800954e-05,
+      "loss": 0.4418,
+      "step": 1042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2839836102348926e-05,
+      "loss": 0.4293,
+      "step": 1043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2827139767554915e-05,
+      "loss": 0.4519,
+      "step": 1044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281443847767161e-05,
+      "loss": 0.44,
+      "step": 1045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2801732254960389e-05,
+      "loss": 0.4173,
+      "step": 1046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2789021121691273e-05,
+      "loss": 0.4319,
+      "step": 1047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2776305100142898e-05,
+      "loss": 0.4436,
+      "step": 1048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2763584212602453e-05,
+      "loss": 0.4252,
+      "step": 1049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2750858481365673e-05,
+      "loss": 0.4296,
+      "step": 1050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738127928736766e-05,
+      "loss": 0.4367,
+      "step": 1051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2725392577028403e-05,
+      "loss": 0.4356,
+      "step": 1052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2712652448561657e-05,
+      "loss": 0.4242,
+      "step": 1053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2699907565665983e-05,
+      "loss": 0.4278,
+      "step": 1054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2687157950679158e-05,
+      "loss": 0.4223,
+      "step": 1055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267440362594726e-05,
+      "loss": 0.4399,
+      "step": 1056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266164461382462e-05,
+      "loss": 0.4197,
+      "step": 1057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2648880936673787e-05,
+      "loss": 0.4378,
+      "step": 1058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2636112616865476e-05,
+      "loss": 0.4487,
+      "step": 1059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2623339676778557e-05,
+      "loss": 0.4454,
+      "step": 1060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2610562138799977e-05,
+      "loss": 0.4299,
+      "step": 1061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2597780025324764e-05,
+      "loss": 0.4302,
+      "step": 1062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2584993358755946e-05,
+      "loss": 0.4242,
+      "step": 1063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2572202161504543e-05,
+      "loss": 0.4176,
+      "step": 1064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2559406455989506e-05,
+      "loss": 0.4245,
+      "step": 1065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.25466062646377e-05,
+      "loss": 0.4433,
+      "step": 1066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2533801609883842e-05,
+      "loss": 0.4422,
+      "step": 1067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252099251417048e-05,
+      "loss": 0.4247,
+      "step": 1068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2508178999947935e-05,
+      "loss": 0.4326,
+      "step": 1069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2495361089674284e-05,
+      "loss": 0.4367,
+      "step": 1070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2482538805815301e-05,
+      "loss": 0.427,
+      "step": 1071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2469712170844431e-05,
+      "loss": 0.4197,
+      "step": 1072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2456881207242732e-05,
+      "loss": 0.4379,
+      "step": 1073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2444045937498874e-05,
+      "loss": 0.435,
+      "step": 1074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2431206384109045e-05,
+      "loss": 0.423,
+      "step": 1075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2418362569576965e-05,
+      "loss": 0.4224,
+      "step": 1076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2405514516413806e-05,
+      "loss": 0.4516,
+      "step": 1077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392662247138181e-05,
+      "loss": 0.4393,
+      "step": 1078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2379805784276081e-05,
+      "loss": 0.4237,
+      "step": 1079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236694515036086e-05,
+      "loss": 0.4544,
+      "step": 1080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2354080367933168e-05,
+      "loss": 0.4309,
+      "step": 1081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234121145954094e-05,
+      "loss": 0.453,
+      "step": 1082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2328338447739331e-05,
+      "loss": 0.4444,
+      "step": 1083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2315461355090698e-05,
+      "loss": 0.4317,
+      "step": 1084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2302580204164542e-05,
+      "loss": 0.4142,
+      "step": 1085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2289695017537483e-05,
+      "loss": 0.4302,
+      "step": 1086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2276805817793208e-05,
+      "loss": 0.4194,
+      "step": 1087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2263912627522449e-05,
+      "loss": 0.4219,
+      "step": 1088
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2251015469322915e-05,
+      "loss": 0.4366,
+      "step": 1089
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2238114365799285e-05,
+      "loss": 0.4152,
+      "step": 1090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.4397,
+      "step": 1091
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2212300413232963e-05,
+      "loss": 0.434,
+      "step": 1092
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219938760943403e-05,
+      "loss": 0.4279,
+      "step": 1093
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2186470950798446e-05,
+      "loss": 0.4347,
+      "step": 1094
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173550459965062e-05,
+      "loss": 0.4305,
+      "step": 1095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2160626159579447e-05,
+      "loss": 0.4393,
+      "step": 1096
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147698072293844e-05,
+      "loss": 0.4165,
+      "step": 1097
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2134766220767135e-05,
+      "loss": 0.4126,
+      "step": 1098
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21218306276648e-05,
+      "loss": 0.4269,
+      "step": 1099
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210889131565888e-05,
+      "loss": 0.4351,
+      "step": 1100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2095948307427925e-05,
+      "loss": 0.4265,
+      "step": 1101
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2083001625656974e-05,
+      "loss": 0.4347,
+      "step": 1102
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2070051293037493e-05,
+      "loss": 0.4205,
+      "step": 1103
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205709733226736e-05,
+      "loss": 0.4392,
+      "step": 1104
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2044139766050803e-05,
+      "loss": 0.4118,
+      "step": 1105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2031178617098372e-05,
+      "loss": 0.4295,
+      "step": 1106
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201821390812689e-05,
+      "loss": 0.4237,
+      "step": 1107
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2005245661859435e-05,
+      "loss": 0.423,
+      "step": 1108
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1992273901025268e-05,
+      "loss": 0.4356,
+      "step": 1109
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1979298648359823e-05,
+      "loss": 0.4272,
+      "step": 1110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966319926604642e-05,
+      "loss": 0.4342,
+      "step": 1111
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195333775850736e-05,
+      "loss": 0.4243,
+      "step": 1112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194035216682164e-05,
+      "loss": 0.4316,
+      "step": 1113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1927363174307155e-05,
+      "loss": 0.4258,
+      "step": 1114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1914370803729533e-05,
+      "loss": 0.441,
+      "step": 1115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1901375077860328e-05,
+      "loss": 0.4258,
+      "step": 1116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1888376019476966e-05,
+      "loss": 0.4269,
+      "step": 1117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1875373651362727e-05,
+      "loss": 0.4414,
+      "step": 1118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1862367996306675e-05,
+      "loss": 0.4327,
+      "step": 1119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184935907710365e-05,
+      "loss": 0.4131,
+      "step": 1120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1836346916554204e-05,
+      "loss": 0.4236,
+      "step": 1121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823331537464576e-05,
+      "loss": 0.4423,
+      "step": 1122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1810312962646643e-05,
+      "loss": 0.4378,
+      "step": 1123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797291214917882e-05,
+      "loss": 0.427,
+      "step": 1124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1784266317101334e-05,
+      "loss": 0.4241,
+      "step": 1125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1771238292025558e-05,
+      "loss": 0.4421,
+      "step": 1126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1758207162524597e-05,
+      "loss": 0.4313,
+      "step": 1127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1745172951437933e-05,
+      "loss": 0.4304,
+      "step": 1128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1732135681610453e-05,
+      "loss": 0.4193,
+      "step": 1129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1719095375892398e-05,
+      "loss": 0.4124,
+      "step": 1130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706052057139335e-05,
+      "loss": 0.4323,
+      "step": 1131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169300574821211e-05,
+      "loss": 0.4249,
+      "step": 1132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1679956471976814e-05,
+      "loss": 0.4172,
+      "step": 1133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1666904251304732e-05,
+      "loss": 0.4296,
+      "step": 1134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1653849109072315e-05,
+      "loss": 0.4251,
+      "step": 1135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164079106816113e-05,
+      "loss": 0.4214,
+      "step": 1136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162773015145783e-05,
+      "loss": 0.4407,
+      "step": 1137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1614666381854107e-05,
+      "loss": 0.4205,
+      "step": 1138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1601599782246646e-05,
+      "loss": 0.4258,
+      "step": 1139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1588530375537102e-05,
+      "loss": 0.4586,
+      "step": 1140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1575458184632045e-05,
+      "loss": 0.412,
+      "step": 1141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1562383232442927e-05,
+      "loss": 0.4226,
+      "step": 1142
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1549305541886033e-05,
+      "loss": 0.423,
+      "step": 1143
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153622513588246e-05,
+      "loss": 0.4378,
+      "step": 1144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152314203735805e-05,
+      "loss": 0.4136,
+      "step": 1145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151005626924338e-05,
+      "loss": 0.4355,
+      "step": 1146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1496967854473688e-05,
+      "loss": 0.4153,
+      "step": 1147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1483876815988867e-05,
+      "loss": 0.4352,
+      "step": 1148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1470783176733396e-05,
+      "loss": 0.4302,
+      "step": 1149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1457686959656322e-05,
+      "loss": 0.428,
+      "step": 1150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1444588187711205e-05,
+      "loss": 0.4408,
+      "step": 1151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1431486883856082e-05,
+      "loss": 0.4282,
+      "step": 1152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141838307105343e-05,
+      "loss": 0.44,
+      "step": 1153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1405276772270126e-05,
+      "loss": 0.4244,
+      "step": 1154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1392168010477398e-05,
+      "loss": 0.4119,
+      "step": 1155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379056808650795e-05,
+      "loss": 0.4237,
+      "step": 1156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136594318977014e-05,
+      "loss": 0.4262,
+      "step": 1157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1352827176819496e-05,
+      "loss": 0.4188,
+      "step": 1158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1339708792787119e-05,
+      "loss": 0.4341,
+      "step": 1159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.132658806066542e-05,
+      "loss": 0.4199,
+      "step": 1160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1313465003450931e-05,
+      "loss": 0.4255,
+      "step": 1161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.130033964414425e-05,
+      "loss": 0.422,
+      "step": 1162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287212005750023e-05,
+      "loss": 0.4515,
+      "step": 1163
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1274082111276877e-05,
+      "loss": 0.4333,
+      "step": 1164
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1260949983737399e-05,
+      "loss": 0.4189,
+      "step": 1165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1247815646148088e-05,
+      "loss": 0.4409,
+      "step": 1166
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1234679121529325e-05,
+      "loss": 0.4207,
+      "step": 1167
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1221540432905309e-05,
+      "loss": 0.4326,
+      "step": 1168
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1208399603304048e-05,
+      "loss": 0.434,
+      "step": 1169
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119525665575729e-05,
+      "loss": 0.4337,
+      "step": 1170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1182111613300502e-05,
+      "loss": 0.4181,
+      "step": 1171
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1168964498972819e-05,
+      "loss": 0.4342,
+      "step": 1172
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115581533581701e-05,
+      "loss": 0.4212,
+      "step": 1173
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1142664146879432e-05,
+      "loss": 0.4347,
+      "step": 1174
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1129510955209996e-05,
+      "loss": 0.4141,
+      "step": 1175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1116355783862122e-05,
+      "loss": 0.418,
+      "step": 1176
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11031986558927e-05,
+      "loss": 0.4268,
+      "step": 1177
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1090039594362046e-05,
+      "loss": 0.4341,
+      "step": 1178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1076878622333869e-05,
+      "loss": 0.4239,
+      "step": 1179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1063715762875225e-05,
+      "loss": 0.4124,
+      "step": 1180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050551039056479e-05,
+      "loss": 0.4154,
+      "step": 1181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103738447395126e-05,
+      "loss": 0.4183,
+      "step": 1182
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1024216090636433e-05,
+      "loss": 0.4308,
+      "step": 1183
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011045912192036e-05,
+      "loss": 0.4351,
+      "step": 1184
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997873961701267e-05,
+      "loss": 0.4182,
+      "step": 1185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0984700262250418e-05,
+      "loss": 0.4176,
+      "step": 1186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.097152483692886e-05,
+      "loss": 0.4285,
+      "step": 1187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958347708828976e-05,
+      "loss": 0.4093,
+      "step": 1188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094516890104614e-05,
+      "loss": 0.4378,
+      "step": 1189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0931988436678666e-05,
+      "loss": 0.4324,
+      "step": 1190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091880633882778e-05,
+      "loss": 0.4189,
+      "step": 1191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0905622630597558e-05,
+      "loss": 0.4109,
+      "step": 1192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892437335094911e-05,
+      "loss": 0.4096,
+      "step": 1193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0879250475429523e-05,
+      "loss": 0.4343,
+      "step": 1194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0866062074713825e-05,
+      "loss": 0.4127,
+      "step": 1195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852872156062947e-05,
+      "loss": 0.411,
+      "step": 1196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0839680742594679e-05,
+      "loss": 0.4335,
+      "step": 1197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0826487857429428e-05,
+      "loss": 0.4291,
+      "step": 1198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0813293523690191e-05,
+      "loss": 0.4262,
+      "step": 1199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080009776450249e-05,
+      "loss": 0.4332,
+      "step": 1200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078690060299436e-05,
+      "loss": 0.4186,
+      "step": 1201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0773702062296273e-05,
+      "loss": 0.4224,
+      "step": 1202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760502165541145e-05,
+      "loss": 0.4235,
+      "step": 1203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0747300935864245e-05,
+      "loss": 0.4099,
+      "step": 1204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0734098396403192e-05,
+      "loss": 0.4205,
+      "step": 1205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0720894570297898e-05,
+      "loss": 0.4313,
+      "step": 1206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707689480690526e-05,
+      "loss": 0.4373,
+      "step": 1207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0694483150725458e-05,
+      "loss": 0.4249,
+      "step": 1208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0681275603549253e-05,
+      "loss": 0.4355,
+      "step": 1209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0668066862310588e-05,
+      "loss": 0.4356,
+      "step": 1210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0654856950160252e-05,
+      "loss": 0.4316,
+      "step": 1211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0641645890251075e-05,
+      "loss": 0.4263,
+      "step": 1212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0628433705737902e-05,
+      "loss": 0.4293,
+      "step": 1213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0615220419777548e-05,
+      "loss": 0.4035,
+      "step": 1214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060200605552876e-05,
+      "loss": 0.3971,
+      "step": 1215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0588790636152169e-05,
+      "loss": 0.4243,
+      "step": 1216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057557418481027e-05,
+      "loss": 0.4562,
+      "step": 1217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0562356724667346e-05,
+      "loss": 0.414,
+      "step": 1218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0549138278889468e-05,
+      "loss": 0.4187,
+      "step": 1219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0535918870644419e-05,
+      "loss": 0.4325,
+      "step": 1220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0522698523101683e-05,
+      "loss": 0.4177,
+      "step": 1221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0509477259432372e-05,
+      "loss": 0.4284,
+      "step": 1222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0496255102809224e-05,
+      "loss": 0.4115,
+      "step": 1223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0483032076406528e-05,
+      "loss": 0.4214,
+      "step": 1224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0469808203400103e-05,
+      "loss": 0.4073,
+      "step": 1225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0456583506967248e-05,
+      "loss": 0.4361,
+      "step": 1226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0443358010286714e-05,
+      "loss": 0.4264,
+      "step": 1227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0430131736538645e-05,
+      "loss": 0.4374,
+      "step": 1228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041690470890455e-05,
+      "loss": 0.4109,
+      "step": 1229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040367695056726e-05,
+      "loss": 0.4247,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0390448484710887e-05,
+      "loss": 0.4154,
+      "step": 1231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0377219334520783e-05,
+      "loss": 0.4328,
+      "step": 1232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0363989523183496e-05,
+      "loss": 0.4183,
+      "step": 1233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.035075907388674e-05,
+      "loss": 0.3974,
+      "step": 1234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0337528009819343e-05,
+      "loss": 0.4175,
+      "step": 1235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0324296354171209e-05,
+      "loss": 0.4126,
+      "step": 1236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0311064130133278e-05,
+      "loss": 0.4291,
+      "step": 1237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0297831360897493e-05,
+      "loss": 0.4282,
+      "step": 1238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0284598069656746e-05,
+      "loss": 0.4286,
+      "step": 1239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0271364279604844e-05,
+      "loss": 0.4305,
+      "step": 1240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0258130013936473e-05,
+      "loss": 0.4274,
+      "step": 1241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0244895295847148e-05,
+      "loss": 0.4366,
+      "step": 1242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0231660148533183e-05,
+      "loss": 0.4145,
+      "step": 1243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0218424595191632e-05,
+      "loss": 0.4207,
+      "step": 1244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0205188659020275e-05,
+      "loss": 0.4178,
+      "step": 1245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0191952363217557e-05,
+      "loss": 0.4149,
+      "step": 1246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.017871573098255e-05,
+      "loss": 0.429,
+      "step": 1247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0165478785514919e-05,
+      "loss": 0.4311,
+      "step": 1248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015224155001488e-05,
+      "loss": 0.4147,
+      "step": 1249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0139004047683152e-05,
+      "loss": 0.4189,
+      "step": 1250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012576630172093e-05,
+      "loss": 0.4231,
+      "step": 1251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0112528335329824e-05,
+      "loss": 0.4229,
+      "step": 1252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099290171711841e-05,
+      "loss": 0.4265,
+      "step": 1253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0086051834069327e-05,
+      "loss": 0.4091,
+      "step": 1254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0072813345604942e-05,
+      "loss": 0.427,
+      "step": 1255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059574729521595e-05,
+      "loss": 0.4186,
+      "step": 1256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0046336009022435e-05,
+      "loss": 0.418,
+      "step": 1257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033097207310784e-05,
+      "loss": 0.4319,
+      "step": 1258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.001985834759011e-05,
+      "loss": 0.4164,
+      "step": 1259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000661945306398e-05,
+      "loss": 0.4231,
+      "step": 1260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993380546936023e-06,
+      "loss": 0.4214,
+      "step": 1261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.980141652409896e-06,
+      "loss": 0.4226,
+      "step": 1262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.966902792689219e-06,
+      "loss": 0.4394,
+      "step": 1263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.953663990977568e-06,
+      "loss": 0.4261,
+      "step": 1264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940425270478407e-06,
+      "loss": 0.4201,
+      "step": 1265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.927186654395063e-06,
+      "loss": 0.4174,
+      "step": 1266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.913948165930676e-06,
+      "loss": 0.4153,
+      "step": 1267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900709828288164e-06,
+      "loss": 0.4119,
+      "step": 1268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88747166467018e-06,
+      "loss": 0.4186,
+      "step": 1269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.874233698279076e-06,
+      "loss": 0.414,
+      "step": 1270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.860995952316851e-06,
+      "loss": 0.4204,
+      "step": 1271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847758449985124e-06,
+      "loss": 0.4111,
+      "step": 1272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.834521214485084e-06,
+      "loss": 0.41,
+      "step": 1273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.821284269017456e-06,
+      "loss": 0.3994,
+      "step": 1274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.808047636782448e-06,
+      "loss": 0.4103,
+      "step": 1275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794811340979725e-06,
+      "loss": 0.4139,
+      "step": 1276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.781575404808372e-06,
+      "loss": 0.4275,
+      "step": 1277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.768339851466818e-06,
+      "loss": 0.4262,
+      "step": 1278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.755104704152855e-06,
+      "loss": 0.4094,
+      "step": 1279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.741869986063527e-06,
+      "loss": 0.419,
+      "step": 1280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.728635720395158e-06,
+      "loss": 0.4284,
+      "step": 1281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715401930343255e-06,
+      "loss": 0.4376,
+      "step": 1282
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.702168639102509e-06,
+      "loss": 0.4168,
+      "step": 1283
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688935869866721e-06,
+      "loss": 0.4153,
+      "step": 1284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.675703645828795e-06,
+      "loss": 0.4096,
+      "step": 1285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.662471990180658e-06,
+      "loss": 0.4087,
+      "step": 1286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649240926113263e-06,
+      "loss": 0.4137,
+      "step": 1287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.636010476816504e-06,
+      "loss": 0.4143,
+      "step": 1288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.622780665479223e-06,
+      "loss": 0.4324,
+      "step": 1289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609551515289117e-06,
+      "loss": 0.4056,
+      "step": 1290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596323049432746e-06,
+      "loss": 0.4104,
+      "step": 1291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.583095291095454e-06,
+      "loss": 0.4042,
+      "step": 1292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.569868263461362e-06,
+      "loss": 0.4122,
+      "step": 1293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.556641989713287e-06,
+      "loss": 0.4211,
+      "step": 1294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.543416493032757e-06,
+      "loss": 0.4035,
+      "step": 1295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5301917965999e-06,
+      "loss": 0.4313,
+      "step": 1296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.516967923593479e-06,
+      "loss": 0.4273,
+      "step": 1297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.503744897190778e-06,
+      "loss": 0.4186,
+      "step": 1298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.490522740567633e-06,
+      "loss": 0.4293,
+      "step": 1299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477301476898322e-06,
+      "loss": 0.4241,
+      "step": 1300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.464081129355586e-06,
+      "loss": 0.4232,
+      "step": 1301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.450861721110535e-06,
+      "loss": 0.4153,
+      "step": 1302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.437643275332654e-06,
+      "loss": 0.4124,
+      "step": 1303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.424425815189734e-06,
+      "loss": 0.4154,
+      "step": 1304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41120936384783e-06,
+      "loss": 0.4323,
+      "step": 1305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397993944471245e-06,
+      "loss": 0.4325,
+      "step": 1306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.384779580222454e-06,
+      "loss": 0.4335,
+      "step": 1307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.371566294262101e-06,
+      "loss": 0.4278,
+      "step": 1308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358354109748927e-06,
+      "loss": 0.4263,
+      "step": 1309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34514304983975e-06,
+      "loss": 0.4073,
+      "step": 1310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.331933137689413e-06,
+      "loss": 0.4132,
+      "step": 1311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31872439645075e-06,
+      "loss": 0.4143,
+      "step": 1312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.305516849274542e-06,
+      "loss": 0.4087,
+      "step": 1313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.292310519309476e-06,
+      "loss": 0.4301,
+      "step": 1314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.279105429702104e-06,
+      "loss": 0.4138,
+      "step": 1315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.265901603596811e-06,
+      "loss": 0.4028,
+      "step": 1316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252699064135759e-06,
+      "loss": 0.4271,
+      "step": 1317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23949783445886e-06,
+      "loss": 0.4088,
+      "step": 1318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.226297937703728e-06,
+      "loss": 0.4078,
+      "step": 1319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.213099397005647e-06,
+      "loss": 0.4329,
+      "step": 1320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.199902235497513e-06,
+      "loss": 0.4233,
+      "step": 1321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.186706476309812e-06,
+      "loss": 0.4451,
+      "step": 1322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173512142570574e-06,
+      "loss": 0.4071,
+      "step": 1323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.160319257405327e-06,
+      "loss": 0.4169,
+      "step": 1324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147127843937055e-06,
+      "loss": 0.4267,
+      "step": 1325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13393792528618e-06,
+      "loss": 0.4158,
+      "step": 1326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12074952457048e-06,
+      "loss": 0.4191,
+      "step": 1327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107562664905094e-06,
+      "loss": 0.4246,
+      "step": 1328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.094377369402445e-06,
+      "loss": 0.4175,
+      "step": 1329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081193661172226e-06,
+      "loss": 0.4131,
+      "step": 1330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068011563321335e-06,
+      "loss": 0.4339,
+      "step": 1331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.054831098953863e-06,
+      "loss": 0.4205,
+      "step": 1332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041652291171029e-06,
+      "loss": 0.4376,
+      "step": 1333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.028475163071142e-06,
+      "loss": 0.4495,
+      "step": 1334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015299737749584e-06,
+      "loss": 0.4064,
+      "step": 1335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002126038298736e-06,
+      "loss": 0.4085,
+      "step": 1336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988954087807968e-06,
+      "loss": 0.4072,
+      "step": 1337
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97578390936357e-06,
+      "loss": 0.4131,
+      "step": 1338
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962615526048742e-06,
+      "loss": 0.4211,
+      "step": 1339
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.949448960943524e-06,
+      "loss": 0.4206,
+      "step": 1340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.936284237124779e-06,
+      "loss": 0.409,
+      "step": 1341
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923121377666135e-06,
+      "loss": 0.4147,
+      "step": 1342
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.909960405637958e-06,
+      "loss": 0.406,
+      "step": 1343
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.896801344107303e-06,
+      "loss": 0.4365,
+      "step": 1344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883644216137881e-06,
+      "loss": 0.4117,
+      "step": 1345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.870489044790007e-06,
+      "loss": 0.4186,
+      "step": 1346
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.857335853120573e-06,
+      "loss": 0.4137,
+      "step": 1347
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.844184664182993e-06,
+      "loss": 0.4253,
+      "step": 1348
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.831035501027186e-06,
+      "loss": 0.4387,
+      "step": 1349
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8178883866995e-06,
+      "loss": 0.4008,
+      "step": 1350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804743344242716e-06,
+      "loss": 0.4242,
+      "step": 1351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.791600396695954e-06,
+      "loss": 0.4307,
+      "step": 1352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.778459567094696e-06,
+      "loss": 0.4335,
+      "step": 1353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.765320878470679e-06,
+      "loss": 0.42,
+      "step": 1354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752184353851917e-06,
+      "loss": 0.4168,
+      "step": 1355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.739050016262605e-06,
+      "loss": 0.4075,
+      "step": 1356
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72591788872313e-06,
+      "loss": 0.4273,
+      "step": 1357
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712787994249979e-06,
+      "loss": 0.4092,
+      "step": 1358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.699660355855747e-06,
+      "loss": 0.422,
+      "step": 1359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68653499654907e-06,
+      "loss": 0.417,
+      "step": 1360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67341193933458e-06,
+      "loss": 0.4002,
+      "step": 1361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.660291207212883e-06,
+      "loss": 0.4176,
+      "step": 1362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.647172823180505e-06,
+      "loss": 0.4289,
+      "step": 1363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634056810229863e-06,
+      "loss": 0.4045,
+      "step": 1364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.620943191349207e-06,
+      "loss": 0.4171,
+      "step": 1365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607831989522604e-06,
+      "loss": 0.4238,
+      "step": 1366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.594723227729875e-06,
+      "loss": 0.4161,
+      "step": 1367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.581616928946571e-06,
+      "loss": 0.4097,
+      "step": 1368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56851311614392e-06,
+      "loss": 0.4242,
+      "step": 1369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.555411812288799e-06,
+      "loss": 0.4151,
+      "step": 1370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54231304034368e-06,
+      "loss": 0.4236,
+      "step": 1371
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.529216823266605e-06,
+      "loss": 0.4306,
+      "step": 1372
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.516123184011136e-06,
+      "loss": 0.4204,
+      "step": 1373
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.503032145526314e-06,
+      "loss": 0.4206,
+      "step": 1374
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489943730756622e-06,
+      "loss": 0.3978,
+      "step": 1375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476857962641951e-06,
+      "loss": 0.4208,
+      "step": 1376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.463774864117542e-06,
+      "loss": 0.4097,
+      "step": 1377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.450694458113969e-06,
+      "loss": 0.4159,
+      "step": 1378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.437616767557078e-06,
+      "loss": 0.3919,
+      "step": 1379
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.424541815367958e-06,
+      "loss": 0.4359,
+      "step": 1380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4114696244629e-06,
+      "loss": 0.4184,
+      "step": 1381
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398400217753357e-06,
+      "loss": 0.4191,
+      "step": 1382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385333618145896e-06,
+      "loss": 0.4316,
+      "step": 1383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.372269848542173e-06,
+      "loss": 0.4111,
+      "step": 1384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359208931838872e-06,
+      "loss": 0.4193,
+      "step": 1385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.346150890927688e-06,
+      "loss": 0.4133,
+      "step": 1386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.333095748695271e-06,
+      "loss": 0.4122,
+      "step": 1387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.320043528023188e-06,
+      "loss": 0.4125,
+      "step": 1388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.306994251787891e-06,
+      "loss": 0.4187,
+      "step": 1389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.293947942860667e-06,
+      "loss": 0.4265,
+      "step": 1390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.280904624107606e-06,
+      "loss": 0.4242,
+      "step": 1391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.267864318389548e-06,
+      "loss": 0.4146,
+      "step": 1392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.254827048562069e-06,
+      "loss": 0.4041,
+      "step": 1393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.241792837475405e-06,
+      "loss": 0.4037,
+      "step": 1394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.228761707974445e-06,
+      "loss": 0.4097,
+      "step": 1395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215733682898669e-06,
+      "loss": 0.4065,
+      "step": 1396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202708785082122e-06,
+      "loss": 0.4075,
+      "step": 1397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18968703735336e-06,
+      "loss": 0.4056,
+      "step": 1398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176668462535427e-06,
+      "loss": 0.4236,
+      "step": 1399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.163653083445799e-06,
+      "loss": 0.4129,
+      "step": 1400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150640922896356e-06,
+      "loss": 0.4255,
+      "step": 1401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.137632003693329e-06,
+      "loss": 0.4084,
+      "step": 1402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.124626348637278e-06,
+      "loss": 0.4058,
+      "step": 1403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.111623980523036e-06,
+      "loss": 0.4114,
+      "step": 1404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.098624922139677e-06,
+      "loss": 0.3907,
+      "step": 1405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085629196270468e-06,
+      "loss": 0.4227,
+      "step": 1406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07263682569285e-06,
+      "loss": 0.4085,
+      "step": 1407
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.059647833178363e-06,
+      "loss": 0.4152,
+      "step": 1408
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.046662241492645e-06,
+      "loss": 0.4195,
+      "step": 1409
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03368007339536e-06,
+      "loss": 0.4241,
+      "step": 1410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.020701351640182e-06,
+      "loss": 0.4196,
+      "step": 1411
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007726098974733e-06,
+      "loss": 0.4021,
+      "step": 1412
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.99475433814057e-06,
+      "loss": 0.4238,
+      "step": 1413
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981786091873112e-06,
+      "loss": 0.4004,
+      "step": 1414
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96882138290163e-06,
+      "loss": 0.4193,
+      "step": 1415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9558602339492e-06,
+      "loss": 0.4195,
+      "step": 1416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.942902667732639e-06,
+      "loss": 0.4144,
+      "step": 1417
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.929948706962508e-06,
+      "loss": 0.4192,
+      "step": 1418
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.916998374343028e-06,
+      "loss": 0.4195,
+      "step": 1419
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.904051692572078e-06,
+      "loss": 0.4165,
+      "step": 1420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.891108684341122e-06,
+      "loss": 0.4193,
+      "step": 1421
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.878169372335201e-06,
+      "loss": 0.4096,
+      "step": 1422
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865233779232867e-06,
+      "loss": 0.409,
+      "step": 1423
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852301927706158e-06,
+      "loss": 0.4133,
+      "step": 1424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.839373840420555e-06,
+      "loss": 0.4278,
+      "step": 1425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.82644954003494e-06,
+      "loss": 0.4361,
+      "step": 1426
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.813529049201555e-06,
+      "loss": 0.394,
+      "step": 1427
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.800612390565974e-06,
+      "loss": 0.4173,
+      "step": 1428
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.787699586767042e-06,
+      "loss": 0.4313,
+      "step": 1429
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.4071,
+      "step": 1430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.761885634200717e-06,
+      "loss": 0.4153,
+      "step": 1431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748984530677088e-06,
+      "loss": 0.4008,
+      "step": 1432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.736087372477554e-06,
+      "loss": 0.4285,
+      "step": 1433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.723194182206793e-06,
+      "loss": 0.4185,
+      "step": 1434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.710304982462518e-06,
+      "loss": 0.4032,
+      "step": 1435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.697419795835461e-06,
+      "loss": 0.4062,
+      "step": 1436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.684538644909303e-06,
+      "loss": 0.4161,
+      "step": 1437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67166155226067e-06,
+      "loss": 0.4294,
+      "step": 1438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.658788540459063e-06,
+      "loss": 0.4211,
+      "step": 1439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.645919632066833e-06,
+      "loss": 0.3978,
+      "step": 1440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.633054849639143e-06,
+      "loss": 0.4064,
+      "step": 1441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.620194215723919e-06,
+      "loss": 0.4031,
+      "step": 1442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607337752861821e-06,
+      "loss": 0.4057,
+      "step": 1443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.594485483586193e-06,
+      "loss": 0.4187,
+      "step": 1444
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.581637430423038e-06,
+      "loss": 0.4088,
+      "step": 1445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.568793615890955e-06,
+      "loss": 0.4098,
+      "step": 1446
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.55595406250113e-06,
+      "loss": 0.4023,
+      "step": 1447
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.543118792757267e-06,
+      "loss": 0.4169,
+      "step": 1448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.530287829155574e-06,
+      "loss": 0.4066,
+      "step": 1449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.517461194184699e-06,
+      "loss": 0.4084,
+      "step": 1450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.504638910325716e-06,
+      "loss": 0.4192,
+      "step": 1451
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4918210000520645e-06,
+      "loss": 0.4287,
+      "step": 1452
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.479007485829523e-06,
+      "loss": 0.414,
+      "step": 1453
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.466198390116157e-06,
+      "loss": 0.4371,
+      "step": 1454
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.453393735362302e-06,
+      "loss": 0.4204,
+      "step": 1455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4405935440104946e-06,
+      "loss": 0.4217,
+      "step": 1456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.427797838495463e-06,
+      "loss": 0.4106,
+      "step": 1457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.415006641244057e-06,
+      "loss": 0.411,
+      "step": 1458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40221997467524e-06,
+      "loss": 0.4233,
+      "step": 1459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.389437861200024e-06,
+      "loss": 0.4133,
+      "step": 1460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.376660323221449e-06,
+      "loss": 0.4268,
+      "step": 1461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.363887383134527e-06,
+      "loss": 0.4151,
+      "step": 1462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3511190633262195e-06,
+      "loss": 0.4177,
+      "step": 1463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.338355386175382e-06,
+      "loss": 0.4057,
+      "step": 1464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3255963740527435e-06,
+      "loss": 0.4078,
+      "step": 1465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312842049320845e-06,
+      "loss": 0.4073,
+      "step": 1466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.300092434334021e-06,
+      "loss": 0.4096,
+      "step": 1467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.287347551438344e-06,
+      "loss": 0.4238,
+      "step": 1468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2746074229716e-06,
+      "loss": 0.415,
+      "step": 1469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2618720712632366e-06,
+      "loss": 0.3879,
+      "step": 1470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2491415186343275e-06,
+      "loss": 0.3827,
+      "step": 1471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236415787397549e-06,
+      "loss": 0.4071,
+      "step": 1472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2236948998571035e-06,
+      "loss": 0.4223,
+      "step": 1473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21097887830873e-06,
+      "loss": 0.3993,
+      "step": 1474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.198267745039612e-06,
+      "loss": 0.4097,
+      "step": 1475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.185561522328395e-06,
+      "loss": 0.4201,
+      "step": 1476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.172860232445085e-06,
+      "loss": 0.4112,
+      "step": 1477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1601638976510756e-06,
+      "loss": 0.4288,
+      "step": 1478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1474725401990465e-06,
+      "loss": 0.415,
+      "step": 1479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.134786182332978e-06,
+      "loss": 0.3968,
+      "step": 1480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.122104846288065e-06,
+      "loss": 0.4046,
+      "step": 1481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.109428554290725e-06,
+      "loss": 0.4157,
+      "step": 1482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.096757328558507e-06,
+      "loss": 0.4148,
+      "step": 1483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.084091191300111e-06,
+      "loss": 0.4127,
+      "step": 1484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.071430164715288e-06,
+      "loss": 0.3963,
+      "step": 1485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0587742709948615e-06,
+      "loss": 0.4201,
+      "step": 1486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.04612353232063e-06,
+      "loss": 0.4158,
+      "step": 1487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033477970865381e-06,
+      "loss": 0.4186,
+      "step": 1488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.02083760879281e-06,
+      "loss": 0.4146,
+      "step": 1489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.008202468257514e-06,
+      "loss": 0.4164,
+      "step": 1490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9955725714049225e-06,
+      "loss": 0.4087,
+      "step": 1491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.982947940371294e-06,
+      "loss": 0.4032,
+      "step": 1492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.970328597283638e-06,
+      "loss": 0.4238,
+      "step": 1493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957714564259712e-06,
+      "loss": 0.4148,
+      "step": 1494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.94510586340795e-06,
+      "loss": 0.4059,
+      "step": 1495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.932502516827462e-06,
+      "loss": 0.4084,
+      "step": 1496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.919904546607954e-06,
+      "loss": 0.434,
+      "step": 1497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.907311974829716e-06,
+      "loss": 0.4174,
+      "step": 1498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894724823563584e-06,
+      "loss": 0.4221,
+      "step": 1499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.882143114870876e-06,
+      "loss": 0.4289,
+      "step": 1500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.869566870803389e-06,
+      "loss": 0.4079,
+      "step": 1501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85699611340333e-06,
+      "loss": 0.4178,
+      "step": 1502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.844430864703298e-06,
+      "loss": 0.4327,
+      "step": 1503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.831871146726228e-06,
+      "loss": 0.3989,
+      "step": 1504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8193169814853725e-06,
+      "loss": 0.4197,
+      "step": 1505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8067683909842376e-06,
+      "loss": 0.4042,
+      "step": 1506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.794225397216575e-06,
+      "loss": 0.4153,
+      "step": 1507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.781688022166312e-06,
+      "loss": 0.4112,
+      "step": 1508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.769156287807539e-06,
+      "loss": 0.405,
+      "step": 1509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.756630216104454e-06,
+      "loss": 0.4172,
+      "step": 1510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744109829011333e-06,
+      "loss": 0.3943,
+      "step": 1511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.731595148472485e-06,
+      "loss": 0.4099,
+      "step": 1512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719086196422224e-06,
+      "loss": 0.4065,
+      "step": 1513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.706582994784815e-06,
+      "loss": 0.4055,
+      "step": 1514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.694085565474453e-06,
+      "loss": 0.4097,
+      "step": 1515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.681593930395209e-06,
+      "loss": 0.4257,
+      "step": 1516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.669108111441004e-06,
+      "loss": 0.4,
+      "step": 1517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6566281304955584e-06,
+      "loss": 0.4051,
+      "step": 1518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.644154009432369e-06,
+      "loss": 0.4326,
+      "step": 1519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.631685770114655e-06,
+      "loss": 0.39,
+      "step": 1520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.61922343439533e-06,
+      "loss": 0.3985,
+      "step": 1521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606767024116958e-06,
+      "loss": 0.4103,
+      "step": 1522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5943165611117244e-06,
+      "loss": 0.3847,
+      "step": 1523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.581872067201378e-06,
+      "loss": 0.4006,
+      "step": 1524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569433564197222e-06,
+      "loss": 0.4018,
+      "step": 1525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5570010739000445e-06,
+      "loss": 0.3956,
+      "step": 1526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5445746181001015e-06,
+      "loss": 0.3915,
+      "step": 1527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532154218577081e-06,
+      "loss": 0.4065,
+      "step": 1528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.519739897100034e-06,
+      "loss": 0.4132,
+      "step": 1529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507331675427388e-06,
+      "loss": 0.423,
+      "step": 1530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.494929575306848e-06,
+      "loss": 0.413,
+      "step": 1531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.482533618475422e-06,
+      "loss": 0.4222,
+      "step": 1532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.470143826659317e-06,
+      "loss": 0.4075,
+      "step": 1533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.457760221573969e-06,
+      "loss": 0.4232,
+      "step": 1534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.445382824923938e-06,
+      "loss": 0.4243,
+      "step": 1535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4330116584029325e-06,
+      "loss": 0.4166,
+      "step": 1536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.420646743693715e-06,
+      "loss": 0.397,
+      "step": 1537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.408288102468114e-06,
+      "loss": 0.4029,
+      "step": 1538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39593575638694e-06,
+      "loss": 0.3975,
+      "step": 1539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.383589727099991e-06,
+      "loss": 0.4021,
+      "step": 1540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.371250036245977e-06,
+      "loss": 0.4092,
+      "step": 1541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.358916705452514e-06,
+      "loss": 0.4098,
+      "step": 1542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34658975633605e-06,
+      "loss": 0.3948,
+      "step": 1543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.334269210501876e-06,
+      "loss": 0.4067,
+      "step": 1544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.321955089544029e-06,
+      "loss": 0.4171,
+      "step": 1545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.309647415045315e-06,
+      "loss": 0.3999,
+      "step": 1546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.297346208577213e-06,
+      "loss": 0.405,
+      "step": 1547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285051491699896e-06,
+      "loss": 0.4115,
+      "step": 1548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.272763285962129e-06,
+      "loss": 0.4186,
+      "step": 1549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.260481612901299e-06,
+      "loss": 0.4175,
+      "step": 1550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.248206494043313e-06,
+      "loss": 0.403,
+      "step": 1551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.235937950902615e-06,
+      "loss": 0.4046,
+      "step": 1552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.223676004982104e-06,
+      "loss": 0.428,
+      "step": 1553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.211420677773131e-06,
+      "loss": 0.3938,
+      "step": 1554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.199171990755442e-06,
+      "loss": 0.4051,
+      "step": 1555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186929965397139e-06,
+      "loss": 0.4222,
+      "step": 1556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.174694623154658e-06,
+      "loss": 0.4049,
+      "step": 1557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.16246598547271e-06,
+      "loss": 0.3995,
+      "step": 1558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.150244073784266e-06,
+      "loss": 0.4187,
+      "step": 1559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.138028909510503e-06,
+      "loss": 0.4096,
+      "step": 1560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.125820514060772e-06,
+      "loss": 0.4007,
+      "step": 1561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.113618908832561e-06,
+      "loss": 0.4044,
+      "step": 1562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1014241152114584e-06,
+      "loss": 0.398,
+      "step": 1563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08923615457111e-06,
+      "loss": 0.4172,
+      "step": 1564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.077055048273193e-06,
+      "loss": 0.4226,
+      "step": 1565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.064880817667359e-06,
+      "loss": 0.4002,
+      "step": 1566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.052713484091223e-06,
+      "loss": 0.4172,
+      "step": 1567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040553068870298e-06,
+      "loss": 0.4224,
+      "step": 1568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.028399593317984e-06,
+      "loss": 0.4059,
+      "step": 1569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.016253078735508e-06,
+      "loss": 0.4261,
+      "step": 1570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0041135464119024e-06,
+      "loss": 0.4379,
+      "step": 1571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9919810176239554e-06,
+      "loss": 0.4087,
+      "step": 1572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979855513636192e-06,
+      "loss": 0.4062,
+      "step": 1573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96773705570081e-06,
+      "loss": 0.4095,
+      "step": 1574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9556256650576715e-06,
+      "loss": 0.4153,
+      "step": 1575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.943521362934241e-06,
+      "loss": 0.4072,
+      "step": 1576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.931424170545568e-06,
+      "loss": 0.4147,
+      "step": 1577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919334109094233e-06,
+      "loss": 0.4247,
+      "step": 1578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.907251199770323e-06,
+      "loss": 0.4161,
+      "step": 1579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8951754637513855e-06,
+      "loss": 0.4172,
+      "step": 1580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883106922202405e-06,
+      "loss": 0.4158,
+      "step": 1581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.871045596275741e-06,
+      "loss": 0.397,
+      "step": 1582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858991507111122e-06,
+      "loss": 0.3935,
+      "step": 1583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.846944675835584e-06,
+      "loss": 0.4081,
+      "step": 1584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.834905123563441e-06,
+      "loss": 0.3893,
+      "step": 1585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.822872871396255e-06,
+      "loss": 0.4234,
+      "step": 1586
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.810847940422786e-06,
+      "loss": 0.4211,
+      "step": 1587
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.798830351718975e-06,
+      "loss": 0.3966,
+      "step": 1588
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.786820126347876e-06,
+      "loss": 0.4144,
+      "step": 1589
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.77481728535965e-06,
+      "loss": 0.4316,
+      "step": 1590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.762821849791516e-06,
+      "loss": 0.3919,
+      "step": 1591
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.750833840667711e-06,
+      "loss": 0.3963,
+      "step": 1592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7388532789994476e-06,
+      "loss": 0.3997,
+      "step": 1593
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.726880185784898e-06,
+      "loss": 0.4184,
+      "step": 1594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.714914582009139e-06,
+      "loss": 0.402,
+      "step": 1595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7029564886441245e-06,
+      "loss": 0.406,
+      "step": 1596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6910059266486315e-06,
+      "loss": 0.4076,
+      "step": 1597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679062916968256e-06,
+      "loss": 0.435,
+      "step": 1598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6671274805353435e-06,
+      "loss": 0.4078,
+      "step": 1599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.655199638268978e-06,
+      "loss": 0.3984,
+      "step": 1600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.643279411074913e-06,
+      "loss": 0.4304,
+      "step": 1601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.631366819845579e-06,
+      "loss": 0.4138,
+      "step": 1602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.619461885460006e-06,
+      "loss": 0.41,
+      "step": 1603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607564628783817e-06,
+      "loss": 0.422,
+      "step": 1604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595675070669162e-06,
+      "loss": 0.4159,
+      "step": 1605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.583793231954713e-06,
+      "loss": 0.409,
+      "step": 1606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571919133465605e-06,
+      "loss": 0.4085,
+      "step": 1607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.560052796013413e-06,
+      "loss": 0.4052,
+      "step": 1608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.548194240396099e-06,
+      "loss": 0.41,
+      "step": 1609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.536343487397991e-06,
+      "loss": 0.4324,
+      "step": 1610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524500557789745e-06,
+      "loss": 0.4177,
+      "step": 1611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.512665472328302e-06,
+      "loss": 0.3979,
+      "step": 1612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.500838251756857e-06,
+      "loss": 0.3967,
+      "step": 1613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.489018916804813e-06,
+      "loss": 0.4031,
+      "step": 1614
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477207488187759e-06,
+      "loss": 0.404,
+      "step": 1615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465403986607426e-06,
+      "loss": 0.4025,
+      "step": 1616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4536084327516535e-06,
+      "loss": 0.4042,
+      "step": 1617
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.441820847294339e-06,
+      "loss": 0.4108,
+      "step": 1618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430041250895428e-06,
+      "loss": 0.398,
+      "step": 1619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.418269664200857e-06,
+      "loss": 0.4162,
+      "step": 1620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406506107842532e-06,
+      "loss": 0.4177,
+      "step": 1621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3947506024382665e-06,
+      "loss": 0.4112,
+      "step": 1622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.383003168591781e-06,
+      "loss": 0.4001,
+      "step": 1623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.37126382689264e-06,
+      "loss": 0.4195,
+      "step": 1624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.359532597916233e-06,
+      "loss": 0.4128,
+      "step": 1625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.347809502223717e-06,
+      "loss": 0.4242,
+      "step": 1626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.336094560362006e-06,
+      "loss": 0.4019,
+      "step": 1627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324387792863719e-06,
+      "loss": 0.4081,
+      "step": 1628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.312689220247152e-06,
+      "loss": 0.4256,
+      "step": 1629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.300998863016223e-06,
+      "loss": 0.4106,
+      "step": 1630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.289316741660466e-06,
+      "loss": 0.4298,
+      "step": 1631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277642876654978e-06,
+      "loss": 0.4163,
+      "step": 1632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.265977288460387e-06,
+      "loss": 0.406,
+      "step": 1633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.254319997522797e-06,
+      "loss": 0.4137,
+      "step": 1634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242671024273798e-06,
+      "loss": 0.4117,
+      "step": 1635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.231030389130375e-06,
+      "loss": 0.4025,
+      "step": 1636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.21939811249492e-06,
+      "loss": 0.4011,
+      "step": 1637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.20777421475515e-06,
+      "loss": 0.4162,
+      "step": 1638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.196158716284129e-06,
+      "loss": 0.4176,
+      "step": 1639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1845516374401786e-06,
+      "loss": 0.3986,
+      "step": 1640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.17295299856686e-06,
+      "loss": 0.4099,
+      "step": 1641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.161362819992954e-06,
+      "loss": 0.4111,
+      "step": 1642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14978112203241e-06,
+      "loss": 0.4034,
+      "step": 1643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138207924984313e-06,
+      "loss": 0.3998,
+      "step": 1644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.126643249132843e-06,
+      "loss": 0.4006,
+      "step": 1645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.11508711474725e-06,
+      "loss": 0.4112,
+      "step": 1646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.103539542081815e-06,
+      "loss": 0.3918,
+      "step": 1647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092000551375814e-06,
+      "loss": 0.4084,
+      "step": 1648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.080470162853473e-06,
+      "loss": 0.4145,
+      "step": 1649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.068948396723947e-06,
+      "loss": 0.4199,
+      "step": 1650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0574352731812816e-06,
+      "loss": 0.3976,
+      "step": 1651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045930812404372e-06,
+      "loss": 0.3973,
+      "step": 1652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0344350345569244e-06,
+      "loss": 0.4358,
+      "step": 1653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0229479597874355e-06,
+      "loss": 0.416,
+      "step": 1654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.011469608229143e-06,
+      "loss": 0.4006,
+      "step": 1655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.4089,
+      "step": 1656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.988539155202631e-06,
+      "loss": 0.4062,
+      "step": 1657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.977087093924299e-06,
+      "loss": 0.3894,
+      "step": 1658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965643836236878e-06,
+      "loss": 0.4121,
+      "step": 1659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954209402196813e-06,
+      "loss": 0.4118,
+      "step": 1660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.942783811845075e-06,
+      "loss": 0.411,
+      "step": 1661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931367085207142e-06,
+      "loss": 0.418,
+      "step": 1662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9199592422929545e-06,
+      "loss": 0.3986,
+      "step": 1663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9085603030968874e-06,
+      "loss": 0.3926,
+      "step": 1664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8971702875977e-06,
+      "loss": 0.4058,
+      "step": 1665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.88578921575852e-06,
+      "loss": 0.4153,
+      "step": 1666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.874417107526796e-06,
+      "loss": 0.3936,
+      "step": 1667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.863053982834266e-06,
+      "loss": 0.3812,
+      "step": 1668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85169986159693e-06,
+      "loss": 0.4031,
+      "step": 1669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.840354763714991e-06,
+      "loss": 0.388,
+      "step": 1670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.829018709072855e-06,
+      "loss": 0.3922,
+      "step": 1671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.817691717539066e-06,
+      "loss": 0.403,
+      "step": 1672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.806373808966293e-06,
+      "loss": 0.392,
+      "step": 1673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.795065003191273e-06,
+      "loss": 0.4176,
+      "step": 1674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.783765320034798e-06,
+      "loss": 0.3993,
+      "step": 1675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772474779301669e-06,
+      "loss": 0.3862,
+      "step": 1676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.761193400780667e-06,
+      "loss": 0.4101,
+      "step": 1677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.749921204244503e-06,
+      "loss": 0.4099,
+      "step": 1678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7386582094498045e-06,
+      "loss": 0.3982,
+      "step": 1679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.72740443613707e-06,
+      "loss": 0.3929,
+      "step": 1680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.716159904030637e-06,
+      "loss": 0.4012,
+      "step": 1681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.704924632838636e-06,
+      "loss": 0.4057,
+      "step": 1682
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6936986422529795e-06,
+      "loss": 0.4293,
+      "step": 1683
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.682481951949306e-06,
+      "loss": 0.4142,
+      "step": 1684
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671274581586959e-06,
+      "loss": 0.4037,
+      "step": 1685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660076550808936e-06,
+      "loss": 0.3992,
+      "step": 1686
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.648887879241879e-06,
+      "loss": 0.3922,
+      "step": 1687
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637708586496018e-06,
+      "loss": 0.4179,
+      "step": 1688
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.62653869216515e-06,
+      "loss": 0.4234,
+      "step": 1689
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.61537821582659e-06,
+      "loss": 0.4007,
+      "step": 1690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.604227177041156e-06,
+      "loss": 0.3958,
+      "step": 1691
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.593085595353122e-06,
+      "loss": 0.4102,
+      "step": 1692
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.581953490290188e-06,
+      "loss": 0.3947,
+      "step": 1693
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57083088136344e-06,
+      "loss": 0.4035,
+      "step": 1694
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.559717788067316e-06,
+      "loss": 0.4128,
+      "step": 1695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5486142298795995e-06,
+      "loss": 0.4085,
+      "step": 1696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.537520226261333e-06,
+      "loss": 0.3929,
+      "step": 1697
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52643579665683e-06,
+      "loss": 0.4055,
+      "step": 1698
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5153609604936125e-06,
+      "loss": 0.4058,
+      "step": 1699
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5042957371824055e-06,
+      "loss": 0.4154,
+      "step": 1700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.493240146117066e-06,
+      "loss": 0.4047,
+      "step": 1701
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482194206674585e-06,
+      "loss": 0.4143,
+      "step": 1702
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.471157938215017e-06,
+      "loss": 0.397,
+      "step": 1703
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4601313600814966e-06,
+      "loss": 0.4103,
+      "step": 1704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449114491600143e-06,
+      "loss": 0.3949,
+      "step": 1705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.438107352080076e-06,
+      "loss": 0.4068,
+      "step": 1706
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.427109960813361e-06,
+      "loss": 0.404,
+      "step": 1707
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416122337074975e-06,
+      "loss": 0.4068,
+      "step": 1708
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.405144500122772e-06,
+      "loss": 0.4083,
+      "step": 1709
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.39417646919746e-06,
+      "loss": 0.394,
+      "step": 1710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.383218263522556e-06,
+      "loss": 0.394,
+      "step": 1711
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3722699023043634e-06,
+      "loss": 0.4051,
+      "step": 1712
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361331404731917e-06,
+      "loss": 0.4002,
+      "step": 1713
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.350402789976975e-06,
+      "loss": 0.4021,
+      "step": 1714
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.339484077193974e-06,
+      "loss": 0.4245,
+      "step": 1715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328575285519994e-06,
+      "loss": 0.3938,
+      "step": 1716
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317676434074718e-06,
+      "loss": 0.3899,
+      "step": 1717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306787541960419e-06,
+      "loss": 0.4149,
+      "step": 1718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29590862826191e-06,
+      "loss": 0.4166,
+      "step": 1719
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.285039712046517e-06,
+      "loss": 0.3991,
+      "step": 1720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2741808123640335e-06,
+      "loss": 0.414,
+      "step": 1721
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.263331948246711e-06,
+      "loss": 0.4003,
+      "step": 1722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2524931387092036e-06,
+      "loss": 0.4093,
+      "step": 1723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241664402748544e-06,
+      "loss": 0.3866,
+      "step": 1724
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230845759344116e-06,
+      "loss": 0.3996,
+      "step": 1725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.220037227457601e-06,
+      "loss": 0.384,
+      "step": 1726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.209238826032965e-06,
+      "loss": 0.3972,
+      "step": 1727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.198450573996423e-06,
+      "loss": 0.4096,
+      "step": 1728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1876724902564e-06,
+      "loss": 0.4082,
+      "step": 1729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176904593703488e-06,
+      "loss": 0.3996,
+      "step": 1730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.166146903210436e-06,
+      "loss": 0.4217,
+      "step": 1731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.155399437632103e-06,
+      "loss": 0.399,
+      "step": 1732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.144662215805426e-06,
+      "loss": 0.4091,
+      "step": 1733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1339352565493825e-06,
+      "loss": 0.4103,
+      "step": 1734
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12321857866497e-06,
+      "loss": 0.403,
+      "step": 1735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1125122009351636e-06,
+      "loss": 0.4069,
+      "step": 1736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1018161421248905e-06,
+      "loss": 0.4009,
+      "step": 1737
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091130420980978e-06,
+      "loss": 0.3998,
+      "step": 1738
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080455056232148e-06,
+      "loss": 0.3958,
+      "step": 1739
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069790066588966e-06,
+      "loss": 0.399,
+      "step": 1740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05913547074382e-06,
+      "loss": 0.4018,
+      "step": 1741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048491287370864e-06,
+      "loss": 0.3958,
+      "step": 1742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.037857535126019e-06,
+      "loss": 0.3958,
+      "step": 1743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.027234232646916e-06,
+      "loss": 0.4187,
+      "step": 1744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.016621398552877e-06,
+      "loss": 0.4185,
+      "step": 1745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0060190514448634e-06,
+      "loss": 0.3905,
+      "step": 1746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9954272099054696e-06,
+      "loss": 0.3904,
+      "step": 1747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984845892498869e-06,
+      "loss": 0.4078,
+      "step": 1748
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974275117770798e-06,
+      "loss": 0.4104,
+      "step": 1749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.963714904248501e-06,
+      "loss": 0.4099,
+      "step": 1750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953165270440721e-06,
+      "loss": 0.4094,
+      "step": 1751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9426262348376676e-06,
+      "loss": 0.3969,
+      "step": 1752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932097815910954e-06,
+      "loss": 0.4018,
+      "step": 1753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921580032113602e-06,
+      "loss": 0.4011,
+      "step": 1754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911072901879979e-06,
+      "loss": 0.4021,
+      "step": 1755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.900576443625803e-06,
+      "loss": 0.419,
+      "step": 1756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890090675748062e-06,
+      "loss": 0.392,
+      "step": 1757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879615616625024e-06,
+      "loss": 0.4001,
+      "step": 1758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.869151284616174e-06,
+      "loss": 0.4011,
+      "step": 1759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858697698062217e-06,
+      "loss": 0.3865,
+      "step": 1760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848254875285e-06,
+      "loss": 0.4041,
+      "step": 1761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8378228345875244e-06,
+      "loss": 0.3919,
+      "step": 1762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.827401594253875e-06,
+      "loss": 0.3906,
+      "step": 1763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81699117254923e-06,
+      "loss": 0.4121,
+      "step": 1764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8065915877197847e-06,
+      "loss": 0.4026,
+      "step": 1765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7962028579927555e-06,
+      "loss": 0.4083,
+      "step": 1766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7858250015763176e-06,
+      "loss": 0.3988,
+      "step": 1767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7754580366596116e-06,
+      "loss": 0.4144,
+      "step": 1768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.3947,
+      "step": 1769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.754756853986402e-06,
+      "loss": 0.3971,
+      "step": 1770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7444226725125767e-06,
+      "loss": 0.423,
+      "step": 1771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.734099455103779e-06,
+      "loss": 0.3929,
+      "step": 1772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.723787219853363e-06,
+      "loss": 0.4064,
+      "step": 1773
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7134859848354486e-06,
+      "loss": 0.4032,
+      "step": 1774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7031957681048604e-06,
+      "loss": 0.4048,
+      "step": 1775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.692916587697134e-06,
+      "loss": 0.3963,
+      "step": 1776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6826484616284388e-06,
+      "loss": 0.4138,
+      "step": 1777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6723914078955824e-06,
+      "loss": 0.4029,
+      "step": 1778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.662145444475963e-06,
+      "loss": 0.4139,
+      "step": 1779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.65191058932754e-06,
+      "loss": 0.3999,
+      "step": 1780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.64168686038881e-06,
+      "loss": 0.3862,
+      "step": 1781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6314742755787537e-06,
+      "loss": 0.3941,
+      "step": 1782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6212728527968345e-06,
+      "loss": 0.3934,
+      "step": 1783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6110826099229457e-06,
+      "loss": 0.4061,
+      "step": 1784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.600903564817392e-06,
+      "loss": 0.4088,
+      "step": 1785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5907357353208374e-06,
+      "loss": 0.4028,
+      "step": 1786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5805791392543033e-06,
+      "loss": 0.411,
+      "step": 1787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.570433794419117e-06,
+      "loss": 0.4117,
+      "step": 1788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.560299718596889e-06,
+      "loss": 0.4105,
+      "step": 1789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.550176929549468e-06,
+      "loss": 0.3809,
+      "step": 1790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540065445018933e-06,
+      "loss": 0.3961,
+      "step": 1791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5299652827275455e-06,
+      "loss": 0.3998,
+      "step": 1792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5198764603777235e-06,
+      "loss": 0.3925,
+      "step": 1793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5097989956520016e-06,
+      "loss": 0.4058,
+      "step": 1794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.49973290621302e-06,
+      "loss": 0.4069,
+      "step": 1795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4896782097034755e-06,
+      "loss": 0.4143,
+      "step": 1796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4796349237461003e-06,
+      "loss": 0.3895,
+      "step": 1797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4696030659436174e-06,
+      "loss": 0.4005,
+      "step": 1798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4595826538787314e-06,
+      "loss": 0.4026,
+      "step": 1799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.449573705114082e-06,
+      "loss": 0.4142,
+      "step": 1800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4395762371922203e-06,
+      "loss": 0.4114,
+      "step": 1801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4295902676355654e-06,
+      "loss": 0.4141,
+      "step": 1802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.419615813946392e-06,
+      "loss": 0.4095,
+      "step": 1803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.409652893606791e-06,
+      "loss": 0.3886,
+      "step": 1804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3997015240786346e-06,
+      "loss": 0.3857,
+      "step": 1805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3897617228035573e-06,
+      "loss": 0.3896,
+      "step": 1806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3798335072029033e-06,
+      "loss": 0.3975,
+      "step": 1807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.369916894677733e-06,
+      "loss": 0.4097,
+      "step": 1808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3600119026087475e-06,
+      "loss": 0.4081,
+      "step": 1809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.350118548356299e-06,
+      "loss": 0.3907,
+      "step": 1810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3402368492603245e-06,
+      "loss": 0.3964,
+      "step": 1811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.330366822640356e-06,
+      "loss": 0.4186,
+      "step": 1812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3205084857954452e-06,
+      "loss": 0.4076,
+      "step": 1813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3106618560041704e-06,
+      "loss": 0.4001,
+      "step": 1814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.300826950524575e-06,
+      "loss": 0.4005,
+      "step": 1815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.291003786594178e-06,
+      "loss": 0.4142,
+      "step": 1816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.281192381429894e-06,
+      "loss": 0.413,
+      "step": 1817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2713927522280455e-06,
+      "loss": 0.405,
+      "step": 1818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2616049161643005e-06,
+      "loss": 0.3996,
+      "step": 1819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2518288903936767e-06,
+      "loss": 0.3832,
+      "step": 1820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2420646920504727e-06,
+      "loss": 0.4075,
+      "step": 1821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2323123382482713e-06,
+      "loss": 0.4309,
+      "step": 1822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2225718460798816e-06,
+      "loss": 0.4127,
+      "step": 1823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.212843232617343e-06,
+      "loss": 0.4044,
+      "step": 1824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.203126514911854e-06,
+      "loss": 0.4077,
+      "step": 1825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1934217099937794e-06,
+      "loss": 0.3776,
+      "step": 1826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1837288348725904e-06,
+      "loss": 0.4026,
+      "step": 1827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.174047906536867e-06,
+      "loss": 0.3993,
+      "step": 1828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1643789419542325e-06,
+      "loss": 0.3929,
+      "step": 1829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.154721958071356e-06,
+      "loss": 0.3928,
+      "step": 1830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.145076971813891e-06,
+      "loss": 0.397,
+      "step": 1831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.135444000086485e-06,
+      "loss": 0.3919,
+      "step": 1832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.125823059772708e-06,
+      "loss": 0.3921,
+      "step": 1833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1162141677350532e-06,
+      "loss": 0.4015,
+      "step": 1834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.106617340814896e-06,
+      "loss": 0.3899,
+      "step": 1835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.097032595832462e-06,
+      "loss": 0.3961,
+      "step": 1836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0874599495868074e-06,
+      "loss": 0.4023,
+      "step": 1837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0778994188557722e-06,
+      "loss": 0.3984,
+      "step": 1838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.068351020395971e-06,
+      "loss": 0.3855,
+      "step": 1839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.058814770942751e-06,
+      "loss": 0.4108,
+      "step": 1840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.04929068721017e-06,
+      "loss": 0.4008,
+      "step": 1841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0397787858909543e-06,
+      "loss": 0.4128,
+      "step": 1842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0302790836564853e-06,
+      "loss": 0.4171,
+      "step": 1843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0207915971567624e-06,
+      "loss": 0.3979,
+      "step": 1844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0113163430203775e-06,
+      "loss": 0.3782,
+      "step": 1845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.001853337854471e-06,
+      "loss": 0.3907,
+      "step": 1846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9924025982447267e-06,
+      "loss": 0.3994,
+      "step": 1847
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9829641407553277e-06,
+      "loss": 0.4052,
+      "step": 1848
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.973537981928932e-06,
+      "loss": 0.3953,
+      "step": 1849
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.964124138286635e-06,
+      "loss": 0.4062,
+      "step": 1850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9547226263279525e-06,
+      "loss": 0.4067,
+      "step": 1851
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.945333462530788e-06,
+      "loss": 0.3821,
+      "step": 1852
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9359566633514034e-06,
+      "loss": 0.404,
+      "step": 1853
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.926592245224381e-06,
+      "loss": 0.4151,
+      "step": 1854
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9172402245626107e-06,
+      "loss": 0.4083,
+      "step": 1855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9079006177572523e-06,
+      "loss": 0.4159,
+      "step": 1856
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89857344117771e-06,
+      "loss": 0.3897,
+      "step": 1857
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8892587111715918e-06,
+      "loss": 0.405,
+      "step": 1858
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.879956444064703e-06,
+      "loss": 0.4029,
+      "step": 1859
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.870666656161e-06,
+      "loss": 0.4042,
+      "step": 1860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.861389363742565e-06,
+      "loss": 0.3966,
+      "step": 1861
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8521245830695864e-06,
+      "loss": 0.3912,
+      "step": 1862
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.842872330380314e-06,
+      "loss": 0.404,
+      "step": 1863
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.833632621891046e-06,
+      "loss": 0.406,
+      "step": 1864
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8244054737960935e-06,
+      "loss": 0.3994,
+      "step": 1865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.815190902267757e-06,
+      "loss": 0.4015,
+      "step": 1866
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.805988923456283e-06,
+      "loss": 0.4018,
+      "step": 1867
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7967995534898595e-06,
+      "loss": 0.3988,
+      "step": 1868
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.787622808474567e-06,
+      "loss": 0.4114,
+      "step": 1869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.778458704494367e-06,
+      "loss": 0.4184,
+      "step": 1870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7693072576110515e-06,
+      "loss": 0.4079,
+      "step": 1871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.760168483864241e-06,
+      "loss": 0.3946,
+      "step": 1872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7510423992713374e-06,
+      "loss": 0.3912,
+      "step": 1873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7419290198275096e-06,
+      "loss": 0.4054,
+      "step": 1874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.732828361505647e-06,
+      "loss": 0.4023,
+      "step": 1875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7237404402563516e-06,
+      "loss": 0.4229,
+      "step": 1876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7146652720079e-06,
+      "loss": 0.4023,
+      "step": 1877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7056028726662175e-06,
+      "loss": 0.3994,
+      "step": 1878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.69655325811484e-06,
+      "loss": 0.3944,
+      "step": 1879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6875164442149147e-06,
+      "loss": 0.4118,
+      "step": 1880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6784924468051342e-06,
+      "loss": 0.3927,
+      "step": 1881
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.669481281701739e-06,
+      "loss": 0.3847,
+      "step": 1882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6604829646984687e-06,
+      "loss": 0.3887,
+      "step": 1883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.651497511566562e-06,
+      "loss": 0.4167,
+      "step": 1884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6425249380546914e-06,
+      "loss": 0.4007,
+      "step": 1885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6335652598889682e-06,
+      "loss": 0.3895,
+      "step": 1886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6246184927728913e-06,
+      "loss": 0.4072,
+      "step": 1887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.615684652387348e-06,
+      "loss": 0.3973,
+      "step": 1888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.60676375439055e-06,
+      "loss": 0.3886,
+      "step": 1889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5978558144180368e-06,
+      "loss": 0.4026,
+      "step": 1890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5889608480826323e-06,
+      "loss": 0.4026,
+      "step": 1891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5800788709744228e-06,
+      "loss": 0.4109,
+      "step": 1892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5712098986607325e-06,
+      "loss": 0.3829,
+      "step": 1893
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5623539466860813e-06,
+      "loss": 0.3969,
+      "step": 1894
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5535110305721777e-06,
+      "loss": 0.4024,
+      "step": 1895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5446811658178816e-06,
+      "loss": 0.4101,
+      "step": 1896
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.535864367899179e-06,
+      "loss": 0.4229,
+      "step": 1897
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.527060652269144e-06,
+      "loss": 0.401,
+      "step": 1898
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5182700343579337e-06,
+      "loss": 0.403,
+      "step": 1899
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5094925295727423e-06,
+      "loss": 0.4176,
+      "step": 1900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.500728153297788e-06,
+      "loss": 0.411,
+      "step": 1901
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4919769208942658e-06,
+      "loss": 0.4046,
+      "step": 1902
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4832388477003444e-06,
+      "loss": 0.4039,
+      "step": 1903
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4745139490311254e-06,
+      "loss": 0.3824,
+      "step": 1904
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4658022401786243e-06,
+      "loss": 0.4124,
+      "step": 1905
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4571037364117255e-06,
+      "loss": 0.4113,
+      "step": 1906
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4484184529761836e-06,
+      "loss": 0.3839,
+      "step": 1907
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4397464050945753e-06,
+      "loss": 0.422,
+      "step": 1908
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4310876079662824e-06,
+      "loss": 0.3982,
+      "step": 1909
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422442076767456e-06,
+      "loss": 0.3982,
+      "step": 1910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4138098266510036e-06,
+      "loss": 0.4028,
+      "step": 1911
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4051908727465513e-06,
+      "loss": 0.4015,
+      "step": 1912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.396585230160425e-06,
+      "loss": 0.3891,
+      "step": 1913
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3879929139756133e-06,
+      "loss": 0.4157,
+      "step": 1914
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.379413939251751e-06,
+      "loss": 0.3974,
+      "step": 1915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.370848321025093e-06,
+      "loss": 0.4036,
+      "step": 1916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.36229607430848e-06,
+      "loss": 0.3749,
+      "step": 1917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.353757214091321e-06,
+      "loss": 0.3918,
+      "step": 1918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3452317553395544e-06,
+      "loss": 0.4076,
+      "step": 1919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3367197129956376e-06,
+      "loss": 0.3826,
+      "step": 1920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.328221101978513e-06,
+      "loss": 0.3847,
+      "step": 1921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3197359371835802e-06,
+      "loss": 0.3902,
+      "step": 1922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3112642334826686e-06,
+      "loss": 0.3974,
+      "step": 1923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3028060057240186e-06,
+      "loss": 0.3992,
+      "step": 1924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2943612687322525e-06,
+      "loss": 0.427,
+      "step": 1925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.285930037308347e-06,
+      "loss": 0.4097,
+      "step": 1926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.277512326229601e-06,
+      "loss": 0.3978,
+      "step": 1927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.269108150249625e-06,
+      "loss": 0.4002,
+      "step": 1928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2607175240983027e-06,
+      "loss": 0.4025,
+      "step": 1929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2523404624817733e-06,
+      "loss": 0.4141,
+      "step": 1930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2439769800823942e-06,
+      "loss": 0.3823,
+      "step": 1931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2356270915587274e-06,
+      "loss": 0.3917,
+      "step": 1932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2272908115455105e-06,
+      "loss": 0.3997,
+      "step": 1933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2189681546536292e-06,
+      "loss": 0.3831,
+      "step": 1934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2106591354700846e-06,
+      "loss": 0.4006,
+      "step": 1935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2023637685579856e-06,
+      "loss": 0.3967,
+      "step": 1936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.194082068456509e-06,
+      "loss": 0.3952,
+      "step": 1937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1858140496808776e-06,
+      "loss": 0.4103,
+      "step": 1938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1775597267223327e-06,
+      "loss": 0.4016,
+      "step": 1939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.169319114048114e-06,
+      "loss": 0.396,
+      "step": 1940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.161092226101432e-06,
+      "loss": 0.4015,
+      "step": 1941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.152879077301443e-06,
+      "loss": 0.4092,
+      "step": 1942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.144679682043217e-06,
+      "loss": 0.3937,
+      "step": 1943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.136494054697722e-06,
+      "loss": 0.4032,
+      "step": 1944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1283222096117984e-06,
+      "loss": 0.397,
+      "step": 1945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1201641611081246e-06,
+      "loss": 0.395,
+      "step": 1946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.112019923485207e-06,
+      "loss": 0.4027,
+      "step": 1947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1038895110173287e-06,
+      "loss": 0.3955,
+      "step": 1948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0957729379545654e-06,
+      "loss": 0.4087,
+      "step": 1949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.087670218522714e-06,
+      "loss": 0.3926,
+      "step": 1950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0795813669233078e-06,
+      "loss": 0.4103,
+      "step": 1951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.071506397333557e-06,
+      "loss": 0.3994,
+      "step": 1952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0634453239063623e-06,
+      "loss": 0.4029,
+      "step": 1953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0553981607702477e-06,
+      "loss": 0.4016,
+      "step": 1954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.04736492202937e-06,
+      "loss": 0.3889,
+      "step": 1955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0393456217634777e-06,
+      "loss": 0.3864,
+      "step": 1956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.031340274027891e-06,
+      "loss": 0.4167,
+      "step": 1957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.023348892853467e-06,
+      "loss": 0.3907,
+      "step": 1958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.015371492246596e-06,
+      "loss": 0.4074,
+      "step": 1959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0074080861891564e-06,
+      "loss": 0.3964,
+      "step": 1960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994586886385046e-06,
+      "loss": 0.4093,
+      "step": 1961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.991523313527437e-06,
+      "loss": 0.3917,
+      "step": 1962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98360197476418e-06,
+      "loss": 0.4116,
+      "step": 1963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9756946862323534e-06,
+      "loss": 0.3861,
+      "step": 1964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9678014617909604e-06,
+      "loss": 0.4012,
+      "step": 1965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95992231527434e-06,
+      "loss": 0.4091,
+      "step": 1966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.952057260492167e-06,
+      "loss": 0.4089,
+      "step": 1967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9442063112294164e-06,
+      "loss": 0.4015,
+      "step": 1968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9363694812463442e-06,
+      "loss": 0.3836,
+      "step": 1969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9285467842784465e-06,
+      "loss": 0.4028,
+      "step": 1970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.920738234036463e-06,
+      "loss": 0.4065,
+      "step": 1971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.912943844206333e-06,
+      "loss": 0.418,
+      "step": 1972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9051636284491758e-06,
+      "loss": 0.4009,
+      "step": 1973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.897397600401274e-06,
+      "loss": 0.4144,
+      "step": 1974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8896457736740316e-06,
+      "loss": 0.4027,
+      "step": 1975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8819081618539725e-06,
+      "loss": 0.3996,
+      "step": 1976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8741847785027046e-06,
+      "loss": 0.3869,
+      "step": 1977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8664756371568981e-06,
+      "loss": 0.3963,
+      "step": 1978
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8587807513282552e-06,
+      "loss": 0.3979,
+      "step": 1979
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8511001345034995e-06,
+      "loss": 0.4004,
+      "step": 1980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.843433800144343e-06,
+      "loss": 0.3951,
+      "step": 1981
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8357817616874696e-06,
+      "loss": 0.3949,
+      "step": 1982
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8281440325444954e-06,
+      "loss": 0.3948,
+      "step": 1983
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.820520626101967e-06,
+      "loss": 0.3898,
+      "step": 1984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8129115557213262e-06,
+      "loss": 0.4088,
+      "step": 1985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8053168347388882e-06,
+      "loss": 0.3956,
+      "step": 1986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7977364764658124e-06,
+      "loss": 0.4194,
+      "step": 1987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7901704941880914e-06,
+      "loss": 0.3966,
+      "step": 1988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7826189011665186e-06,
+      "loss": 0.3962,
+      "step": 1989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7750817106366714e-06,
+      "loss": 0.3752,
+      "step": 1990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7675589358088763e-06,
+      "loss": 0.3744,
+      "step": 1991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7600505898681996e-06,
+      "loss": 0.4043,
+      "step": 1992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.752556685974417e-06,
+      "loss": 0.4042,
+      "step": 1993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.745077237261994e-06,
+      "loss": 0.3982,
+      "step": 1994
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7376122568400533e-06,
+      "loss": 0.4122,
+      "step": 1995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7301617577923668e-06,
+      "loss": 0.402,
+      "step": 1996
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7227257531773223e-06,
+      "loss": 0.3923,
+      "step": 1997
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7153042560279064e-06,
+      "loss": 0.3888,
+      "step": 1998
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7078972793516713e-06,
+      "loss": 0.3975,
+      "step": 1999
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7005048361307264e-06,
+      "loss": 0.3907,
+      "step": 2000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6931269393217054e-06,
+      "loss": 0.4058,
+      "step": 2001
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6857636018557466e-06,
+      "loss": 0.3919,
+      "step": 2002
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6784148366384756e-06,
+      "loss": 0.4089,
+      "step": 2003
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6710806565499648e-06,
+      "loss": 0.3871,
+      "step": 2004
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6637610744447407e-06,
+      "loss": 0.4015,
+      "step": 2005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6564561031517278e-06,
+      "loss": 0.3789,
+      "step": 2006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6491657554742556e-06,
+      "loss": 0.3937,
+      "step": 2007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6418900441900088e-06,
+      "loss": 0.3838,
+      "step": 2008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6346289820510364e-06,
+      "loss": 0.3824,
+      "step": 2009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6273825817836964e-06,
+      "loss": 0.3833,
+      "step": 2010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6201508560886602e-06,
+      "loss": 0.3926,
+      "step": 2011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.612933817640868e-06,
+      "loss": 0.4211,
+      "step": 2012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.605731479089534e-06,
+      "loss": 0.4094,
+      "step": 2013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.598543853058091e-06,
+      "loss": 0.3921,
+      "step": 2014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5913709521441989e-06,
+      "loss": 0.3951,
+      "step": 2015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5842127889196957e-06,
+      "loss": 0.4027,
+      "step": 2016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5770693759306055e-06,
+      "loss": 0.4127,
+      "step": 2017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5699407256970834e-06,
+      "loss": 0.3895,
+      "step": 2018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5628268507134225e-06,
+      "loss": 0.3865,
+      "step": 2019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5557277634480084e-06,
+      "loss": 0.4047,
+      "step": 2020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5486434763433222e-06,
+      "loss": 0.414,
+      "step": 2021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.541574001815892e-06,
+      "loss": 0.4013,
+      "step": 2022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5345193522562918e-06,
+      "loss": 0.4064,
+      "step": 2023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5274795400291044e-06,
+      "loss": 0.4067,
+      "step": 2024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5204545774729207e-06,
+      "loss": 0.412,
+      "step": 2025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.513444476900291e-06,
+      "loss": 0.3968,
+      "step": 2026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5064492505977234e-06,
+      "loss": 0.4053,
+      "step": 2027
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4994689108256577e-06,
+      "loss": 0.3984,
+      "step": 2028
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4925034698184393e-06,
+      "loss": 0.3918,
+      "step": 2029
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4855529397843039e-06,
+      "loss": 0.3805,
+      "step": 2030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4786173329053466e-06,
+      "loss": 0.4003,
+      "step": 2031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4716966613375116e-06,
+      "loss": 0.3958,
+      "step": 2032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4647909372105673e-06,
+      "loss": 0.3953,
+      "step": 2033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4579001726280828e-06,
+      "loss": 0.3891,
+      "step": 2034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.451024379667404e-06,
+      "loss": 0.3833,
+      "step": 2035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4441635703796409e-06,
+      "loss": 0.3941,
+      "step": 2036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4373177567896412e-06,
+      "loss": 0.3982,
+      "step": 2037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4304869508959707e-06,
+      "loss": 0.4042,
+      "step": 2038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4236711646708844e-06,
+      "loss": 0.3963,
+      "step": 2039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4168704100603214e-06,
+      "loss": 0.3949,
+      "step": 2040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.41008469898387e-06,
+      "loss": 0.4233,
+      "step": 2041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.403314043334757e-06,
+      "loss": 0.3899,
+      "step": 2042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.396558454979814e-06,
+      "loss": 0.392,
+      "step": 2043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3898179457594686e-06,
+      "loss": 0.4051,
+      "step": 2044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3830925274877216e-06,
+      "loss": 0.3847,
+      "step": 2045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.376382211952123e-06,
+      "loss": 0.4143,
+      "step": 2046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3696870109137462e-06,
+      "loss": 0.4017,
+      "step": 2047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.363006936107183e-06,
+      "loss": 0.3885,
+      "step": 2048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.356341999240507e-06,
+      "loss": 0.4192,
+      "step": 2049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.349692211995266e-06,
+      "loss": 0.3878,
+      "step": 2050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.343057586026446e-06,
+      "loss": 0.3931,
+      "step": 2051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3364381329624688e-06,
+      "loss": 0.3902,
+      "step": 2052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3298338644051579e-06,
+      "loss": 0.4105,
+      "step": 2053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3232447919297276e-06,
+      "loss": 0.3908,
+      "step": 2054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.316670927084751e-06,
+      "loss": 0.4079,
+      "step": 2055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.310112281392153e-06,
+      "loss": 0.3949,
+      "step": 2056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3035688663471836e-06,
+      "loss": 0.3912,
+      "step": 2057
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2970406934183955e-06,
+      "loss": 0.3861,
+      "step": 2058
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2905277740476318e-06,
+      "loss": 0.3966,
+      "step": 2059
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2840301196499894e-06,
+      "loss": 0.3884,
+      "step": 2060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2775477416138292e-06,
+      "loss": 0.4034,
+      "step": 2061
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.271080651300719e-06,
+      "loss": 0.3957,
+      "step": 2062
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2646288600454448e-06,
+      "loss": 0.3859,
+      "step": 2063
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2581923791559647e-06,
+      "loss": 0.4006,
+      "step": 2064
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2517712199134225e-06,
+      "loss": 0.4054,
+      "step": 2065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2453653935720866e-06,
+      "loss": 0.3983,
+      "step": 2066
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2389749113593686e-06,
+      "loss": 0.3997,
+      "step": 2067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.232599784475772e-06,
+      "loss": 0.382,
+      "step": 2068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2262400240949023e-06,
+      "loss": 0.3935,
+      "step": 2069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.21989564136342e-06,
+      "loss": 0.3993,
+      "step": 2070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.213566647401041e-06,
+      "loss": 0.4026,
+      "step": 2071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2072530533005012e-06,
+      "loss": 0.3946,
+      "step": 2072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2009548701275598e-06,
+      "loss": 0.3825,
+      "step": 2073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1946721089209478e-06,
+      "loss": 0.4118,
+      "step": 2074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1884047806923816e-06,
+      "loss": 0.3942,
+      "step": 2075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.182152896426515e-06,
+      "loss": 0.3868,
+      "step": 2076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1759164670809486e-06,
+      "loss": 0.3895,
+      "step": 2077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.169695503586179e-06,
+      "loss": 0.3929,
+      "step": 2078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.163490016845611e-06,
+      "loss": 0.3963,
+      "step": 2079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1573000177355086e-06,
+      "loss": 0.4026,
+      "step": 2080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1511255171050084e-06,
+      "loss": 0.4048,
+      "step": 2081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1449665257760656e-06,
+      "loss": 0.3963,
+      "step": 2082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1388230545434652e-06,
+      "loss": 0.3882,
+      "step": 2083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.132695114174779e-06,
+      "loss": 0.3958,
+      "step": 2084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1265827154103703e-06,
+      "loss": 0.3951,
+      "step": 2085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.120485868963358e-06,
+      "loss": 0.3952,
+      "step": 2086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1144045855195972e-06,
+      "loss": 0.3899,
+      "step": 2087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1083388757376712e-06,
+      "loss": 0.4062,
+      "step": 2088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1022887502488687e-06,
+      "loss": 0.4048,
+      "step": 2089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0962542196571636e-06,
+      "loss": 0.3973,
+      "step": 2090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0902352945391903e-06,
+      "loss": 0.3956,
+      "step": 2091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0842319854442396e-06,
+      "loss": 0.4276,
+      "step": 2092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.078244302894229e-06,
+      "loss": 0.3825,
+      "step": 2093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0722722573836908e-06,
+      "loss": 0.3882,
+      "step": 2094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.066315859379743e-06,
+      "loss": 0.3943,
+      "step": 2095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0603751193220846e-06,
+      "loss": 0.3893,
+      "step": 2096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0544500476229713e-06,
+      "loss": 0.3825,
+      "step": 2097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.048540654667195e-06,
+      "loss": 0.4019,
+      "step": 2098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0426469508120662e-06,
+      "loss": 0.393,
+      "step": 2099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0367689463874009e-06,
+      "loss": 0.4116,
+      "step": 2100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0309066516954958e-06,
+      "loss": 0.4027,
+      "step": 2101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0250600770111186e-06,
+      "loss": 0.404,
+      "step": 2102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0192292325814757e-06,
+      "loss": 0.4051,
+      "step": 2103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0134141286262112e-06,
+      "loss": 0.3967,
+      "step": 2104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0076147753373787e-06,
+      "loss": 0.3889,
+      "step": 2105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.001831182879427e-06,
+      "loss": 0.4187,
+      "step": 2106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.960633613891757e-07,
+      "loss": 0.3971,
+      "step": 2107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.3923,
+      "step": 2108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.845750717208502e-07,
+      "loss": 0.4188,
+      "step": 2109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.78854623678146e-07,
+      "loss": 0.3903,
+      "step": 2110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.731499868738448e-07,
+      "loss": 0.3884,
+      "step": 2111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.674611713063863e-07,
+      "loss": 0.3931,
+      "step": 2112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.617881869464807e-07,
+      "loss": 0.3989,
+      "step": 2113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.561310437370907e-07,
+      "loss": 0.4173,
+      "step": 2114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.504897515934153e-07,
+      "loss": 0.4118,
+      "step": 2115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.448643204028662e-07,
+      "loss": 0.3935,
+      "step": 2116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.392547600250634e-07,
+      "loss": 0.4258,
+      "step": 2117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.336610802918044e-07,
+      "loss": 0.3928,
+      "step": 2118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.280832910070592e-07,
+      "loss": 0.3931,
+      "step": 2119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.225214019469387e-07,
+      "loss": 0.3945,
+      "step": 2120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.169754228596905e-07,
+      "loss": 0.3929,
+      "step": 2121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.114453634656784e-07,
+      "loss": 0.3994,
+      "step": 2122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.059312334573634e-07,
+      "loss": 0.4092,
+      "step": 2123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.004330424992813e-07,
+      "loss": 0.4043,
+      "step": 2124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.949508002280383e-07,
+      "loss": 0.4087,
+      "step": 2125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.894845162522869e-07,
+      "loss": 0.3828,
+      "step": 2126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.840342001527091e-07,
+      "loss": 0.3902,
+      "step": 2127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.785998614819957e-07,
+      "loss": 0.3957,
+      "step": 2128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.731815097648432e-07,
+      "loss": 0.4072,
+      "step": 2129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.677791544979175e-07,
+      "loss": 0.4109,
+      "step": 2130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.623928051498575e-07,
+      "loss": 0.3929,
+      "step": 2131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.570224711612385e-07,
+      "loss": 0.4133,
+      "step": 2132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.516681619445788e-07,
+      "loss": 0.3776,
+      "step": 2133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.463298868842973e-07,
+      "loss": 0.3952,
+      "step": 2134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.410076553367208e-07,
+      "loss": 0.39,
+      "step": 2135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.357014766300442e-07,
+      "loss": 0.3962,
+      "step": 2136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.304113600643437e-07,
+      "loss": 0.3924,
+      "step": 2137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.251373149115294e-07,
+      "loss": 0.3892,
+      "step": 2138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.198793504153491e-07,
+      "loss": 0.3831,
+      "step": 2139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.146374757913622e-07,
+      "loss": 0.4136,
+      "step": 2140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.094117002269363e-07,
+      "loss": 0.4126,
+      "step": 2141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.042020328812162e-07,
+      "loss": 0.3985,
+      "step": 2142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.990084828851108e-07,
+      "loss": 0.3942,
+      "step": 2143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.938310593412879e-07,
+      "loss": 0.4082,
+      "step": 2144
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.886697713241453e-07,
+      "loss": 0.4066,
+      "step": 2145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.835246278798037e-07,
+      "loss": 0.3888,
+      "step": 2146
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.783956380260837e-07,
+      "loss": 0.3947,
+      "step": 2147
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.73282810752497e-07,
+      "loss": 0.3925,
+      "step": 2148
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.681861550202252e-07,
+      "loss": 0.3875,
+      "step": 2149
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.631056797621106e-07,
+      "loss": 0.4057,
+      "step": 2150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.580413938826292e-07,
+      "loss": 0.3928,
+      "step": 2151
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.529933062578865e-07,
+      "loss": 0.403,
+      "step": 2152
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.479614257355972e-07,
+      "loss": 0.4001,
+      "step": 2153
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.42945761135071e-07,
+      "loss": 0.399,
+      "step": 2154
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.379463212471916e-07,
+      "loss": 0.3895,
+      "step": 2155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.329631148344119e-07,
+      "loss": 0.3807,
+      "step": 2156
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.279961506307288e-07,
+      "loss": 0.3991,
+      "step": 2157
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.23045437341674e-07,
+      "loss": 0.3923,
+      "step": 2158
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.181109836442913e-07,
+      "loss": 0.4167,
+      "step": 2159
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.131927981871345e-07,
+      "loss": 0.4044,
+      "step": 2160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.082908895902374e-07,
+      "loss": 0.3975,
+      "step": 2161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.034052664451118e-07,
+      "loss": 0.4059,
+      "step": 2162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.985359373147183e-07,
+      "loss": 0.4067,
+      "step": 2163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.936829107334663e-07,
+      "loss": 0.412,
+      "step": 2164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.88846195207189e-07,
+      "loss": 0.4069,
+      "step": 2165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.840257992131316e-07,
+      "loss": 0.4071,
+      "step": 2166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.79221731199936e-07,
+      "loss": 0.4037,
+      "step": 2167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.744339995876259e-07,
+      "loss": 0.4078,
+      "step": 2168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.696626127675943e-07,
+      "loss": 0.3892,
+      "step": 2169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.649075791025861e-07,
+      "loss": 0.4017,
+      "step": 2170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.601689069266837e-07,
+      "loss": 0.3997,
+      "step": 2171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.554466045452923e-07,
+      "loss": 0.3945,
+      "step": 2172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.507406802351269e-07,
+      "loss": 0.4102,
+      "step": 2173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.460511422441984e-07,
+      "loss": 0.3906,
+      "step": 2174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.413779987917956e-07,
+      "loss": 0.4013,
+      "step": 2175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.367212580684712e-07,
+      "loss": 0.3967,
+      "step": 2176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.320809282360319e-07,
+      "loss": 0.402,
+      "step": 2177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.274570174275218e-07,
+      "loss": 0.391,
+      "step": 2178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.228495337472074e-07,
+      "loss": 0.3812,
+      "step": 2179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.182584852705587e-07,
+      "loss": 0.4061,
+      "step": 2180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.136838800442457e-07,
+      "loss": 0.388,
+      "step": 2181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.091257260861172e-07,
+      "loss": 0.4057,
+      "step": 2182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.045840313851881e-07,
+      "loss": 0.3791,
+      "step": 2183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.000588039016209e-07,
+      "loss": 0.4118,
+      "step": 2184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.955500515667234e-07,
+      "loss": 0.3832,
+      "step": 2185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.910577822829233e-07,
+      "loss": 0.4207,
+      "step": 2186
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.865820039237624e-07,
+      "loss": 0.3988,
+      "step": 2187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.821227243338712e-07,
+      "loss": 0.3819,
+      "step": 2188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.776799513289711e-07,
+      "loss": 0.38,
+      "step": 2189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.732536926958487e-07,
+      "loss": 0.3936,
+      "step": 2190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.688439561923509e-07,
+      "loss": 0.4015,
+      "step": 2191
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.644507495473573e-07,
+      "loss": 0.3875,
+      "step": 2192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.600740804607829e-07,
+      "loss": 0.4019,
+      "step": 2193
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.557139566035574e-07,
+      "loss": 0.409,
+      "step": 2194
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.513703856176112e-07,
+      "loss": 0.3936,
+      "step": 2195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.470433751158577e-07,
+      "loss": 0.3908,
+      "step": 2196
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.427329326821906e-07,
+      "loss": 0.4114,
+      "step": 2197
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.384390658714688e-07,
+      "loss": 0.3822,
+      "step": 2198
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.341617822094869e-07,
+      "loss": 0.3954,
+      "step": 2199
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.299010891929857e-07,
+      "loss": 0.3799,
+      "step": 2200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.256569942896217e-07,
+      "loss": 0.4015,
+      "step": 2201
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.214295049379658e-07,
+      "loss": 0.3902,
+      "step": 2202
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.172186285474756e-07,
+      "loss": 0.3837,
+      "step": 2203
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.130243724984995e-07,
+      "loss": 0.3886,
+      "step": 2204
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.088467441422529e-07,
+      "loss": 0.4153,
+      "step": 2205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.046857508008085e-07,
+      "loss": 0.3956,
+      "step": 2206
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.005413997670816e-07,
+      "loss": 0.3888,
+      "step": 2207
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.964136983048184e-07,
+      "loss": 0.3902,
+      "step": 2208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.923026536485876e-07,
+      "loss": 0.3854,
+      "step": 2209
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.882082730037607e-07,
+      "loss": 0.3856,
+      "step": 2210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.84130563546501e-07,
+      "loss": 0.4223,
+      "step": 2211
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.800695324237548e-07,
+      "loss": 0.3997,
+      "step": 2212
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.760251867532362e-07,
+      "loss": 0.4042,
+      "step": 2213
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.719975336234162e-07,
+      "loss": 0.402,
+      "step": 2214
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6798658009350463e-07,
+      "loss": 0.3945,
+      "step": 2215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6399233319344703e-07,
+      "loss": 0.3696,
+      "step": 2216
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.600147999239035e-07,
+      "loss": 0.394,
+      "step": 2217
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.560539872562464e-07,
+      "loss": 0.4138,
+      "step": 2218
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.521099021325337e-07,
+      "loss": 0.4012,
+      "step": 2219
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.481825514655114e-07,
+      "loss": 0.3865,
+      "step": 2220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4427194213859216e-07,
+      "loss": 0.3993,
+      "step": 2221
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.403780810058511e-07,
+      "loss": 0.3945,
+      "step": 2222
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3650097489200125e-07,
+      "loss": 0.3864,
+      "step": 2223
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.326406305923958e-07,
+      "loss": 0.3978,
+      "step": 2224
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2879705487300693e-07,
+      "loss": 0.4004,
+      "step": 2225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2497025447041705e-07,
+      "loss": 0.3927,
+      "step": 2226
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.211602360918099e-07,
+      "loss": 0.3921,
+      "step": 2227
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.173670064149482e-07,
+      "loss": 0.3821,
+      "step": 2228
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1359057208817613e-07,
+      "loss": 0.3935,
+      "step": 2229
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.098309397303979e-07,
+      "loss": 0.3805,
+      "step": 2230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0608811593107254e-07,
+      "loss": 0.4055,
+      "step": 2231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0236210725019265e-07,
+      "loss": 0.3916,
+      "step": 2232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.986529202182832e-07,
+      "loss": 0.3834,
+      "step": 2233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.949605613363883e-07,
+      "loss": 0.3793,
+      "step": 2234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.912850370760535e-07,
+      "loss": 0.4117,
+      "step": 2235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8762635387931903e-07,
+      "loss": 0.3999,
+      "step": 2236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8398451815870984e-07,
+      "loss": 0.3756,
+      "step": 2237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8035953629722234e-07,
+      "loss": 0.3938,
+      "step": 2238
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.76751414648312e-07,
+      "loss": 0.4035,
+      "step": 2239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.731601595358847e-07,
+      "loss": 0.3919,
+      "step": 2240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6958577725428437e-07,
+      "loss": 0.3985,
+      "step": 2241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.660282740682808e-07,
+      "loss": 0.3825,
+      "step": 2242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6248765621306413e-07,
+      "loss": 0.4133,
+      "step": 2243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.589639298942238e-07,
+      "loss": 0.4117,
+      "step": 2244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5545710128774837e-07,
+      "loss": 0.4203,
+      "step": 2245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.519671765400079e-07,
+      "loss": 0.3958,
+      "step": 2246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.484941617677473e-07,
+      "loss": 0.3959,
+      "step": 2247
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4503806305807076e-07,
+      "loss": 0.3975,
+      "step": 2248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4159888646843497e-07,
+      "loss": 0.4112,
+      "step": 2249
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3817663802663935e-07,
+      "loss": 0.4248,
+      "step": 2250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3477132373081257e-07,
+      "loss": 0.3891,
+      "step": 2251
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.313829495493992e-07,
+      "loss": 0.4032,
+      "step": 2252
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2801152142115764e-07,
+      "loss": 0.4128,
+      "step": 2253
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2465704525514784e-07,
+      "loss": 0.4059,
+      "step": 2254
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.21319526930709e-07,
+      "loss": 0.3797,
+      "step": 2255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.179989722974686e-07,
+      "loss": 0.4004,
+      "step": 2256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1469538717531224e-07,
+      "loss": 0.4043,
+      "step": 2257
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.114087773543939e-07,
+      "loss": 0.3957,
+      "step": 2258
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0813914859510575e-07,
+      "loss": 0.4,
+      "step": 2259
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0488650662808483e-07,
+      "loss": 0.3759,
+      "step": 2260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.016508571541876e-07,
+      "loss": 0.4091,
+      "step": 2261
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9843220584449774e-07,
+      "loss": 0.4069,
+      "step": 2262
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.952305583402981e-07,
+      "loss": 0.396,
+      "step": 2263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9204592025307565e-07,
+      "loss": 0.3832,
+      "step": 2264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8887829716449877e-07,
+      "loss": 0.391,
+      "step": 2265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.857276946264198e-07,
+      "loss": 0.3858,
+      "step": 2266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.825941181608549e-07,
+      "loss": 0.3916,
+      "step": 2267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7947757325998305e-07,
+      "loss": 0.4119,
+      "step": 2268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7637806538612586e-07,
+      "loss": 0.3977,
+      "step": 2269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.732955999717546e-07,
+      "loss": 0.4066,
+      "step": 2270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7023018241945863e-07,
+      "loss": 0.4072,
+      "step": 2271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.67181818101957e-07,
+      "loss": 0.3923,
+      "step": 2272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6415051236207355e-07,
+      "loss": 0.392,
+      "step": 2273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.611362705127396e-07,
+      "loss": 0.3852,
+      "step": 2274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5813909783697354e-07,
+      "loss": 0.4105,
+      "step": 2275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.551589995878789e-07,
+      "loss": 0.395,
+      "step": 2276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.521959809886343e-07,
+      "loss": 0.4061,
+      "step": 2277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.492500472324832e-07,
+      "loss": 0.4021,
+      "step": 2278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4632120348272e-07,
+      "loss": 0.4001,
+      "step": 2279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4340945487269176e-07,
+      "loss": 0.4021,
+      "step": 2280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.405148065057794e-07,
+      "loss": 0.3993,
+      "step": 2281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.376372634553936e-07,
+      "loss": 0.393,
+      "step": 2282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.347768307649667e-07,
+      "loss": 0.3868,
+      "step": 2283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3193351344793836e-07,
+      "loss": 0.4077,
+      "step": 2284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2910731648775108e-07,
+      "loss": 0.4025,
+      "step": 2285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.262982448378437e-07,
+      "loss": 0.3957,
+      "step": 2286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.235063034216378e-07,
+      "loss": 0.3859,
+      "step": 2287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2073149713252918e-07,
+      "loss": 0.4008,
+      "step": 2288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1797383083388412e-07,
+      "loss": 0.408,
+      "step": 2289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1523330935902642e-07,
+      "loss": 0.402,
+      "step": 2290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.125099375112316e-07,
+      "loss": 0.3917,
+      "step": 2291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0980372006371486e-07,
+      "loss": 0.3946,
+      "step": 2292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.071146617596276e-07,
+      "loss": 0.3996,
+      "step": 2293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0444276731204416e-07,
+      "loss": 0.391,
+      "step": 2294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.017880414039608e-07,
+      "loss": 0.4105,
+      "step": 2295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9915048868827558e-07,
+      "loss": 0.4067,
+      "step": 2296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9653011378779285e-07,
+      "loss": 0.4109,
+      "step": 2297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9392692129520884e-07,
+      "loss": 0.3939,
+      "step": 2298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9134091577310278e-07,
+      "loss": 0.3966,
+      "step": 2299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.88772101753929e-07,
+      "loss": 0.4003,
+      "step": 2300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8622048374001612e-07,
+      "loss": 0.4126,
+      "step": 2301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.836860662035489e-07,
+      "loss": 0.4056,
+      "step": 2302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8116885358656743e-07,
+      "loss": 0.4076,
+      "step": 2303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.786688503009537e-07,
+      "loss": 0.3901,
+      "step": 2304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7618606072842938e-07,
+      "loss": 0.3944,
+      "step": 2305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7372048922054906e-07,
+      "loss": 0.3937,
+      "step": 2306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7127214009868387e-07,
+      "loss": 0.3898,
+      "step": 2307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6884101765402117e-07,
+      "loss": 0.4053,
+      "step": 2308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6642712614755697e-07,
+      "loss": 0.4246,
+      "step": 2309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6403046981008807e-07,
+      "loss": 0.3957,
+      "step": 2310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6165105284219774e-07,
+      "loss": 0.3798,
+      "step": 2311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5928887941426108e-07,
+      "loss": 0.3956,
+      "step": 2312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5694395366642413e-07,
+      "loss": 0.3918,
+      "step": 2313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5461627970860814e-07,
+      "loss": 0.3856,
+      "step": 2314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.523058616204942e-07,
+      "loss": 0.3908,
+      "step": 2315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5001270345152308e-07,
+      "loss": 0.3939,
+      "step": 2316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4773680922087863e-07,
+      "loss": 0.419,
+      "step": 2317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4547818291749116e-07,
+      "loss": 0.3952,
+      "step": 2318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4323682850002407e-07,
+      "loss": 0.4149,
+      "step": 2319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4101274989687052e-07,
+      "loss": 0.4057,
+      "step": 2320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.388059510061379e-07,
+      "loss": 0.3874,
+      "step": 2321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3661643569565786e-07,
+      "loss": 0.3864,
+      "step": 2322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3444420780296062e-07,
+      "loss": 0.4123,
+      "step": 2323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3228927113528189e-07,
+      "loss": 0.4076,
+      "step": 2324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3015162946954707e-07,
+      "loss": 0.3804,
+      "step": 2325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2803128655237694e-07,
+      "loss": 0.4178,
+      "step": 2326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2592824610006215e-07,
+      "loss": 0.4021,
+      "step": 2327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2384251179857642e-07,
+      "loss": 0.3953,
+      "step": 2328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2177408730355555e-07,
+      "loss": 0.3772,
+      "step": 2329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1972297624030072e-07,
+      "loss": 0.381,
+      "step": 2330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1768918220376624e-07,
+      "loss": 0.3967,
+      "step": 2331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1567270875855408e-07,
+      "loss": 0.3903,
+      "step": 2332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1367355943890823e-07,
+      "loss": 0.4004,
+      "step": 2333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1169173774871478e-07,
+      "loss": 0.4032,
+      "step": 2334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0972724716148186e-07,
+      "loss": 0.4025,
+      "step": 2335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0778009112034749e-07,
+      "loss": 0.4006,
+      "step": 2336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0585027303806395e-07,
+      "loss": 0.4011,
+      "step": 2337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0393779629699785e-07,
+      "loss": 0.4066,
+      "step": 2338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0204266424912124e-07,
+      "loss": 0.3847,
+      "step": 2339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.001648802160049e-07,
+      "loss": 0.4042,
+      "step": 2340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.830444748881729e-08,
+      "loss": 0.3987,
+      "step": 2341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.64613693283123e-08,
+      "loss": 0.3976,
+      "step": 2342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.463564896482813e-08,
+      "loss": 0.3972,
+      "step": 2343
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.282728959827958e-08,
+      "loss": 0.4076,
+      "step": 2344
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.103629439815353e-08,
+      "loss": 0.4128,
+      "step": 2345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.926266650350235e-08,
+      "loss": 0.3921,
+      "step": 2346
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.750640902294161e-08,
+      "loss": 0.3991,
+      "step": 2347
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.576752503463681e-08,
+      "loss": 0.3984,
+      "step": 2348
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.404601758630892e-08,
+      "loss": 0.3987,
+      "step": 2349
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.234188969521884e-08,
+      "loss": 0.3904,
+      "step": 2350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.065514434816846e-08,
+      "loss": 0.3991,
+      "step": 2351
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.898578450149408e-08,
+      "loss": 0.3923,
+      "step": 2352
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.73338130810597e-08,
+      "loss": 0.3933,
+      "step": 2353
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.569923298225146e-08,
+      "loss": 0.3842,
+      "step": 2354
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.408204706997879e-08,
+      "loss": 0.3963,
+      "step": 2355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.248225817865883e-08,
+      "loss": 0.3799,
+      "step": 2356
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.08998691122198e-08,
+      "loss": 0.3937,
+      "step": 2357
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.933488264409538e-08,
+      "loss": 0.3967,
+      "step": 2358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.77873015172148e-08,
+      "loss": 0.4094,
+      "step": 2359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.625712844400056e-08,
+      "loss": 0.3971,
+      "step": 2360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.474436610636958e-08,
+      "loss": 0.4005,
+      "step": 2361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.324901715571652e-08,
+      "loss": 0.3838,
+      "step": 2362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.177108421292266e-08,
+      "loss": 0.3896,
+      "step": 2363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.031056986833706e-08,
+      "loss": 0.4003,
+      "step": 2364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.886747668178538e-08,
+      "loss": 0.4039,
+      "step": 2365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.744180718255776e-08,
+      "loss": 0.3794,
+      "step": 2366
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6033563869404285e-08,
+      "loss": 0.3997,
+      "step": 2367
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4642749210535026e-08,
+      "loss": 0.3836,
+      "step": 2368
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.326936564361118e-08,
+      "loss": 0.3958,
+      "step": 2369
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.191341557574392e-08,
+      "loss": 0.397,
+      "step": 2370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0574901383487754e-08,
+      "loss": 0.3793,
+      "step": 2371
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.92538254128383e-08,
+      "loss": 0.4054,
+      "step": 2372
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7950189979227844e-08,
+      "loss": 0.3868,
+      "step": 2373
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.666399736751981e-08,
+      "loss": 0.4087,
+      "step": 2374
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.53952498320076e-08,
+      "loss": 0.391,
+      "step": 2375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.414394959640578e-08,
+      "loss": 0.4091,
+      "step": 2376
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.291009885385333e-08,
+      "loss": 0.393,
+      "step": 2377
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.169369976690263e-08,
+      "loss": 0.389,
+      "step": 2378
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0494754467518274e-08,
+      "loss": 0.4029,
+      "step": 2379
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.931326505707822e-08,
+      "loss": 0.4081,
+      "step": 2380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.814923360636158e-08,
+      "loss": 0.3857,
+      "step": 2381
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.700266215554971e-08,
+      "loss": 0.4101,
+      "step": 2382
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.587355271422288e-08,
+      "loss": 0.4016,
+      "step": 2383
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4761907261356976e-08,
+      "loss": 0.3978,
+      "step": 2384
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3667727745316794e-08,
+      "loss": 0.3861,
+      "step": 2385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2591016083856066e-08,
+      "loss": 0.4024,
+      "step": 2386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.153177416411191e-08,
+      "loss": 0.4051,
+      "step": 2387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.049000384260592e-08,
+      "loss": 0.4033,
+      "step": 2388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9465706945230876e-08,
+      "loss": 0.3739,
+      "step": 2389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8458885267260706e-08,
+      "loss": 0.3898,
+      "step": 2390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.746954057333606e-08,
+      "loss": 0.3973,
+      "step": 2391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6497674597469882e-08,
+      "loss": 0.3912,
+      "step": 2392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.554328904303738e-08,
+      "loss": 0.3781,
+      "step": 2393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.460638558277606e-08,
+      "loss": 0.3958,
+      "step": 2394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3686965858786824e-08,
+      "loss": 0.4095,
+      "step": 2395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2785031482521757e-08,
+      "loss": 0.3927,
+      "step": 2396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1900584034788565e-08,
+      "loss": 0.4106,
+      "step": 2397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1033625065747244e-08,
+      "loss": 0.4023,
+      "step": 2398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018415609490565e-08,
+      "loss": 0.4106,
+      "step": 2399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9352178611115045e-08,
+      "loss": 0.4147,
+      "step": 2400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.853769407257122e-08,
+      "loss": 0.3939,
+      "step": 2401
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7740703906810042e-08,
+      "loss": 0.3871,
+      "step": 2402
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6961209510707454e-08,
+      "loss": 0.3828,
+      "step": 2403
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.61992122504695e-08,
+      "loss": 0.4026,
+      "step": 2404
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.545471346164007e-08,
+      "loss": 0.3827,
+      "step": 2405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4727714449090936e-08,
+      "loss": 0.3964,
+      "step": 2406
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.401821648702506e-08,
+      "loss": 0.3856,
+      "step": 2407
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3326220818968839e-08,
+      "loss": 0.3915,
+      "step": 2408
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2651728657773198e-08,
+      "loss": 0.3954,
+      "step": 2409
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1994741185612501e-08,
+      "loss": 0.392,
+      "step": 2410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1355259553978981e-08,
+      "loss": 0.39,
+      "step": 2411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0733284883682748e-08,
+      "loss": 0.4095,
+      "step": 2412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0128818264851791e-08,
+      "loss": 0.3973,
+      "step": 2413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.541860756925315e-09,
+      "loss": 0.3962,
+      "step": 2414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.972413388657064e-09,
+      "loss": 0.3991,
+      "step": 2415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.420477158107565e-09,
+      "loss": 0.4001,
+      "step": 2416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.886053032649665e-09,
+      "loss": 0.3835,
+      "step": 2417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.369141948960768e-09,
+      "loss": 0.4017,
+      "step": 2418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.8697448130239374e-09,
+      "loss": 0.3963,
+      "step": 2419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.387862500125686e-09,
+      "loss": 0.3895,
+      "step": 2420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.923495854853745e-09,
+      "loss": 0.3846,
+      "step": 2421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.476645691098181e-09,
+      "loss": 0.3845,
+      "step": 2422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.047312792046955e-09,
+      "loss": 0.3995,
+      "step": 2423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.635497910187026e-09,
+      "loss": 0.3903,
+      "step": 2424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.241201767298808e-09,
+      "loss": 0.3906,
+      "step": 2425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8644250544594975e-09,
+      "loss": 0.4005,
+      "step": 2426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5051684320430712e-09,
+      "loss": 0.3988,
+      "step": 2427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.163432529711408e-09,
+      "loss": 0.4072,
+      "step": 2428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.839217946422057e-09,
+      "loss": 0.3803,
+      "step": 2429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.53252525042047e-09,
+      "loss": 0.3967,
+      "step": 2430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.243354979242218e-09,
+      "loss": 0.3943,
+      "step": 2431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9717076397129942e-09,
+      "loss": 0.3935,
+      "step": 2432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7175837079452806e-09,
+      "loss": 0.4102,
+      "step": 2433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4809836293394607e-09,
+      "loss": 0.4025,
+      "step": 2434
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2619078185793775e-09,
+      "loss": 0.3861,
+      "step": 2435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0603566596367743e-09,
+      "loss": 0.3964,
+      "step": 2436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.763305057690741e-10,
+      "loss": 0.4165,
+      "step": 2437
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.098296795138293e-10,
+      "loss": 0.3902,
+      "step": 2438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.608544726976028e-10,
+      "loss": 0.4069,
+      "step": 2439
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.294051464259763e-10,
+      "loss": 0.3905,
+      "step": 2440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1548193108688063e-10,
+      "loss": 0.3998,
+      "step": 2441
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1908502635503703e-10,
+      "loss": 0.4318,
+      "step": 2442
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4021460118085473e-10,
+      "loss": 0.3982,
+      "step": 2443
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.887079380153317e-11,
+      "loss": 0.3986,
+      "step": 2444
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.505371173329053e-11,
+      "loss": 0.3857,
+      "step": 2445
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.763431773584075e-12,
+      "loss": 0.4021,
+      "step": 2446
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.4082,
+      "step": 2447
+    },
+    {
+      "epoch": 1.0,
+      "step": 2447,
+      "total_flos": 3.083117523382567e+18,
+      "train_loss": 0.33612436193761214,
+      "train_runtime": 19890.5704,
+      "train_samples_per_second": 7.871,
+      "train_steps_per_second": 0.123
+    }
+  ],
+  "max_steps": 2447,
+  "num_train_epochs": 1,
+  "total_flos": 3.083117523382567e+18,
+  "trial_name": null,
+  "trial_params": null
+}