diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,90900 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999668994737017,
+  "eval_steps": 1000,
+  "global_step": 15105,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.044052863436123e-06,
+      "loss": 1.0012,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0881057268722468e-06,
+      "loss": 0.9422,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.13215859030837e-06,
+      "loss": 0.877,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.176211453744493e-06,
+      "loss": 0.9228,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.220264317180617e-06,
+      "loss": 0.9627,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.26431718061674e-06,
+      "loss": 1.0284,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3083700440528633e-06,
+      "loss": 0.9719,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.352422907488987e-06,
+      "loss": 1.0821,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.39647577092511e-06,
+      "loss": 0.9774,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4405286343612334e-06,
+      "loss": 0.8796,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.484581497797357e-06,
+      "loss": 0.8694,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5286343612334802e-06,
+      "loss": 0.7757,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5726872246696035e-06,
+      "loss": 0.8783,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.616740088105727e-06,
+      "loss": 0.8749,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6607929515418503e-06,
+      "loss": 0.8695,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7048458149779735e-06,
+      "loss": 0.8608,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7488986784140968e-06,
+      "loss": 0.7477,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7929515418502204e-06,
+      "loss": 0.9024,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8370044052863436e-06,
+      "loss": 0.9179,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.881057268722467e-06,
+      "loss": 0.9025,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.92511013215859e-06,
+      "loss": 0.8773,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9691629955947137e-06,
+      "loss": 0.8596,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.013215859030837e-06,
+      "loss": 0.8448,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.05726872246696e-06,
+      "loss": 0.841,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1013215859030838e-06,
+      "loss": 0.853,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.145374449339207e-06,
+      "loss": 0.93,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1894273127753302e-06,
+      "loss": 0.7478,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.233480176211454e-06,
+      "loss": 0.8652,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.277533039647577e-06,
+      "loss": 0.8377,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3215859030837003e-06,
+      "loss": 0.7629,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.365638766519824e-06,
+      "loss": 0.8316,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4096916299559476e-06,
+      "loss": 0.9251,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4537444933920704e-06,
+      "loss": 0.8401,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.497797356828194e-06,
+      "loss": 0.7998,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5418502202643172e-06,
+      "loss": 0.8827,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5859030837004405e-06,
+      "loss": 0.8577,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.629955947136564e-06,
+      "loss": 0.8366,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6740088105726873e-06,
+      "loss": 0.8656,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7180616740088105e-06,
+      "loss": 0.8099,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.762114537444934e-06,
+      "loss": 0.8663,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.806167400881057e-06,
+      "loss": 0.8217,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.850220264317181e-06,
+      "loss": 0.8128,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.894273127753304e-06,
+      "loss": 0.8511,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.938325991189428e-06,
+      "loss": 0.8402,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.982378854625551e-06,
+      "loss": 0.8146,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0264317180616735e-06,
+      "loss": 0.8893,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.070484581497797e-06,
+      "loss": 0.7561,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.114537444933921e-06,
+      "loss": 0.7531,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1585903083700444e-06,
+      "loss": 0.727,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.202643171806167e-06,
+      "loss": 0.7762,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.246696035242291e-06,
+      "loss": 0.7381,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.290748898678414e-06,
+      "loss": 0.8699,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.334801762114537e-06,
+      "loss": 0.7906,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.378854625550661e-06,
+      "loss": 0.8063,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.422907488986785e-06,
+      "loss": 0.844,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.466960352422908e-06,
+      "loss": 0.8,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.511013215859031e-06,
+      "loss": 0.7289,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.555066079295154e-06,
+      "loss": 0.8351,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5991189427312775e-06,
+      "loss": 0.7914,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.643171806167401e-06,
+      "loss": 0.8872,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.687224669603525e-06,
+      "loss": 0.761,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.731277533039648e-06,
+      "loss": 0.7488,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.775330396475771e-06,
+      "loss": 0.8498,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.819383259911895e-06,
+      "loss": 0.7855,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.863436123348018e-06,
+      "loss": 0.8369,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.907488986784141e-06,
+      "loss": 0.7783,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.951541850220265e-06,
+      "loss": 0.7256,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995594713656388e-06,
+      "loss": 0.7928,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.0396475770925105e-06,
+      "loss": 0.7609,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.083700440528634e-06,
+      "loss": 0.7664,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.127753303964758e-06,
+      "loss": 0.6855,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.1718061674008814e-06,
+      "loss": 0.7598,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.215859030837005e-06,
+      "loss": 0.8296,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.259911894273128e-06,
+      "loss": 0.8355,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.3039647577092515e-06,
+      "loss": 0.7983,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.348017621145374e-06,
+      "loss": 0.7743,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.392070484581498e-06,
+      "loss": 0.8814,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.436123348017622e-06,
+      "loss": 0.7939,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.480176211453744e-06,
+      "loss": 0.7686,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.524229074889868e-06,
+      "loss": 0.8077,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.568281938325991e-06,
+      "loss": 0.8831,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6123348017621145e-06,
+      "loss": 0.8205,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.656387665198238e-06,
+      "loss": 0.7211,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.700440528634362e-06,
+      "loss": 0.7793,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.744493392070485e-06,
+      "loss": 0.8217,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.788546255506608e-06,
+      "loss": 0.8181,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.832599118942731e-06,
+      "loss": 0.8635,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.876651982378855e-06,
+      "loss": 0.7728,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.920704845814978e-06,
+      "loss": 0.8454,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.964757709251102e-06,
+      "loss": 0.8146,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.0088105726872256e-06,
+      "loss": 0.7756,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.0528634361233475e-06,
+      "loss": 0.8364,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.096916299559471e-06,
+      "loss": 0.7595,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.140969162995595e-06,
+      "loss": 0.7088,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.1850220264317184e-06,
+      "loss": 0.7164,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.229074889867842e-06,
+      "loss": 0.7433,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.273127753303966e-06,
+      "loss": 0.8102,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.317180616740089e-06,
+      "loss": 0.8551,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.361233480176211e-06,
+      "loss": 0.8394,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.405286343612335e-06,
+      "loss": 0.8165,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.449339207048459e-06,
+      "loss": 0.8475,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.493392070484582e-06,
+      "loss": 0.8602,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.537444933920704e-06,
+      "loss": 0.8183,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.581497797356828e-06,
+      "loss": 0.797,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6255506607929515e-06,
+      "loss": 0.7536,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.669603524229075e-06,
+      "loss": 0.8275,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.713656387665199e-06,
+      "loss": 0.7812,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.757709251101322e-06,
+      "loss": 0.878,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.801762114537446e-06,
+      "loss": 0.7552,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.845814977973568e-06,
+      "loss": 0.8072,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.889867841409692e-06,
+      "loss": 0.7343,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.933920704845815e-06,
+      "loss": 0.7977,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.977973568281939e-06,
+      "loss": 0.7628,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0220264317180626e-06,
+      "loss": 0.7422,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.066079295154186e-06,
+      "loss": 0.7286,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.110132158590308e-06,
+      "loss": 0.8517,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.154185022026432e-06,
+      "loss": 0.8157,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.1982378854625554e-06,
+      "loss": 0.7897,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.242290748898679e-06,
+      "loss": 0.8179,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.286343612334803e-06,
+      "loss": 0.7894,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.330396475770926e-06,
+      "loss": 0.7677,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.374449339207048e-06,
+      "loss": 0.8183,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.418502202643172e-06,
+      "loss": 0.7886,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.462555066079296e-06,
+      "loss": 0.8681,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5066079295154176e-06,
+      "loss": 0.7815,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.550660792951543e-06,
+      "loss": 0.7741,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5947136563876665e-06,
+      "loss": 0.7475,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.63876651982379e-06,
+      "loss": 0.8722,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.682819383259912e-06,
+      "loss": 0.7957,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.726872246696036e-06,
+      "loss": 0.7405,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.77092511013216e-06,
+      "loss": 0.8789,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.814977973568281e-06,
+      "loss": 0.789,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.859030837004405e-06,
+      "loss": 0.7534,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.903083700440529e-06,
+      "loss": 0.7705,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.947136563876652e-06,
+      "loss": 0.7959,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.991189427312776e-06,
+      "loss": 0.7473,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.0352422907489e-06,
+      "loss": 0.7557,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.079295154185022e-06,
+      "loss": 0.7235,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.123348017621145e-06,
+      "loss": 0.7588,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.167400881057269e-06,
+      "loss": 0.8087,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.211453744493392e-06,
+      "loss": 0.7683,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.255506607929516e-06,
+      "loss": 0.8109,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.29955947136564e-06,
+      "loss": 0.7908,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.343612334801763e-06,
+      "loss": 0.7367,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.387665198237885e-06,
+      "loss": 0.7571,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.431718061674009e-06,
+      "loss": 0.7931,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.475770925110133e-06,
+      "loss": 0.8223,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.519823788546256e-06,
+      "loss": 0.7662,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.56387665198238e-06,
+      "loss": 0.7369,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.607929515418504e-06,
+      "loss": 0.7132,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.651982378854625e-06,
+      "loss": 0.8949,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.696035242290749e-06,
+      "loss": 0.874,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.740088105726873e-06,
+      "loss": 0.7503,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.784140969162996e-06,
+      "loss": 0.8024,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.828193832599118e-06,
+      "loss": 0.7166,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.872246696035244e-06,
+      "loss": 0.7802,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.916299559471367e-06,
+      "loss": 0.7412,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.96035242290749e-06,
+      "loss": 0.8302,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.004405286343613e-06,
+      "loss": 0.8212,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.048458149779737e-06,
+      "loss": 0.7559,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.092511013215859e-06,
+      "loss": 0.8057,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.136563876651982e-06,
+      "loss": 0.7655,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.180616740088106e-06,
+      "loss": 0.9226,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.22466960352423e-06,
+      "loss": 0.8592,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.268722466960353e-06,
+      "loss": 0.7385,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.312775330396477e-06,
+      "loss": 0.6949,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.3568281938326e-06,
+      "loss": 0.8062,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.400881057268722e-06,
+      "loss": 0.8028,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.444933920704846e-06,
+      "loss": 0.7977,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.48898678414097e-06,
+      "loss": 0.7521,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.533039647577093e-06,
+      "loss": 0.7965,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.577092511013217e-06,
+      "loss": 0.7642,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.62114537444934e-06,
+      "loss": 0.7937,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.665198237885462e-06,
+      "loss": 0.7425,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.709251101321586e-06,
+      "loss": 0.7377,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.75330396475771e-06,
+      "loss": 0.8116,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.797356828193832e-06,
+      "loss": 0.8043,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.841409691629955e-06,
+      "loss": 0.7822,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.885462555066079e-06,
+      "loss": 0.7907,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.929515418502203e-06,
+      "loss": 0.8517,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.973568281938326e-06,
+      "loss": 0.7805,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.001762114537445e-05,
+      "loss": 0.7345,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0061674008810574e-05,
+      "loss": 0.8502,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0105726872246696e-05,
+      "loss": 0.7323,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.014977973568282e-05,
+      "loss": 0.6981,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0193832599118943e-05,
+      "loss": 0.8584,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0237885462555066e-05,
+      "loss": 0.8213,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.028193832599119e-05,
+      "loss": 0.7568,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0325991189427314e-05,
+      "loss": 0.8178,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0370044052863436e-05,
+      "loss": 0.7966,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.041409691629956e-05,
+      "loss": 0.7838,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0458149779735683e-05,
+      "loss": 0.8155,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0502202643171807e-05,
+      "loss": 0.7182,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.054625550660793e-05,
+      "loss": 0.7943,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0590308370044054e-05,
+      "loss": 0.7599,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0634361233480178e-05,
+      "loss": 0.7609,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.06784140969163e-05,
+      "loss": 0.7557,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0722466960352423e-05,
+      "loss": 0.7978,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0766519823788547e-05,
+      "loss": 0.7885,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0810572687224669e-05,
+      "loss": 0.8399,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0854625550660792e-05,
+      "loss": 0.8176,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0898678414096916e-05,
+      "loss": 0.7805,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.094273127753304e-05,
+      "loss": 0.7524,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0986784140969163e-05,
+      "loss": 0.8317,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1030837004405287e-05,
+      "loss": 0.7712,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1074889867841409e-05,
+      "loss": 0.7419,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1118942731277533e-05,
+      "loss": 0.7362,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1162995594713656e-05,
+      "loss": 0.8627,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.120704845814978e-05,
+      "loss": 0.7028,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1251101321585903e-05,
+      "loss": 0.7633,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1295154185022027e-05,
+      "loss": 0.7864,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.133920704845815e-05,
+      "loss": 0.8099,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1383259911894273e-05,
+      "loss": 0.8558,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1427312775330396e-05,
+      "loss": 0.7526,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.147136563876652e-05,
+      "loss": 0.7369,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1515418502202644e-05,
+      "loss": 0.8272,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1559471365638767e-05,
+      "loss": 0.7403,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1603524229074891e-05,
+      "loss": 0.7781,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1647577092511013e-05,
+      "loss": 0.7183,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1691629955947136e-05,
+      "loss": 0.7694,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.173568281938326e-05,
+      "loss": 0.8305,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1779735682819384e-05,
+      "loss": 0.8188,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1823788546255507e-05,
+      "loss": 0.701,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1867841409691631e-05,
+      "loss": 0.8046,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1911894273127755e-05,
+      "loss": 0.796,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1955947136563877e-05,
+      "loss": 0.7426,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 0.7703,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2044052863436124e-05,
+      "loss": 0.7608,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2088105726872248e-05,
+      "loss": 0.7833,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2132158590308371e-05,
+      "loss": 0.8308,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2176211453744493e-05,
+      "loss": 0.7902,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2220264317180617e-05,
+      "loss": 0.7804,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.226431718061674e-05,
+      "loss": 0.7617,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2308370044052864e-05,
+      "loss": 0.8175,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2352422907488988e-05,
+      "loss": 0.8071,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2396475770925111e-05,
+      "loss": 0.831,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2440528634361233e-05,
+      "loss": 0.8141,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2484581497797357e-05,
+      "loss": 0.8099,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.252863436123348e-05,
+      "loss": 0.7497,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2572687224669604e-05,
+      "loss": 0.6906,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2616740088105728e-05,
+      "loss": 0.7278,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2660792951541852e-05,
+      "loss": 0.8136,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2704845814977975e-05,
+      "loss": 0.797,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2748898678414097e-05,
+      "loss": 0.7626,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.279295154185022e-05,
+      "loss": 0.7508,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2837004405286344e-05,
+      "loss": 0.8202,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2881057268722468e-05,
+      "loss": 0.8384,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2925110132158592e-05,
+      "loss": 0.7354,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2969162995594715e-05,
+      "loss": 0.7772,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3013215859030836e-05,
+      "loss": 0.8101,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.305726872246696e-05,
+      "loss": 0.8628,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3101321585903085e-05,
+      "loss": 0.7926,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3145374449339208e-05,
+      "loss": 0.7644,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3189427312775332e-05,
+      "loss": 0.8379,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3233480176211456e-05,
+      "loss": 0.7777,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.327753303964758e-05,
+      "loss": 0.7026,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.33215859030837e-05,
+      "loss": 0.7143,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3365638766519823e-05,
+      "loss": 0.8555,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3409691629955947e-05,
+      "loss": 0.7897,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.345374449339207e-05,
+      "loss": 0.7292,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3497797356828194e-05,
+      "loss": 0.785,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.354185022026432e-05,
+      "loss": 0.8268,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.358590308370044e-05,
+      "loss": 0.8531,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3629955947136563e-05,
+      "loss": 0.6988,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3674008810572687e-05,
+      "loss": 0.7517,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.371806167400881e-05,
+      "loss": 0.7827,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3762114537444934e-05,
+      "loss": 0.7457,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3806167400881058e-05,
+      "loss": 0.82,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3850220264317181e-05,
+      "loss": 0.7838,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3894273127753303e-05,
+      "loss": 0.7572,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3938325991189427e-05,
+      "loss": 0.7828,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.398237885462555e-05,
+      "loss": 0.7819,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4026431718061674e-05,
+      "loss": 0.7402,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4070484581497798e-05,
+      "loss": 0.8074,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4114537444933922e-05,
+      "loss": 0.7725,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4158590308370044e-05,
+      "loss": 0.8569,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4202643171806167e-05,
+      "loss": 0.7597,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.424669603524229e-05,
+      "loss": 0.7649,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4290748898678414e-05,
+      "loss": 0.7592,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4334801762114538e-05,
+      "loss": 0.768,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4378854625550662e-05,
+      "loss": 0.8304,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4422907488986785e-05,
+      "loss": 0.7704,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4466960352422907e-05,
+      "loss": 0.8018,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4511013215859031e-05,
+      "loss": 0.6958,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4555066079295155e-05,
+      "loss": 0.806,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4599118942731278e-05,
+      "loss": 0.7165,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4643171806167402e-05,
+      "loss": 0.8323,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4687224669603526e-05,
+      "loss": 0.8319,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4731277533039648e-05,
+      "loss": 0.7695,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4775330396475771e-05,
+      "loss": 0.7077,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4819383259911895e-05,
+      "loss": 0.8042,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4863436123348018e-05,
+      "loss": 0.7741,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4907488986784142e-05,
+      "loss": 0.7749,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4951541850220266e-05,
+      "loss": 0.861,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.499559471365639e-05,
+      "loss": 0.8001,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5039647577092511e-05,
+      "loss": 0.6838,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5083700440528635e-05,
+      "loss": 0.7296,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5127753303964759e-05,
+      "loss": 0.8137,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5171806167400882e-05,
+      "loss": 0.7604,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5215859030837006e-05,
+      "loss": 0.8569,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.525991189427313e-05,
+      "loss": 0.8279,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5303964757709253e-05,
+      "loss": 0.7332,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5348017621145377e-05,
+      "loss": 0.739,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.53920704845815e-05,
+      "loss": 0.8337,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5436123348017624e-05,
+      "loss": 0.7999,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5480176211453748e-05,
+      "loss": 0.7806,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.552422907488987e-05,
+      "loss": 0.7635,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5568281938325995e-05,
+      "loss": 0.7544,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5612334801762115e-05,
+      "loss": 0.7745,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.565638766519824e-05,
+      "loss": 0.769,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5700440528634363e-05,
+      "loss": 0.7617,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5744493392070486e-05,
+      "loss": 0.8339,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.578854625550661e-05,
+      "loss": 0.8515,
+      "step": 313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5832599118942733e-05,
+      "loss": 0.6619,
+      "step": 314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5876651982378854e-05,
+      "loss": 0.7053,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5920704845814977e-05,
+      "loss": 0.8404,
+      "step": 316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.59647577092511e-05,
+      "loss": 0.8436,
+      "step": 317
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6008810572687225e-05,
+      "loss": 0.8112,
+      "step": 318
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.605286343612335e-05,
+      "loss": 0.7865,
+      "step": 319
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6096916299559472e-05,
+      "loss": 0.7775,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6140969162995596e-05,
+      "loss": 0.7533,
+      "step": 321
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.618502202643172e-05,
+      "loss": 0.8076,
+      "step": 322
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6229074889867843e-05,
+      "loss": 0.7915,
+      "step": 323
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6273127753303967e-05,
+      "loss": 0.7872,
+      "step": 324
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.631718061674009e-05,
+      "loss": 0.7698,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6361233480176214e-05,
+      "loss": 0.7357,
+      "step": 326
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6405286343612337e-05,
+      "loss": 0.8215,
+      "step": 327
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6449339207048458e-05,
+      "loss": 0.8071,
+      "step": 328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.649339207048458e-05,
+      "loss": 0.8028,
+      "step": 329
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6537444933920705e-05,
+      "loss": 0.744,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.658149779735683e-05,
+      "loss": 0.7686,
+      "step": 331
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6625550660792952e-05,
+      "loss": 0.654,
+      "step": 332
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6669603524229076e-05,
+      "loss": 0.8494,
+      "step": 333
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.67136563876652e-05,
+      "loss": 0.7442,
+      "step": 334
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6757709251101323e-05,
+      "loss": 0.762,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6801762114537447e-05,
+      "loss": 0.7393,
+      "step": 336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.684581497797357e-05,
+      "loss": 0.7669,
+      "step": 337
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6889867841409694e-05,
+      "loss": 0.7957,
+      "step": 338
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6933920704845818e-05,
+      "loss": 0.7752,
+      "step": 339
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.697797356828194e-05,
+      "loss": 0.7844,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.702202643171806e-05,
+      "loss": 0.8526,
+      "step": 341
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7066079295154185e-05,
+      "loss": 0.7726,
+      "step": 342
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.711013215859031e-05,
+      "loss": 0.7316,
+      "step": 343
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7154185022026433e-05,
+      "loss": 0.7485,
+      "step": 344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7198237885462556e-05,
+      "loss": 0.7591,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.724229074889868e-05,
+      "loss": 0.724,
+      "step": 346
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7286343612334804e-05,
+      "loss": 0.7721,
+      "step": 347
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7330396475770927e-05,
+      "loss": 0.7892,
+      "step": 348
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.737444933920705e-05,
+      "loss": 0.7969,
+      "step": 349
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7418502202643174e-05,
+      "loss": 0.8073,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7462555066079298e-05,
+      "loss": 0.7931,
+      "step": 351
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7506607929515422e-05,
+      "loss": 0.76,
+      "step": 352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7550660792951545e-05,
+      "loss": 0.7815,
+      "step": 353
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7594713656387666e-05,
+      "loss": 0.7576,
+      "step": 354
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.763876651982379e-05,
+      "loss": 0.7179,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7682819383259913e-05,
+      "loss": 0.7643,
+      "step": 356
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7726872246696037e-05,
+      "loss": 0.7846,
+      "step": 357
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.777092511013216e-05,
+      "loss": 0.761,
+      "step": 358
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7814977973568284e-05,
+      "loss": 0.7687,
+      "step": 359
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7859030837004408e-05,
+      "loss": 0.8539,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.790308370044053e-05,
+      "loss": 0.7845,
+      "step": 361
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7947136563876655e-05,
+      "loss": 0.7252,
+      "step": 362
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.799118942731278e-05,
+      "loss": 0.8032,
+      "step": 363
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8035242290748902e-05,
+      "loss": 0.7861,
+      "step": 364
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8079295154185026e-05,
+      "loss": 0.674,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.812334801762115e-05,
+      "loss": 0.7299,
+      "step": 366
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.816740088105727e-05,
+      "loss": 0.7852,
+      "step": 367
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8211453744493393e-05,
+      "loss": 0.776,
+      "step": 368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8255506607929517e-05,
+      "loss": 0.7338,
+      "step": 369
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.829955947136564e-05,
+      "loss": 0.7631,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8343612334801764e-05,
+      "loss": 0.7598,
+      "step": 371
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8387665198237888e-05,
+      "loss": 0.7133,
+      "step": 372
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.843171806167401e-05,
+      "loss": 0.7129,
+      "step": 373
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8475770925110135e-05,
+      "loss": 0.756,
+      "step": 374
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.851982378854626e-05,
+      "loss": 0.7648,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8563876651982382e-05,
+      "loss": 0.6836,
+      "step": 376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8607929515418506e-05,
+      "loss": 0.8009,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.865198237885463e-05,
+      "loss": 0.7362,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8696035242290753e-05,
+      "loss": 0.7713,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8740088105726874e-05,
+      "loss": 0.7702,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8784140969162997e-05,
+      "loss": 0.8045,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.882819383259912e-05,
+      "loss": 0.7846,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8872246696035245e-05,
+      "loss": 0.8161,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8916299559471368e-05,
+      "loss": 0.8363,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8960352422907492e-05,
+      "loss": 0.7181,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9004405286343615e-05,
+      "loss": 0.7207,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.904845814977974e-05,
+      "loss": 0.8387,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9092511013215863e-05,
+      "loss": 0.8002,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9136563876651986e-05,
+      "loss": 0.7556,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.918061674008811e-05,
+      "loss": 0.7966,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9224669603524234e-05,
+      "loss": 0.7734,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9268722466960357e-05,
+      "loss": 0.7925,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9312775330396478e-05,
+      "loss": 0.8397,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.93568281938326e-05,
+      "loss": 0.769,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9400881057268725e-05,
+      "loss": 0.7373,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.944493392070485e-05,
+      "loss": 0.7529,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9488986784140972e-05,
+      "loss": 0.8076,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9533039647577096e-05,
+      "loss": 0.7122,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9577092511013216e-05,
+      "loss": 0.6996,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.962114537444934e-05,
+      "loss": 0.8942,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9665198237885463e-05,
+      "loss": 0.7384,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9709251101321587e-05,
+      "loss": 0.655,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.975330396475771e-05,
+      "loss": 0.7626,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9797356828193834e-05,
+      "loss": 0.8161,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.984140969162996e-05,
+      "loss": 0.7542,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.988546255506608e-05,
+      "loss": 0.7778,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9929515418502205e-05,
+      "loss": 0.8464,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.997356828193833e-05,
+      "loss": 0.7924,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0017621145374452e-05,
+      "loss": 0.708,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0061674008810576e-05,
+      "loss": 0.8645,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.01057268722467e-05,
+      "loss": 0.8032,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.014977973568282e-05,
+      "loss": 0.8304,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0193832599118944e-05,
+      "loss": 0.7822,
+      "step": 413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0237885462555067e-05,
+      "loss": 0.7441,
+      "step": 414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.028193832599119e-05,
+      "loss": 0.7936,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0325991189427315e-05,
+      "loss": 0.7309,
+      "step": 416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0370044052863438e-05,
+      "loss": 0.7406,
+      "step": 417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0414096916299562e-05,
+      "loss": 0.7119,
+      "step": 418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0458149779735685e-05,
+      "loss": 0.7216,
+      "step": 419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.050220264317181e-05,
+      "loss": 0.7961,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0546255506607933e-05,
+      "loss": 0.7121,
+      "step": 421
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0590308370044056e-05,
+      "loss": 0.8234,
+      "step": 422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.063436123348018e-05,
+      "loss": 0.7365,
+      "step": 423
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0678414096916304e-05,
+      "loss": 0.8105,
+      "step": 424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0722466960352424e-05,
+      "loss": 0.8111,
+      "step": 425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0766519823788548e-05,
+      "loss": 0.7937,
+      "step": 426
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.081057268722467e-05,
+      "loss": 0.7575,
+      "step": 427
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0854625550660795e-05,
+      "loss": 0.8419,
+      "step": 428
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.089867841409692e-05,
+      "loss": 0.6707,
+      "step": 429
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0942731277533042e-05,
+      "loss": 0.7277,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0986784140969166e-05,
+      "loss": 0.7677,
+      "step": 431
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.103083700440529e-05,
+      "loss": 0.7881,
+      "step": 432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1074889867841413e-05,
+      "loss": 0.7607,
+      "step": 433
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1118942731277537e-05,
+      "loss": 0.7285,
+      "step": 434
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.116299559471366e-05,
+      "loss": 0.7832,
+      "step": 435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1207048458149784e-05,
+      "loss": 0.7887,
+      "step": 436
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1251101321585908e-05,
+      "loss": 0.8294,
+      "step": 437
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1295154185022028e-05,
+      "loss": 0.7653,
+      "step": 438
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.133920704845815e-05,
+      "loss": 0.6828,
+      "step": 439
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1383259911894275e-05,
+      "loss": 0.7497,
+      "step": 440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.14273127753304e-05,
+      "loss": 0.7487,
+      "step": 441
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1471365638766522e-05,
+      "loss": 0.731,
+      "step": 442
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1515418502202646e-05,
+      "loss": 0.7809,
+      "step": 443
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.155947136563877e-05,
+      "loss": 0.6759,
+      "step": 444
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1603524229074893e-05,
+      "loss": 0.6841,
+      "step": 445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1647577092511017e-05,
+      "loss": 0.7691,
+      "step": 446
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.169162995594714e-05,
+      "loss": 0.7557,
+      "step": 447
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1735682819383264e-05,
+      "loss": 0.7344,
+      "step": 448
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1779735682819388e-05,
+      "loss": 0.8004,
+      "step": 449
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.182378854625551e-05,
+      "loss": 0.7641,
+      "step": 450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1867841409691632e-05,
+      "loss": 0.7465,
+      "step": 451
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1911894273127756e-05,
+      "loss": 0.7943,
+      "step": 452
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.195594713656388e-05,
+      "loss": 0.7616,
+      "step": 453
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.7655,
+      "step": 454
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999999770101993e-05,
+      "loss": 0.714,
+      "step": 455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999999080407974e-05,
+      "loss": 0.806,
+      "step": 456
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199999793091798e-05,
+      "loss": 0.7338,
+      "step": 457
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199999632163206e-05,
+      "loss": 0.7612,
+      "step": 458
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199999425255029e-05,
+      "loss": 0.7513,
+      "step": 459
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999991723672764e-05,
+      "loss": 0.768,
+      "step": 460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999988734999602e-05,
+      "loss": 0.7648,
+      "step": 461
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999985286530936e-05,
+      "loss": 0.8096,
+      "step": 462
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199998137826693e-05,
+      "loss": 0.6628,
+      "step": 463
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199997701020776e-05,
+      "loss": 0.7017,
+      "step": 464
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999972182353626e-05,
+      "loss": 0.7522,
+      "step": 465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999966894704754e-05,
+      "loss": 0.7682,
+      "step": 466
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999961147261383e-05,
+      "loss": 0.758,
+      "step": 467
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999954940023782e-05,
+      "loss": 0.8038,
+      "step": 468
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999948272992228e-05,
+      "loss": 0.7601,
+      "step": 469
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199994114616704e-05,
+      "loss": 0.6908,
+      "step": 470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999933559548535e-05,
+      "loss": 0.8163,
+      "step": 471
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999925513137063e-05,
+      "loss": 0.7719,
+      "step": 472
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999917006933e-05,
+      "loss": 0.8697,
+      "step": 473
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999908040936734e-05,
+      "loss": 0.7114,
+      "step": 474
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999898615148676e-05,
+      "loss": 0.7422,
+      "step": 475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999888729569264e-05,
+      "loss": 0.824,
+      "step": 476
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199987838419894e-05,
+      "loss": 0.7288,
+      "step": 477
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999867579038196e-05,
+      "loss": 0.7673,
+      "step": 478
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199985631408752e-05,
+      "loss": 0.7191,
+      "step": 479
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199984458934743e-05,
+      "loss": 0.7264,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199983240481847e-05,
+      "loss": 0.7542,
+      "step": 481
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999819760501194e-05,
+      "loss": 0.7198,
+      "step": 482
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999806656396184e-05,
+      "loss": 0.7253,
+      "step": 483
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999793092504048e-05,
+      "loss": 0.8518,
+      "step": 484
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999779068825403e-05,
+      "loss": 0.8117,
+      "step": 485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.19997645853609e-05,
+      "loss": 0.7603,
+      "step": 486
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.19997496421112e-05,
+      "loss": 0.7836,
+      "step": 487
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999734239076993e-05,
+      "loss": 0.8302,
+      "step": 488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999718376258986e-05,
+      "loss": 0.7865,
+      "step": 489
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199970205365791e-05,
+      "loss": 0.7509,
+      "step": 490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999685271274513e-05,
+      "loss": 0.7607,
+      "step": 491
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999668029109566e-05,
+      "loss": 0.7462,
+      "step": 492
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999650327163866e-05,
+      "loss": 0.7707,
+      "step": 493
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999632165438223e-05,
+      "loss": 0.8438,
+      "step": 494
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999613543933473e-05,
+      "loss": 0.6845,
+      "step": 495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999594462650475e-05,
+      "loss": 0.7216,
+      "step": 496
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999574921590106e-05,
+      "loss": 0.8073,
+      "step": 497
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999554920753256e-05,
+      "loss": 0.7344,
+      "step": 498
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999534460140858e-05,
+      "loss": 0.8231,
+      "step": 499
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999513539753843e-05,
+      "loss": 0.7789,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999492159593177e-05,
+      "loss": 0.7575,
+      "step": 501
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199947031965984e-05,
+      "loss": 0.7573,
+      "step": 502
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199944801995484e-05,
+      "loss": 0.7565,
+      "step": 503
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.19994252604792e-05,
+      "loss": 0.7428,
+      "step": 504
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999402041233967e-05,
+      "loss": 0.6994,
+      "step": 505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999378362220207e-05,
+      "loss": 0.7625,
+      "step": 506
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999354223439013e-05,
+      "loss": 0.7235,
+      "step": 507
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999329624891493e-05,
+      "loss": 0.7446,
+      "step": 508
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999304566578775e-05,
+      "loss": 0.7732,
+      "step": 509
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999279048502015e-05,
+      "loss": 0.7546,
+      "step": 510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199925307066238e-05,
+      "loss": 0.7166,
+      "step": 511
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199922663306108e-05,
+      "loss": 0.7152,
+      "step": 512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199919973569931e-05,
+      "loss": 0.7319,
+      "step": 513
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999172378578317e-05,
+      "loss": 0.8048,
+      "step": 514
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999144561699363e-05,
+      "loss": 0.7574,
+      "step": 515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999116285063722e-05,
+      "loss": 0.7685,
+      "step": 516
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199908754867269e-05,
+      "loss": 0.8529,
+      "step": 517
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999058352527597e-05,
+      "loss": 0.7776,
+      "step": 518
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199902869662978e-05,
+      "loss": 0.8078,
+      "step": 519
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998998580980605e-05,
+      "loss": 0.7469,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998968005581456e-05,
+      "loss": 0.7313,
+      "step": 521
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998936970433734e-05,
+      "loss": 0.797,
+      "step": 522
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998905475538876e-05,
+      "loss": 0.8266,
+      "step": 523
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199887352089832e-05,
+      "loss": 0.74,
+      "step": 524
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199884110651354e-05,
+      "loss": 0.8408,
+      "step": 525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199880823238603e-05,
+      "loss": 0.7451,
+      "step": 526
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998774898517293e-05,
+      "loss": 0.8036,
+      "step": 527
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199874110490887e-05,
+      "loss": 0.816,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998706851562313e-05,
+      "loss": 0.7769,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998672138479187e-05,
+      "loss": 0.8163,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998636965661102e-05,
+      "loss": 0.7606,
+      "step": 531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998601333109668e-05,
+      "loss": 0.7248,
+      "step": 532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199856524082653e-05,
+      "loss": 0.7619,
+      "step": 533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998528688813335e-05,
+      "loss": 0.6746,
+      "step": 534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998491677071776e-05,
+      "loss": 0.6844,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199845420560355e-05,
+      "loss": 0.7709,
+      "step": 536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998416274410377e-05,
+      "loss": 0.7979,
+      "step": 537
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998377883494002e-05,
+      "loss": 0.752,
+      "step": 538
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998339032856197e-05,
+      "loss": 0.7286,
+      "step": 539
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998299722498738e-05,
+      "loss": 0.8097,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998259952423443e-05,
+      "loss": 0.7672,
+      "step": 541
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998219722632133e-05,
+      "loss": 0.7516,
+      "step": 542
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199817903312666e-05,
+      "loss": 0.7831,
+      "step": 543
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998137883908892e-05,
+      "loss": 0.7678,
+      "step": 544
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199809627498073e-05,
+      "loss": 0.7966,
+      "step": 545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199805420634408e-05,
+      "loss": 0.751,
+      "step": 546
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998011678000877e-05,
+      "loss": 0.7881,
+      "step": 547
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997968689953073e-05,
+      "loss": 0.7611,
+      "step": 548
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997925242202654e-05,
+      "loss": 0.7519,
+      "step": 549
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997881334751608e-05,
+      "loss": 0.7322,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997836967601958e-05,
+      "loss": 0.7472,
+      "step": 551
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997792140755747e-05,
+      "loss": 0.7869,
+      "step": 552
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997746854215035e-05,
+      "loss": 0.7017,
+      "step": 553
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19977011079819e-05,
+      "loss": 0.7124,
+      "step": 554
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997654902058446e-05,
+      "loss": 0.6657,
+      "step": 555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19976082364468e-05,
+      "loss": 0.7912,
+      "step": 556
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997561111149108e-05,
+      "loss": 0.8064,
+      "step": 557
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997513526167536e-05,
+      "loss": 0.7007,
+      "step": 558
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997465481504273e-05,
+      "loss": 0.737,
+      "step": 559
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997416977161526e-05,
+      "loss": 0.7795,
+      "step": 560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997368013141523e-05,
+      "loss": 0.7833,
+      "step": 561
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199731858944652e-05,
+      "loss": 0.7984,
+      "step": 562
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199726870607879e-05,
+      "loss": 0.7884,
+      "step": 563
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997218363040624e-05,
+      "loss": 0.7817,
+      "step": 564
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997167560334338e-05,
+      "loss": 0.7199,
+      "step": 565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997116297962264e-05,
+      "loss": 0.7309,
+      "step": 566
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997064575926762e-05,
+      "loss": 0.6765,
+      "step": 567
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997012394230214e-05,
+      "loss": 0.8322,
+      "step": 568
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996959752875014e-05,
+      "loss": 0.7375,
+      "step": 569
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996906651863586e-05,
+      "loss": 0.721,
+      "step": 570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996853091198365e-05,
+      "loss": 0.7118,
+      "step": 571
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199679907088182e-05,
+      "loss": 0.7423,
+      "step": 572
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996744590916435e-05,
+      "loss": 0.7754,
+      "step": 573
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996689651304713e-05,
+      "loss": 0.7242,
+      "step": 574
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199663425204918e-05,
+      "loss": 0.7876,
+      "step": 575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199657839315238e-05,
+      "loss": 0.8052,
+      "step": 576
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199652207461689e-05,
+      "loss": 0.8278,
+      "step": 577
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199646529644529e-05,
+      "loss": 0.677,
+      "step": 578
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19964080586402e-05,
+      "loss": 0.8365,
+      "step": 579
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996350361204242e-05,
+      "loss": 0.7428,
+      "step": 580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996292204140076e-05,
+      "loss": 0.6982,
+      "step": 581
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199623358745037e-05,
+      "loss": 0.8194,
+      "step": 582
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996174511137827e-05,
+      "loss": 0.6948,
+      "step": 583
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199611497520516e-05,
+      "loss": 0.69,
+      "step": 584
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996054979655103e-05,
+      "loss": 0.7804,
+      "step": 585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199599452449042e-05,
+      "loss": 0.7534,
+      "step": 586
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995933609713886e-05,
+      "loss": 0.8424,
+      "step": 587
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19958722353283e-05,
+      "loss": 0.6533,
+      "step": 588
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995810401336493e-05,
+      "loss": 0.7219,
+      "step": 589
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19957481077413e-05,
+      "loss": 0.7305,
+      "step": 590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995685354545592e-05,
+      "loss": 0.7757,
+      "step": 591
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995622141752245e-05,
+      "loss": 0.7715,
+      "step": 592
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995558469364173e-05,
+      "loss": 0.7527,
+      "step": 593
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995494337384302e-05,
+      "loss": 0.8404,
+      "step": 594
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995429745815582e-05,
+      "loss": 0.7436,
+      "step": 595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995364694660977e-05,
+      "loss": 0.7896,
+      "step": 596
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995299183923487e-05,
+      "loss": 0.7153,
+      "step": 597
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995233213606116e-05,
+      "loss": 0.7765,
+      "step": 598
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19951667837119e-05,
+      "loss": 0.7595,
+      "step": 599
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995099894243897e-05,
+      "loss": 0.7242,
+      "step": 600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199503254520518e-05,
+      "loss": 0.715,
+      "step": 601
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199496473659884e-05,
+      "loss": 0.7727,
+      "step": 602
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994896468428004e-05,
+      "loss": 0.7919,
+      "step": 603
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199482774069581e-05,
+      "loss": 0.7266,
+      "step": 604
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994758553405412e-05,
+      "loss": 0.739,
+      "step": 605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994688906559992e-05,
+      "loss": 0.809,
+      "step": 606
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199461880016276e-05,
+      "loss": 0.7667,
+      "step": 607
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994548234216928e-05,
+      "loss": 0.7673,
+      "step": 608
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994477208725752e-05,
+      "loss": 0.7436,
+      "step": 609
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199440572369249e-05,
+      "loss": 0.7643,
+      "step": 610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199433377912043e-05,
+      "loss": 0.7374,
+      "step": 611
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994261375012884e-05,
+      "loss": 0.789,
+      "step": 612
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994188511373178e-05,
+      "loss": 0.7072,
+      "step": 613
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994115188204664e-05,
+      "loss": 0.8557,
+      "step": 614
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994041405510707e-05,
+      "loss": 0.7143,
+      "step": 615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993967163294708e-05,
+      "loss": 0.7392,
+      "step": 616
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199389246156008e-05,
+      "loss": 0.766,
+      "step": 617
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199381730031025e-05,
+      "loss": 0.6921,
+      "step": 618
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993741679548677e-05,
+      "loss": 0.773,
+      "step": 619
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993665599278846e-05,
+      "loss": 0.8464,
+      "step": 620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993589059504243e-05,
+      "loss": 0.9339,
+      "step": 621
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993512060228398e-05,
+      "loss": 0.7492,
+      "step": 622
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993434601454845e-05,
+      "loss": 0.827,
+      "step": 623
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993356683187148e-05,
+      "loss": 0.7785,
+      "step": 624
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993278305428884e-05,
+      "loss": 0.7967,
+      "step": 625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993199468183667e-05,
+      "loss": 0.7323,
+      "step": 626
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993120171455114e-05,
+      "loss": 0.7408,
+      "step": 627
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993040415246874e-05,
+      "loss": 0.6866,
+      "step": 628
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992960199562614e-05,
+      "loss": 0.7346,
+      "step": 629
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992879524406022e-05,
+      "loss": 0.7159,
+      "step": 630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992798389780807e-05,
+      "loss": 0.7949,
+      "step": 631
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19927167956907e-05,
+      "loss": 0.756,
+      "step": 632
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992634742139453e-05,
+      "loss": 0.7481,
+      "step": 633
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992552229130837e-05,
+      "loss": 0.829,
+      "step": 634
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199246925666865e-05,
+      "loss": 0.7501,
+      "step": 635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19923858247567e-05,
+      "loss": 0.714,
+      "step": 636
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199230193339883e-05,
+      "loss": 0.7911,
+      "step": 637
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992217582598896e-05,
+      "loss": 0.789,
+      "step": 638
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992132772360772e-05,
+      "loss": 0.713,
+      "step": 639
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992047502688363e-05,
+      "loss": 0.8124,
+      "step": 640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991961773585588e-05,
+      "loss": 0.7981,
+      "step": 641
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991875585056387e-05,
+      "loss": 0.7795,
+      "step": 642
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991788937104726e-05,
+      "loss": 0.8049,
+      "step": 643
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991701829734584e-05,
+      "loss": 0.7341,
+      "step": 644
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199161426294997e-05,
+      "loss": 0.752,
+      "step": 645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199152623675491e-05,
+      "loss": 0.787,
+      "step": 646
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199143775115345e-05,
+      "loss": 0.7707,
+      "step": 647
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991348806149664e-05,
+      "loss": 0.7178,
+      "step": 648
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199125940174763e-05,
+      "loss": 0.7409,
+      "step": 649
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199116953795147e-05,
+      "loss": 0.7734,
+      "step": 650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199107921476531e-05,
+      "loss": 0.8119,
+      "step": 651
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199098843219331e-05,
+      "loss": 0.6971,
+      "step": 652
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990897190239634e-05,
+      "loss": 0.7918,
+      "step": 653
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990805488908484e-05,
+      "loss": 0.7371,
+      "step": 654
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990713328204073e-05,
+      "loss": 0.7211,
+      "step": 655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990620708130644e-05,
+      "loss": 0.7967,
+      "step": 656
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990527628692447e-05,
+      "loss": 0.7736,
+      "step": 657
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990434089893772e-05,
+      "loss": 0.7751,
+      "step": 658
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199034009173891e-05,
+      "loss": 0.7225,
+      "step": 659
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199024563423219e-05,
+      "loss": 0.749,
+      "step": 660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199015071737795e-05,
+      "loss": 0.8245,
+      "step": 661
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990055341180556e-05,
+      "loss": 0.7843,
+      "step": 662
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989959505644398e-05,
+      "loss": 0.7151,
+      "step": 663
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198986321077388e-05,
+      "loss": 0.7215,
+      "step": 664
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989766456573423e-05,
+      "loss": 0.7515,
+      "step": 665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989669243047484e-05,
+      "loss": 0.8103,
+      "step": 666
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989571570200528e-05,
+      "loss": 0.734,
+      "step": 667
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198947343803705e-05,
+      "loss": 0.7703,
+      "step": 668
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989374846561555e-05,
+      "loss": 0.7685,
+      "step": 669
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198927579577859e-05,
+      "loss": 0.7623,
+      "step": 670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198917628569269e-05,
+      "loss": 0.6574,
+      "step": 671
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198907631630845e-05,
+      "loss": 0.7383,
+      "step": 672
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198897588763045e-05,
+      "loss": 0.7535,
+      "step": 673
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988874999663322e-05,
+      "loss": 0.663,
+      "step": 674
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988773652411693e-05,
+      "loss": 0.7332,
+      "step": 675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988671845880226e-05,
+      "loss": 0.8099,
+      "step": 676
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988569580073607e-05,
+      "loss": 0.7585,
+      "step": 677
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988466854996536e-05,
+      "loss": 0.7244,
+      "step": 678
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988363670653735e-05,
+      "loss": 0.7044,
+      "step": 679
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1988260027049944e-05,
+      "loss": 0.7347,
+      "step": 680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1988155924189938e-05,
+      "loss": 0.729,
+      "step": 681
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1988051362078494e-05,
+      "loss": 0.7341,
+      "step": 682
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987946340720425e-05,
+      "loss": 0.7133,
+      "step": 683
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987840860120564e-05,
+      "loss": 0.7868,
+      "step": 684
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987734920283753e-05,
+      "loss": 0.7241,
+      "step": 685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987628521214866e-05,
+      "loss": 0.7598,
+      "step": 686
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987521662918796e-05,
+      "loss": 0.8038,
+      "step": 687
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987414345400453e-05,
+      "loss": 0.7143,
+      "step": 688
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198730656866478e-05,
+      "loss": 0.6699,
+      "step": 689
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198719833271672e-05,
+      "loss": 0.7405,
+      "step": 690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198708963756126e-05,
+      "loss": 0.7691,
+      "step": 691
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986980483203396e-05,
+      "loss": 0.7444,
+      "step": 692
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986870869648143e-05,
+      "loss": 0.6604,
+      "step": 693
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198676079690054e-05,
+      "loss": 0.7492,
+      "step": 694
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986650264965652e-05,
+      "loss": 0.7964,
+      "step": 695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986539273848564e-05,
+      "loss": 0.8054,
+      "step": 696
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198642782355437e-05,
+      "loss": 0.8001,
+      "step": 697
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986315914088203e-05,
+      "loss": 0.7243,
+      "step": 698
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986203545455206e-05,
+      "loss": 0.7518,
+      "step": 699
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986090717660545e-05,
+      "loss": 0.8156,
+      "step": 700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198597743070941e-05,
+      "loss": 0.8309,
+      "step": 701
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985863684607005e-05,
+      "loss": 0.7944,
+      "step": 702
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985749479358563e-05,
+      "loss": 0.8178,
+      "step": 703
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985634814969336e-05,
+      "loss": 0.7169,
+      "step": 704
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985519691444593e-05,
+      "loss": 0.7211,
+      "step": 705
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198540410878963e-05,
+      "loss": 0.7564,
+      "step": 706
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985288067009764e-05,
+      "loss": 0.7414,
+      "step": 707
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985171566110325e-05,
+      "loss": 0.8795,
+      "step": 708
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985054606096673e-05,
+      "loss": 0.7923,
+      "step": 709
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984937186974186e-05,
+      "loss": 0.7213,
+      "step": 710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984819308748258e-05,
+      "loss": 0.7299,
+      "step": 711
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984700971424316e-05,
+      "loss": 0.8615,
+      "step": 712
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19845821750078e-05,
+      "loss": 0.7891,
+      "step": 713
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984462919504167e-05,
+      "loss": 0.7853,
+      "step": 714
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984343204918903e-05,
+      "loss": 0.7635,
+      "step": 715
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984223031257516e-05,
+      "loss": 0.7164,
+      "step": 716
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984102398525526e-05,
+      "loss": 0.732,
+      "step": 717
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983981306728482e-05,
+      "loss": 0.7694,
+      "step": 718
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198385975587195e-05,
+      "loss": 0.799,
+      "step": 719
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983737745961526e-05,
+      "loss": 0.7981,
+      "step": 720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198361527700281e-05,
+      "loss": 0.6878,
+      "step": 721
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983492349001442e-05,
+      "loss": 0.7202,
+      "step": 722
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983368961963066e-05,
+      "loss": 0.8217,
+      "step": 723
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198324511589336e-05,
+      "loss": 0.7854,
+      "step": 724
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198312081079802e-05,
+      "loss": 0.7204,
+      "step": 725
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982996046682756e-05,
+      "loss": 0.7668,
+      "step": 726
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198287082355331e-05,
+      "loss": 0.7572,
+      "step": 727
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198274514141544e-05,
+      "loss": 0.778,
+      "step": 728
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982619000274917e-05,
+      "loss": 0.8327,
+      "step": 729
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982492400137548e-05,
+      "loss": 0.7225,
+      "step": 730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198236534100915e-05,
+      "loss": 0.8046,
+      "step": 731
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198223782289557e-05,
+      "loss": 0.7954,
+      "step": 732
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982109845802668e-05,
+      "loss": 0.7185,
+      "step": 733
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198198140973633e-05,
+      "loss": 0.7129,
+      "step": 734
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198185251470246e-05,
+      "loss": 0.6874,
+      "step": 735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198172316070698e-05,
+      "loss": 0.8473,
+      "step": 736
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981593347755852e-05,
+      "loss": 0.7766,
+      "step": 737
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981463075855028e-05,
+      "loss": 0.7864,
+      "step": 738
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981332345010506e-05,
+      "loss": 0.6867,
+      "step": 739
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19812011552283e-05,
+      "loss": 0.8207,
+      "step": 740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981069506514434e-05,
+      "loss": 0.7747,
+      "step": 741
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980937398874967e-05,
+      "loss": 0.7665,
+      "step": 742
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980804832315972e-05,
+      "loss": 0.7858,
+      "step": 743
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198067180684354e-05,
+      "loss": 0.7726,
+      "step": 744
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980538322463794e-05,
+      "loss": 0.7429,
+      "step": 745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198040437918287e-05,
+      "loss": 0.7313,
+      "step": 746
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980269977006923e-05,
+      "loss": 0.7401,
+      "step": 747
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980135115942136e-05,
+      "loss": 0.7241,
+      "step": 748
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979999795994713e-05,
+      "loss": 0.776,
+      "step": 749
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979864017170867e-05,
+      "loss": 0.7384,
+      "step": 750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197972777947685e-05,
+      "loss": 0.7234,
+      "step": 751
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197959108291892e-05,
+      "loss": 0.8134,
+      "step": 752
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979453927503368e-05,
+      "loss": 0.7663,
+      "step": 753
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979316313236492e-05,
+      "loss": 0.7498,
+      "step": 754
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979178240124626e-05,
+      "loss": 0.7552,
+      "step": 755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197903970817412e-05,
+      "loss": 0.7987,
+      "step": 756
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978900717391337e-05,
+      "loss": 0.7343,
+      "step": 757
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978761267782673e-05,
+      "loss": 0.7311,
+      "step": 758
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197862135935454e-05,
+      "loss": 0.779,
+      "step": 759
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978480992113365e-05,
+      "loss": 0.7663,
+      "step": 760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197834016606561e-05,
+      "loss": 0.7971,
+      "step": 761
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197819888121774e-05,
+      "loss": 0.8198,
+      "step": 762
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197805713757626e-05,
+      "loss": 0.7855,
+      "step": 763
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197791493514769e-05,
+      "loss": 0.8288,
+      "step": 764
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977772273938558e-05,
+      "loss": 0.7086,
+      "step": 765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197762915395543e-05,
+      "loss": 0.7508,
+      "step": 766
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197748557520489e-05,
+      "loss": 0.8826,
+      "step": 767
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197734153769353e-05,
+      "loss": 0.7063,
+      "step": 768
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977197041427977e-05,
+      "loss": 0.7197,
+      "step": 769
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977052086414876e-05,
+      "loss": 0.7859,
+      "step": 770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976906672660894e-05,
+      "loss": 0.823,
+      "step": 771
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976760800172716e-05,
+      "loss": 0.798,
+      "step": 772
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197661446895704e-05,
+      "loss": 0.7834,
+      "step": 773
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197646767902061e-05,
+      "loss": 0.8101,
+      "step": 774
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976320430370166e-05,
+      "loss": 0.8618,
+      "step": 775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976172723012482e-05,
+      "loss": 0.7586,
+      "step": 776
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197602455695434e-05,
+      "loss": 0.7262,
+      "step": 777
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975875932202567e-05,
+      "loss": 0.7614,
+      "step": 778
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975726848763986e-05,
+      "loss": 0.7594,
+      "step": 779
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975577306645457e-05,
+      "loss": 0.7854,
+      "step": 780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975427305853853e-05,
+      "loss": 0.7931,
+      "step": 781
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975276846396074e-05,
+      "loss": 0.8465,
+      "step": 782
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975125928279034e-05,
+      "loss": 0.8242,
+      "step": 783
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197497455150968e-05,
+      "loss": 0.7595,
+      "step": 784
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974822716094964e-05,
+      "loss": 0.7563,
+      "step": 785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974670422041867e-05,
+      "loss": 0.787,
+      "step": 786
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974517669357396e-05,
+      "loss": 0.799,
+      "step": 787
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974364458048576e-05,
+      "loss": 0.7929,
+      "step": 788
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974210788122447e-05,
+      "loss": 0.7357,
+      "step": 789
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974056659586077e-05,
+      "loss": 0.7935,
+      "step": 790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973902072446554e-05,
+      "loss": 0.6834,
+      "step": 791
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197374702671098e-05,
+      "loss": 0.8046,
+      "step": 792
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197359152238649e-05,
+      "loss": 0.7263,
+      "step": 793
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197343555948023e-05,
+      "loss": 0.7322,
+      "step": 794
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973279137999378e-05,
+      "loss": 0.7195,
+      "step": 795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197312225795112e-05,
+      "loss": 0.8052,
+      "step": 796
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972964919342665e-05,
+      "loss": 0.7456,
+      "step": 797
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972807122181257e-05,
+      "loss": 0.802,
+      "step": 798
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972648866474147e-05,
+      "loss": 0.6868,
+      "step": 799
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197249015222861e-05,
+      "loss": 0.7462,
+      "step": 800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197233097945195e-05,
+      "loss": 0.7721,
+      "step": 801
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972171348151477e-05,
+      "loss": 0.8275,
+      "step": 802
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972011258334538e-05,
+      "loss": 0.7484,
+      "step": 803
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197185071000849e-05,
+      "loss": 0.7533,
+      "step": 804
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971689703180717e-05,
+      "loss": 0.7034,
+      "step": 805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197152823785862e-05,
+      "loss": 0.7444,
+      "step": 806
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971366314049623e-05,
+      "loss": 0.7581,
+      "step": 807
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197120393176117e-05,
+      "loss": 0.8143,
+      "step": 808
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971041091000736e-05,
+      "loss": 0.8689,
+      "step": 809
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970877791775798e-05,
+      "loss": 0.7128,
+      "step": 810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970714034093867e-05,
+      "loss": 0.7513,
+      "step": 811
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970549817962475e-05,
+      "loss": 0.7473,
+      "step": 812
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197038514338917e-05,
+      "loss": 0.8023,
+      "step": 813
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970220010381527e-05,
+      "loss": 0.76,
+      "step": 814
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970054418947137e-05,
+      "loss": 0.7784,
+      "step": 815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969888369093612e-05,
+      "loss": 0.6987,
+      "step": 816
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196972186082859e-05,
+      "loss": 0.7943,
+      "step": 817
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969554894159724e-05,
+      "loss": 0.8163,
+      "step": 818
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969387469094694e-05,
+      "loss": 0.7592,
+      "step": 819
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969219585641197e-05,
+      "loss": 0.817,
+      "step": 820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196905124380695e-05,
+      "loss": 0.7237,
+      "step": 821
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968882443599696e-05,
+      "loss": 0.7864,
+      "step": 822
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968713185027196e-05,
+      "loss": 0.7312,
+      "step": 823
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968543468097233e-05,
+      "loss": 0.789,
+      "step": 824
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196837329281761e-05,
+      "loss": 0.7244,
+      "step": 825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968202659196147e-05,
+      "loss": 0.8464,
+      "step": 826
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968031567240697e-05,
+      "loss": 0.702,
+      "step": 827
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196786001695912e-05,
+      "loss": 0.7938,
+      "step": 828
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1967688008359314e-05,
+      "loss": 0.7006,
+      "step": 829
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1967515541449176e-05,
+      "loss": 0.8046,
+      "step": 830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1967342616236643e-05,
+      "loss": 0.7322,
+      "step": 831
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1967169232729665e-05,
+      "loss": 0.731,
+      "step": 832
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196699539093621e-05,
+      "loss": 0.7815,
+      "step": 833
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196682109086428e-05,
+      "loss": 0.8239,
+      "step": 834
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196664633252188e-05,
+      "loss": 0.7694,
+      "step": 835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196647111591705e-05,
+      "loss": 0.7382,
+      "step": 836
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196629544105784e-05,
+      "loss": 0.7223,
+      "step": 837
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1966119307952338e-05,
+      "loss": 0.7846,
+      "step": 838
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965942716608638e-05,
+      "loss": 0.7375,
+      "step": 839
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965765667034858e-05,
+      "loss": 0.672,
+      "step": 840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965588159239144e-05,
+      "loss": 0.7883,
+      "step": 841
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965410193229647e-05,
+      "loss": 0.7615,
+      "step": 842
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965231769014558e-05,
+      "loss": 0.8001,
+      "step": 843
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196505288660208e-05,
+      "loss": 0.7912,
+      "step": 844
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964873546000436e-05,
+      "loss": 0.7543,
+      "step": 845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964693747217874e-05,
+      "loss": 0.7322,
+      "step": 846
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196451349026266e-05,
+      "loss": 0.7893,
+      "step": 847
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964332775143082e-05,
+      "loss": 0.6821,
+      "step": 848
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964151601867455e-05,
+      "loss": 0.7299,
+      "step": 849
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963969970444096e-05,
+      "loss": 0.6677,
+      "step": 850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196378788088137e-05,
+      "loss": 0.8428,
+      "step": 851
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963605333187638e-05,
+      "loss": 0.7667,
+      "step": 852
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963422327371303e-05,
+      "loss": 0.6882,
+      "step": 853
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963238863440772e-05,
+      "loss": 0.7796,
+      "step": 854
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963054941404487e-05,
+      "loss": 0.8173,
+      "step": 855
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19628705612709e-05,
+      "loss": 0.7865,
+      "step": 856
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196268572304849e-05,
+      "loss": 0.7781,
+      "step": 857
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196250042674576e-05,
+      "loss": 0.7208,
+      "step": 858
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962314672371223e-05,
+      "loss": 0.7366,
+      "step": 859
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962128459933423e-05,
+      "loss": 0.7709,
+      "step": 860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1961941789440923e-05,
+      "loss": 0.8223,
+      "step": 861
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1961754660902307e-05,
+      "loss": 0.7378,
+      "step": 862
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1961567074326172e-05,
+      "loss": 0.714,
+      "step": 863
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1961379029721155e-05,
+      "loss": 0.8353,
+      "step": 864
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1961190527095893e-05,
+      "loss": 0.7349,
+      "step": 865
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196100156645905e-05,
+      "loss": 0.8753,
+      "step": 866
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196081214781933e-05,
+      "loss": 0.7198,
+      "step": 867
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960622271185426e-05,
+      "loss": 0.8033,
+      "step": 868
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960431936566077e-05,
+      "loss": 0.7454,
+      "step": 869
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196024114397003e-05,
+      "loss": 0.7567,
+      "step": 870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960049893406064e-05,
+      "loss": 0.7332,
+      "step": 871
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959858184882967e-05,
+      "loss": 0.7189,
+      "step": 872
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959666018409556e-05,
+      "loss": 0.7928,
+      "step": 873
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959473393994666e-05,
+      "loss": 0.6958,
+      "step": 874
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959280311647152e-05,
+      "loss": 0.7648,
+      "step": 875
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959086771375896e-05,
+      "loss": 0.7346,
+      "step": 876
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958892773189793e-05,
+      "loss": 0.6482,
+      "step": 877
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958698317097766e-05,
+      "loss": 0.7639,
+      "step": 878
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958503403108754e-05,
+      "loss": 0.79,
+      "step": 879
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195830803123172e-05,
+      "loss": 0.7272,
+      "step": 880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195811220147565e-05,
+      "loss": 0.809,
+      "step": 881
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195791591384954e-05,
+      "loss": 0.6921,
+      "step": 882
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195771916836242e-05,
+      "loss": 0.7313,
+      "step": 883
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195752196502334e-05,
+      "loss": 0.7856,
+      "step": 884
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1957324303841363e-05,
+      "loss": 0.7798,
+      "step": 885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1957126184825575e-05,
+      "loss": 0.7446,
+      "step": 886
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956927607985095e-05,
+      "loss": 0.6962,
+      "step": 887
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956728573329042e-05,
+      "loss": 0.8293,
+      "step": 888
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956529080866576e-05,
+      "loss": 0.7665,
+      "step": 889
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956329130606864e-05,
+      "loss": 0.7307,
+      "step": 890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19561287225591e-05,
+      "loss": 0.7283,
+      "step": 891
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955927856732506e-05,
+      "loss": 0.8276,
+      "step": 892
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955726533136308e-05,
+      "loss": 0.7449,
+      "step": 893
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955524751779768e-05,
+      "loss": 0.7548,
+      "step": 894
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955322512672163e-05,
+      "loss": 0.7941,
+      "step": 895
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955119815822793e-05,
+      "loss": 0.8361,
+      "step": 896
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954916661240976e-05,
+      "loss": 0.7536,
+      "step": 897
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954713048936052e-05,
+      "loss": 0.8073,
+      "step": 898
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954508978917386e-05,
+      "loss": 0.7922,
+      "step": 899
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195430445119436e-05,
+      "loss": 0.7631,
+      "step": 900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954099465776378e-05,
+      "loss": 0.7299,
+      "step": 901
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1953894022672863e-05,
+      "loss": 0.7607,
+      "step": 902
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195368812189327e-05,
+      "loss": 0.721,
+      "step": 903
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1953481763447052e-05,
+      "loss": 0.7722,
+      "step": 904
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1953274947343706e-05,
+      "loss": 0.7971,
+      "step": 905
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195306767359274e-05,
+      "loss": 0.7278,
+      "step": 906
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952859942203684e-05,
+      "loss": 0.7503,
+      "step": 907
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952651753186095e-05,
+      "loss": 0.8093,
+      "step": 908
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952443106549536e-05,
+      "loss": 0.6911,
+      "step": 909
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952234002303604e-05,
+      "loss": 0.7322,
+      "step": 910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952024440457917e-05,
+      "loss": 0.7243,
+      "step": 911
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951814421022104e-05,
+      "loss": 0.7137,
+      "step": 912
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195160394400583e-05,
+      "loss": 0.7288,
+      "step": 913
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951393009418765e-05,
+      "loss": 0.7249,
+      "step": 914
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951181617270614e-05,
+      "loss": 0.7862,
+      "step": 915
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195096976757109e-05,
+      "loss": 0.744,
+      "step": 916
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1950757460329937e-05,
+      "loss": 0.709,
+      "step": 917
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195054469555692e-05,
+      "loss": 0.7673,
+      "step": 918
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1950331473261818e-05,
+      "loss": 0.7595,
+      "step": 919
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1950117793454435e-05,
+      "loss": 0.7125,
+      "step": 920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1949903656144596e-05,
+      "loss": 0.8032,
+      "step": 921
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194968906134215e-05,
+      "loss": 0.751,
+      "step": 922
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194947400905696e-05,
+      "loss": 0.829,
+      "step": 923
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1949258499298917e-05,
+      "loss": 0.7749,
+      "step": 924
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194904253207793e-05,
+      "loss": 0.7083,
+      "step": 925
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1948826107403924e-05,
+      "loss": 0.6984,
+      "step": 926
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1948609225286854e-05,
+      "loss": 0.7674,
+      "step": 927
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1948391885736695e-05,
+      "loss": 0.781,
+      "step": 928
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1948174088763435e-05,
+      "loss": 0.7829,
+      "step": 929
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194795583437709e-05,
+      "loss": 0.8324,
+      "step": 930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1947737122587698e-05,
+      "loss": 0.7647,
+      "step": 931
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1947517953405313e-05,
+      "loss": 0.7646,
+      "step": 932
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1947298326840013e-05,
+      "loss": 0.8071,
+      "step": 933
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1947078242901892e-05,
+      "loss": 0.757,
+      "step": 934
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1946857701601077e-05,
+      "loss": 0.7174,
+      "step": 935
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19466367029477e-05,
+      "loss": 0.749,
+      "step": 936
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194641524695193e-05,
+      "loss": 0.6854,
+      "step": 937
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1946193333623944e-05,
+      "loss": 0.8204,
+      "step": 938
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194597096297395e-05,
+      "loss": 0.7826,
+      "step": 939
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194574813501217e-05,
+      "loss": 0.6914,
+      "step": 940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194552484974885e-05,
+      "loss": 0.7808,
+      "step": 941
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1945301107194256e-05,
+      "loss": 0.6914,
+      "step": 942
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1945076907358676e-05,
+      "loss": 0.7909,
+      "step": 943
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1944852250252417e-05,
+      "loss": 0.7866,
+      "step": 944
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1944627135885813e-05,
+      "loss": 0.6947,
+      "step": 945
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1944401564269215e-05,
+      "loss": 0.7375,
+      "step": 946
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1944175535412986e-05,
+      "loss": 0.7306,
+      "step": 947
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1943949049327525e-05,
+      "loss": 0.7094,
+      "step": 948
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1943722106023248e-05,
+      "loss": 0.8083,
+      "step": 949
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1943494705510585e-05,
+      "loss": 0.7463,
+      "step": 950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1943266847799994e-05,
+      "loss": 0.7124,
+      "step": 951
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1943038532901953e-05,
+      "loss": 0.7814,
+      "step": 952
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1942809760826956e-05,
+      "loss": 0.842,
+      "step": 953
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1942580531585522e-05,
+      "loss": 0.7138,
+      "step": 954
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1942350845188196e-05,
+      "loss": 0.7957,
+      "step": 955
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1942120701645538e-05,
+      "loss": 0.7669,
+      "step": 956
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1941890100968123e-05,
+      "loss": 0.804,
+      "step": 957
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1941659043166562e-05,
+      "loss": 0.8275,
+      "step": 958
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1941427528251474e-05,
+      "loss": 0.6915,
+      "step": 959
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1941195556233508e-05,
+      "loss": 0.7438,
+      "step": 960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1940963127123326e-05,
+      "loss": 0.7063,
+      "step": 961
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1940730240931617e-05,
+      "loss": 0.7479,
+      "step": 962
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194049689766909e-05,
+      "loss": 0.774,
+      "step": 963
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194026309734647e-05,
+      "loss": 0.8656,
+      "step": 964
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194002883997451e-05,
+      "loss": 0.7646,
+      "step": 965
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193979412556398e-05,
+      "loss": 0.7137,
+      "step": 966
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1939558954125675e-05,
+      "loss": 0.8199,
+      "step": 967
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1939323325670404e-05,
+      "loss": 0.766,
+      "step": 968
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1939087240209005e-05,
+      "loss": 0.7075,
+      "step": 969
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1938850697752332e-05,
+      "loss": 0.7249,
+      "step": 970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1938613698311257e-05,
+      "loss": 0.7661,
+      "step": 971
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1938376241896683e-05,
+      "loss": 0.7244,
+      "step": 972
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1938138328519523e-05,
+      "loss": 0.7161,
+      "step": 973
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1937899958190724e-05,
+      "loss": 0.7957,
+      "step": 974
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1937661130921236e-05,
+      "loss": 0.7182,
+      "step": 975
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1937421846722047e-05,
+      "loss": 0.739,
+      "step": 976
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1937182105604156e-05,
+      "loss": 0.6929,
+      "step": 977
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1936941907578587e-05,
+      "loss": 0.7336,
+      "step": 978
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1936701252656388e-05,
+      "loss": 0.7087,
+      "step": 979
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193646014084862e-05,
+      "loss": 0.7381,
+      "step": 980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1936218572166366e-05,
+      "loss": 0.7894,
+      "step": 981
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1935976546620745e-05,
+      "loss": 0.7386,
+      "step": 982
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1935734064222872e-05,
+      "loss": 0.7202,
+      "step": 983
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19354911249839e-05,
+      "loss": 0.8009,
+      "step": 984
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1935247728915007e-05,
+      "loss": 0.8051,
+      "step": 985
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1935003876027374e-05,
+      "loss": 0.7837,
+      "step": 986
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193475956633222e-05,
+      "loss": 0.7582,
+      "step": 987
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1934514799840774e-05,
+      "loss": 0.7568,
+      "step": 988
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1934269576564295e-05,
+      "loss": 0.8372,
+      "step": 989
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1934023896514054e-05,
+      "loss": 0.8312,
+      "step": 990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1933777759701347e-05,
+      "loss": 0.8002,
+      "step": 991
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1933531166137494e-05,
+      "loss": 0.8194,
+      "step": 992
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193328411583383e-05,
+      "loss": 0.696,
+      "step": 993
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1933036608801722e-05,
+      "loss": 0.7655,
+      "step": 994
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193278864505254e-05,
+      "loss": 0.749,
+      "step": 995
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1932540224597694e-05,
+      "loss": 0.7077,
+      "step": 996
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19322913474486e-05,
+      "loss": 0.7891,
+      "step": 997
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1932042013616703e-05,
+      "loss": 0.7268,
+      "step": 998
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1931792223113466e-05,
+      "loss": 0.7338,
+      "step": 999
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193154197595038e-05,
+      "loss": 0.7078,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07,
+      "eval_text_loss": 0.6576274037361145,
+      "eval_text_runtime": 1.7988,
+      "eval_text_samples_per_second": 41.139,
+      "eval_text_steps_per_second": 1.112,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07,
+      "eval_svit_loss": 0.9333049058914185,
+      "eval_svit_runtime": 6.6093,
+      "eval_svit_samples_per_second": 123.462,
+      "eval_svit_steps_per_second": 1.967,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1931291272138947e-05,
+      "loss": 0.7199,
+      "step": 1001
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1931040111690693e-05,
+      "loss": 0.7079,
+      "step": 1002
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193078849461717e-05,
+      "loss": 0.8062,
+      "step": 1003
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1930536420929945e-05,
+      "loss": 0.8128,
+      "step": 1004
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1930283890640605e-05,
+      "loss": 0.7561,
+      "step": 1005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1930030903760767e-05,
+      "loss": 0.7483,
+      "step": 1006
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1929777460302063e-05,
+      "loss": 0.7394,
+      "step": 1007
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192952356027614e-05,
+      "loss": 0.6708,
+      "step": 1008
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192926920369468e-05,
+      "loss": 0.7801,
+      "step": 1009
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192901439056937e-05,
+      "loss": 0.759,
+      "step": 1010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1928759120911934e-05,
+      "loss": 0.7215,
+      "step": 1011
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1928503394734108e-05,
+      "loss": 0.7368,
+      "step": 1012
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1928247212047644e-05,
+      "loss": 0.8239,
+      "step": 1013
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1927990572864327e-05,
+      "loss": 0.7922,
+      "step": 1014
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1927733477195957e-05,
+      "loss": 0.7856,
+      "step": 1015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1927475925054353e-05,
+      "loss": 0.6443,
+      "step": 1016
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1927217916451355e-05,
+      "loss": 0.7468,
+      "step": 1017
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1926959451398833e-05,
+      "loss": 0.7579,
+      "step": 1018
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1926700529908665e-05,
+      "loss": 0.8261,
+      "step": 1019
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1926441151992757e-05,
+      "loss": 0.8418,
+      "step": 1020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1926181317663035e-05,
+      "loss": 0.7147,
+      "step": 1021
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1925921026931454e-05,
+      "loss": 0.7559,
+      "step": 1022
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1925660279809968e-05,
+      "loss": 0.7714,
+      "step": 1023
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192539907631058e-05,
+      "loss": 0.7283,
+      "step": 1024
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192513741644529e-05,
+      "loss": 0.8055,
+      "step": 1025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1924875300226135e-05,
+      "loss": 0.6472,
+      "step": 1026
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192461272766516e-05,
+      "loss": 0.7374,
+      "step": 1027
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1924349698774443e-05,
+      "loss": 0.744,
+      "step": 1028
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192408621356608e-05,
+      "loss": 0.7836,
+      "step": 1029
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1923822272052182e-05,
+      "loss": 0.7783,
+      "step": 1030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1923557874244887e-05,
+      "loss": 0.8364,
+      "step": 1031
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1923293020156353e-05,
+      "loss": 0.7487,
+      "step": 1032
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1923027709798753e-05,
+      "loss": 0.752,
+      "step": 1033
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192276194318429e-05,
+      "loss": 0.7644,
+      "step": 1034
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1922495720325182e-05,
+      "loss": 0.8115,
+      "step": 1035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1922229041233677e-05,
+      "loss": 0.6721,
+      "step": 1036
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1921961905922023e-05,
+      "loss": 0.7789,
+      "step": 1037
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192169431440251e-05,
+      "loss": 0.8519,
+      "step": 1038
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1921426266687447e-05,
+      "loss": 0.7362,
+      "step": 1039
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192115776278915e-05,
+      "loss": 0.7906,
+      "step": 1040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1920888802719967e-05,
+      "loss": 0.7382,
+      "step": 1041
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192061938649227e-05,
+      "loss": 0.6227,
+      "step": 1042
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192034951411844e-05,
+      "loss": 0.7798,
+      "step": 1043
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1920079185610892e-05,
+      "loss": 0.7531,
+      "step": 1044
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1919808400982048e-05,
+      "loss": 0.7772,
+      "step": 1045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1919537160244367e-05,
+      "loss": 0.7454,
+      "step": 1046
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1919265463410312e-05,
+      "loss": 0.7351,
+      "step": 1047
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191899331049238e-05,
+      "loss": 0.7706,
+      "step": 1048
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1918720701503084e-05,
+      "loss": 0.6752,
+      "step": 1049
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1918447636454958e-05,
+      "loss": 0.7052,
+      "step": 1050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191817411536056e-05,
+      "loss": 0.6558,
+      "step": 1051
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1917900138232462e-05,
+      "loss": 0.7944,
+      "step": 1052
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1917625705083266e-05,
+      "loss": 0.7601,
+      "step": 1053
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191735081592559e-05,
+      "loss": 0.7461,
+      "step": 1054
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1917075470772065e-05,
+      "loss": 0.7766,
+      "step": 1055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191679966963536e-05,
+      "loss": 0.8635,
+      "step": 1056
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1916523412528158e-05,
+      "loss": 0.7597,
+      "step": 1057
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1916246699463153e-05,
+      "loss": 0.721,
+      "step": 1058
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1915969530453072e-05,
+      "loss": 0.7271,
+      "step": 1059
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191569190551066e-05,
+      "loss": 0.7102,
+      "step": 1060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1915413824648685e-05,
+      "loss": 0.7457,
+      "step": 1061
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191513528787992e-05,
+      "loss": 0.8621,
+      "step": 1062
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191485629521719e-05,
+      "loss": 0.812,
+      "step": 1063
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191457684667331e-05,
+      "loss": 0.8143,
+      "step": 1064
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1914296942261138e-05,
+      "loss": 0.7377,
+      "step": 1065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1914016581993534e-05,
+      "loss": 0.8589,
+      "step": 1066
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19137357658834e-05,
+      "loss": 0.7548,
+      "step": 1067
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1913454493943635e-05,
+      "loss": 0.7684,
+      "step": 1068
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191317276618718e-05,
+      "loss": 0.7284,
+      "step": 1069
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191289058262699e-05,
+      "loss": 0.8383,
+      "step": 1070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1912607943276036e-05,
+      "loss": 0.7158,
+      "step": 1071
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1912324848147313e-05,
+      "loss": 0.6836,
+      "step": 1072
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191204129725384e-05,
+      "loss": 0.7429,
+      "step": 1073
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1911757290608657e-05,
+      "loss": 0.7373,
+      "step": 1074
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191147282822482e-05,
+      "loss": 0.7242,
+      "step": 1075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1911187910115403e-05,
+      "loss": 0.8009,
+      "step": 1076
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1910902536293513e-05,
+      "loss": 0.7463,
+      "step": 1077
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191061670677227e-05,
+      "loss": 0.7487,
+      "step": 1078
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1910330421564816e-05,
+      "loss": 0.7422,
+      "step": 1079
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1910043680684312e-05,
+      "loss": 0.6805,
+      "step": 1080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190975648414395e-05,
+      "loss": 0.7517,
+      "step": 1081
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1909468831956926e-05,
+      "loss": 0.6681,
+      "step": 1082
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1909180724136468e-05,
+      "loss": 0.7719,
+      "step": 1083
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1908892160695826e-05,
+      "loss": 0.7104,
+      "step": 1084
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190860314164827e-05,
+      "loss": 0.7451,
+      "step": 1085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1908313667007083e-05,
+      "loss": 0.7359,
+      "step": 1086
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190802373678558e-05,
+      "loss": 0.739,
+      "step": 1087
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1907733350997088e-05,
+      "loss": 0.7212,
+      "step": 1088
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1907442509654962e-05,
+      "loss": 0.7371,
+      "step": 1089
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190715121277257e-05,
+      "loss": 0.7064,
+      "step": 1090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1906859460363308e-05,
+      "loss": 0.782,
+      "step": 1091
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1906567252440597e-05,
+      "loss": 0.7425,
+      "step": 1092
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1906274589017865e-05,
+      "loss": 0.7184,
+      "step": 1093
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190598147010857e-05,
+      "loss": 0.7422,
+      "step": 1094
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1905687895726192e-05,
+      "loss": 0.7258,
+      "step": 1095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1905393865884224e-05,
+      "loss": 0.7867,
+      "step": 1096
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1905099380596195e-05,
+      "loss": 0.7466,
+      "step": 1097
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1904804439875636e-05,
+      "loss": 0.7318,
+      "step": 1098
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190450904373611e-05,
+      "loss": 0.7023,
+      "step": 1099
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19042131921912e-05,
+      "loss": 0.7167,
+      "step": 1100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1903916885254515e-05,
+      "loss": 0.7762,
+      "step": 1101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1903620122939673e-05,
+      "loss": 0.7584,
+      "step": 1102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190332290526032e-05,
+      "loss": 0.7094,
+      "step": 1103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190302523223012e-05,
+      "loss": 0.7413,
+      "step": 1104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1902727103862763e-05,
+      "loss": 0.8186,
+      "step": 1105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1902428520171954e-05,
+      "loss": 0.6879,
+      "step": 1106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1902129481171426e-05,
+      "loss": 0.737,
+      "step": 1107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1901829986874926e-05,
+      "loss": 0.7258,
+      "step": 1108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190153003729622e-05,
+      "loss": 0.7675,
+      "step": 1109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1901229632449113e-05,
+      "loss": 0.7422,
+      "step": 1110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1900928772347405e-05,
+      "loss": 0.6859,
+      "step": 1111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1900627457004933e-05,
+      "loss": 0.7885,
+      "step": 1112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190032568643555e-05,
+      "loss": 0.7865,
+      "step": 1113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1900023460653132e-05,
+      "loss": 0.7967,
+      "step": 1114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189972077967158e-05,
+      "loss": 0.7499,
+      "step": 1115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1899417643504803e-05,
+      "loss": 0.7132,
+      "step": 1116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189911405216675e-05,
+      "loss": 0.7299,
+      "step": 1117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189881000567137e-05,
+      "loss": 0.7992,
+      "step": 1118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1898505504032648e-05,
+      "loss": 0.8052,
+      "step": 1119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189820054726458e-05,
+      "loss": 0.7805,
+      "step": 1120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189789513538119e-05,
+      "loss": 0.7621,
+      "step": 1121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1897589268396526e-05,
+      "loss": 0.7618,
+      "step": 1122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1897282946324645e-05,
+      "loss": 0.7628,
+      "step": 1123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1896976169179634e-05,
+      "loss": 0.7907,
+      "step": 1124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.18966689369756e-05,
+      "loss": 0.7268,
+      "step": 1125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1896361249726667e-05,
+      "loss": 0.78,
+      "step": 1126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1896053107446983e-05,
+      "loss": 0.7364,
+      "step": 1127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1895744510150714e-05,
+      "loss": 0.7221,
+      "step": 1128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1895435457852055e-05,
+      "loss": 0.7656,
+      "step": 1129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1895125950565213e-05,
+      "loss": 0.7206,
+      "step": 1130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1894815988304416e-05,
+      "loss": 0.7349,
+      "step": 1131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189450557108392e-05,
+      "loss": 0.6691,
+      "step": 1132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1894194698918e-05,
+      "loss": 0.7641,
+      "step": 1133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189388337182094e-05,
+      "loss": 0.7201,
+      "step": 1134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1893571589807067e-05,
+      "loss": 0.6541,
+      "step": 1135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1893259352890704e-05,
+      "loss": 0.7413,
+      "step": 1136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189294666108622e-05,
+      "loss": 0.787,
+      "step": 1137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1892633514407986e-05,
+      "loss": 0.8387,
+      "step": 1138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18923199128704e-05,
+      "loss": 0.733,
+      "step": 1139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189200585648788e-05,
+      "loss": 0.7073,
+      "step": 1140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1891691345274873e-05,
+      "loss": 0.736,
+      "step": 1141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1891376379245832e-05,
+      "loss": 0.741,
+      "step": 1142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1891060958415244e-05,
+      "loss": 0.7983,
+      "step": 1143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189074508279761e-05,
+      "loss": 0.7191,
+      "step": 1144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1890428752407454e-05,
+      "loss": 0.6914,
+      "step": 1145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189011196725932e-05,
+      "loss": 0.6686,
+      "step": 1146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1889794727367777e-05,
+      "loss": 0.7802,
+      "step": 1147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1889477032747407e-05,
+      "loss": 0.7789,
+      "step": 1148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188915888341282e-05,
+      "loss": 0.669,
+      "step": 1149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1888840279378645e-05,
+      "loss": 0.7022,
+      "step": 1150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1888521220659533e-05,
+      "loss": 0.7643,
+      "step": 1151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1888201707270147e-05,
+      "loss": 0.7823,
+      "step": 1152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1887881739225185e-05,
+      "loss": 0.7477,
+      "step": 1153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1887561316539357e-05,
+      "loss": 0.7829,
+      "step": 1154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1887240439227395e-05,
+      "loss": 0.6604,
+      "step": 1155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1886919107304052e-05,
+      "loss": 0.7879,
+      "step": 1156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1886597320784106e-05,
+      "loss": 0.78,
+      "step": 1157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1886275079682354e-05,
+      "loss": 0.7566,
+      "step": 1158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1885952384013604e-05,
+      "loss": 0.6426,
+      "step": 1159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1885629233792703e-05,
+      "loss": 0.7304,
+      "step": 1160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18853056290345e-05,
+      "loss": 0.6292,
+      "step": 1161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1884981569753882e-05,
+      "loss": 0.8192,
+      "step": 1162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1884657055965743e-05,
+      "loss": 0.7113,
+      "step": 1163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1884332087685014e-05,
+      "loss": 0.7826,
+      "step": 1164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1884006664926626e-05,
+      "loss": 0.742,
+      "step": 1165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1883680787705547e-05,
+      "loss": 0.7574,
+      "step": 1166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188335445603676e-05,
+      "loss": 0.7502,
+      "step": 1167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188302766993527e-05,
+      "loss": 0.7921,
+      "step": 1168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1882700429416102e-05,
+      "loss": 0.6537,
+      "step": 1169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18823727344943e-05,
+      "loss": 0.6775,
+      "step": 1170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1882044585184935e-05,
+      "loss": 0.7771,
+      "step": 1171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1881715981503092e-05,
+      "loss": 0.6751,
+      "step": 1172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1881386923463883e-05,
+      "loss": 0.734,
+      "step": 1173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188105741108244e-05,
+      "loss": 0.6551,
+      "step": 1174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1880727444373904e-05,
+      "loss": 0.7131,
+      "step": 1175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1880397023353457e-05,
+      "loss": 0.7975,
+      "step": 1176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1880066148036288e-05,
+      "loss": 0.7652,
+      "step": 1177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1879734818437613e-05,
+      "loss": 0.7606,
+      "step": 1178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187940303457266e-05,
+      "loss": 0.7375,
+      "step": 1179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1879070796456688e-05,
+      "loss": 0.6874,
+      "step": 1180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1878738104104976e-05,
+      "loss": 0.8258,
+      "step": 1181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1878404957532818e-05,
+      "loss": 0.7854,
+      "step": 1182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1878071356755533e-05,
+      "loss": 0.7539,
+      "step": 1183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1877737301788458e-05,
+      "loss": 0.8005,
+      "step": 1184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1877402792646954e-05,
+      "loss": 0.766,
+      "step": 1185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18770678293464e-05,
+      "loss": 0.835,
+      "step": 1186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18767324119022e-05,
+      "loss": 0.7128,
+      "step": 1187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1876396540329776e-05,
+      "loss": 0.7936,
+      "step": 1188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187606021464457e-05,
+      "loss": 0.7678,
+      "step": 1189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1875723434862044e-05,
+      "loss": 0.7436,
+      "step": 1190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187538620099769e-05,
+      "loss": 0.7889,
+      "step": 1191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187504851306701e-05,
+      "loss": 0.8146,
+      "step": 1192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1874710371085527e-05,
+      "loss": 0.7623,
+      "step": 1193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1874371775068793e-05,
+      "loss": 0.6187,
+      "step": 1194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1874032725032375e-05,
+      "loss": 0.815,
+      "step": 1195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1873693220991863e-05,
+      "loss": 0.7672,
+      "step": 1196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187335326296287e-05,
+      "loss": 0.859,
+      "step": 1197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187301285096102e-05,
+      "loss": 0.7596,
+      "step": 1198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1872671985001972e-05,
+      "loss": 0.7115,
+      "step": 1199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1872330665101396e-05,
+      "loss": 0.7971,
+      "step": 1200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187198889127499e-05,
+      "loss": 0.7965,
+      "step": 1201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1871646663538457e-05,
+      "loss": 0.7829,
+      "step": 1202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1871303981907545e-05,
+      "loss": 0.7588,
+      "step": 1203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1870960846398e-05,
+      "loss": 0.7813,
+      "step": 1204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187061725702561e-05,
+      "loss": 0.7347,
+      "step": 1205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1870273213806165e-05,
+      "loss": 0.805,
+      "step": 1206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1869928716755487e-05,
+      "loss": 0.7157,
+      "step": 1207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1869583765889415e-05,
+      "loss": 0.7453,
+      "step": 1208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1869238361223815e-05,
+      "loss": 0.7247,
+      "step": 1209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186889250277456e-05,
+      "loss": 0.7532,
+      "step": 1210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1868546190557556e-05,
+      "loss": 0.8273,
+      "step": 1211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1868199424588728e-05,
+      "loss": 0.6918,
+      "step": 1212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1867852204884015e-05,
+      "loss": 0.7075,
+      "step": 1213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186750453145939e-05,
+      "loss": 0.6717,
+      "step": 1214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186715640433083e-05,
+      "loss": 0.7586,
+      "step": 1215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1866807823514354e-05,
+      "loss": 0.7542,
+      "step": 1216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1866458789025973e-05,
+      "loss": 0.7579,
+      "step": 1217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186610930088175e-05,
+      "loss": 0.686,
+      "step": 1218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186575935909775e-05,
+      "loss": 0.6783,
+      "step": 1219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1865408963690056e-05,
+      "loss": 0.7843,
+      "step": 1220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186505811467479e-05,
+      "loss": 0.8216,
+      "step": 1221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186470681206808e-05,
+      "loss": 0.7815,
+      "step": 1222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1864355055886072e-05,
+      "loss": 0.7319,
+      "step": 1223
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186400284614495e-05,
+      "loss": 0.8344,
+      "step": 1224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18636501828609e-05,
+      "loss": 0.7756,
+      "step": 1225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1863297066050145e-05,
+      "loss": 0.7489,
+      "step": 1226
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1862943495728914e-05,
+      "loss": 0.8302,
+      "step": 1227
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186258947191347e-05,
+      "loss": 0.7533,
+      "step": 1228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186223499462008e-05,
+      "loss": 0.7215,
+      "step": 1229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186188006386506e-05,
+      "loss": 0.7749,
+      "step": 1230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186152467966472e-05,
+      "loss": 0.7123,
+      "step": 1231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1861168842035398e-05,
+      "loss": 0.719,
+      "step": 1232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186081255099346e-05,
+      "loss": 0.7965,
+      "step": 1233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186045580655528e-05,
+      "loss": 0.7951,
+      "step": 1234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1860098608737275e-05,
+      "loss": 0.751,
+      "step": 1235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185974095755586e-05,
+      "loss": 0.7487,
+      "step": 1236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1859382853027478e-05,
+      "loss": 0.7242,
+      "step": 1237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1859024295168594e-05,
+      "loss": 0.789,
+      "step": 1238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1858665283995702e-05,
+      "loss": 0.7224,
+      "step": 1239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1858305819525306e-05,
+      "loss": 0.7656,
+      "step": 1240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185794590177393e-05,
+      "loss": 0.7491,
+      "step": 1241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1857585530758127e-05,
+      "loss": 0.7406,
+      "step": 1242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1857224706494464e-05,
+      "loss": 0.6882,
+      "step": 1243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1856863428999534e-05,
+      "loss": 0.7428,
+      "step": 1244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1856501698289946e-05,
+      "loss": 0.7004,
+      "step": 1245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1856139514382335e-05,
+      "loss": 0.7289,
+      "step": 1246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185577687729335e-05,
+      "loss": 0.8033,
+      "step": 1247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185541378703967e-05,
+      "loss": 0.7458,
+      "step": 1248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1855050243637985e-05,
+      "loss": 0.818,
+      "step": 1249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1854686247105018e-05,
+      "loss": 0.8015,
+      "step": 1250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1854321797457494e-05,
+      "loss": 0.7323,
+      "step": 1251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1853956894712182e-05,
+      "loss": 0.6988,
+      "step": 1252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185359153888585e-05,
+      "loss": 0.7296,
+      "step": 1253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1853225729995303e-05,
+      "loss": 0.7151,
+      "step": 1254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185285946805736e-05,
+      "loss": 0.833,
+      "step": 1255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185249275308886e-05,
+      "loss": 0.7381,
+      "step": 1256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1852125585106666e-05,
+      "loss": 0.6075,
+      "step": 1257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1851757964127658e-05,
+      "loss": 0.8289,
+      "step": 1258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185138989016874e-05,
+      "loss": 0.7595,
+      "step": 1259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1851021363246836e-05,
+      "loss": 0.6167,
+      "step": 1260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1850652383378895e-05,
+      "loss": 0.778,
+      "step": 1261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185028295058188e-05,
+      "loss": 0.7263,
+      "step": 1262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184991306487277e-05,
+      "loss": 0.7341,
+      "step": 1263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184954272626858e-05,
+      "loss": 0.7757,
+      "step": 1264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1849171934786338e-05,
+      "loss": 0.7169,
+      "step": 1265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184880069044309e-05,
+      "loss": 0.8354,
+      "step": 1266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184842899325591e-05,
+      "loss": 0.7333,
+      "step": 1267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184805684324188e-05,
+      "loss": 0.7172,
+      "step": 1268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1847684240418115e-05,
+      "loss": 0.7443,
+      "step": 1269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1847311184801752e-05,
+      "loss": 0.7909,
+      "step": 1270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184693767640994e-05,
+      "loss": 0.7891,
+      "step": 1271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184656371525986e-05,
+      "loss": 0.7273,
+      "step": 1272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1846189301368692e-05,
+      "loss": 0.8095,
+      "step": 1273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1845814434753665e-05,
+      "loss": 0.7573,
+      "step": 1274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1845439115432003e-05,
+      "loss": 0.7813,
+      "step": 1275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1845063343420975e-05,
+      "loss": 0.7123,
+      "step": 1276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1844687118737853e-05,
+      "loss": 0.7509,
+      "step": 1277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1844310441399932e-05,
+      "loss": 0.7969,
+      "step": 1278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184393331142454e-05,
+      "loss": 0.8302,
+      "step": 1279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1843555728829014e-05,
+      "loss": 0.7091,
+      "step": 1280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1843177693630708e-05,
+      "loss": 0.8032,
+      "step": 1281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1842799205847015e-05,
+      "loss": 0.7486,
+      "step": 1282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1842420265495333e-05,
+      "loss": 0.6412,
+      "step": 1283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1842040872593083e-05,
+      "loss": 0.7251,
+      "step": 1284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.184166102715771e-05,
+      "loss": 0.7381,
+      "step": 1285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1841280729206682e-05,
+      "loss": 0.7129,
+      "step": 1286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1840899978757484e-05,
+      "loss": 0.7441,
+      "step": 1287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1840518775827622e-05,
+      "loss": 0.6572,
+      "step": 1288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1840137120434623e-05,
+      "loss": 0.8259,
+      "step": 1289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1839755012596038e-05,
+      "loss": 0.7561,
+      "step": 1290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1839372452329437e-05,
+      "loss": 0.7724,
+      "step": 1291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1838989439652403e-05,
+      "loss": 0.7554,
+      "step": 1292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183860597458255e-05,
+      "loss": 0.669,
+      "step": 1293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1838222057137516e-05,
+      "loss": 0.7721,
+      "step": 1294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1837837687334942e-05,
+      "loss": 0.7858,
+      "step": 1295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1837452865192515e-05,
+      "loss": 0.7309,
+      "step": 1296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1837067590727914e-05,
+      "loss": 0.7732,
+      "step": 1297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1836681863958866e-05,
+      "loss": 0.7522,
+      "step": 1298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.18362956849031e-05,
+      "loss": 0.7933,
+      "step": 1299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1835909053578375e-05,
+      "loss": 0.7394,
+      "step": 1300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1835521970002465e-05,
+      "loss": 0.766,
+      "step": 1301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1835134434193168e-05,
+      "loss": 0.6898,
+      "step": 1302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1834746446168312e-05,
+      "loss": 0.7734,
+      "step": 1303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1834358005945725e-05,
+      "loss": 0.7561,
+      "step": 1304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183396911354327e-05,
+      "loss": 0.7929,
+      "step": 1305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183357976897883e-05,
+      "loss": 0.8303,
+      "step": 1306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183318997227031e-05,
+      "loss": 0.7112,
+      "step": 1307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1832799723435623e-05,
+      "loss": 0.7265,
+      "step": 1308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1832409022492725e-05,
+      "loss": 0.7234,
+      "step": 1309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1832017869459575e-05,
+      "loss": 0.7971,
+      "step": 1310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1831626264354155e-05,
+      "loss": 0.6981,
+      "step": 1311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1831234207194472e-05,
+      "loss": 0.6596,
+      "step": 1312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183084169799855e-05,
+      "loss": 0.7743,
+      "step": 1313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1830448736784448e-05,
+      "loss": 0.7786,
+      "step": 1314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183005532357022e-05,
+      "loss": 0.7242,
+      "step": 1315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1829661458373965e-05,
+      "loss": 0.7046,
+      "step": 1316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182926714121379e-05,
+      "loss": 0.7656,
+      "step": 1317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1828872372107822e-05,
+      "loss": 0.6535,
+      "step": 1318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1828477151074215e-05,
+      "loss": 0.7573,
+      "step": 1319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1828081478131142e-05,
+      "loss": 0.6966,
+      "step": 1320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1827685353296795e-05,
+      "loss": 0.7485,
+      "step": 1321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182728877658939e-05,
+      "loss": 0.8224,
+      "step": 1322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1826891748027154e-05,
+      "loss": 0.7401,
+      "step": 1323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182649426762835e-05,
+      "loss": 0.7619,
+      "step": 1324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182609633541125e-05,
+      "loss": 0.6888,
+      "step": 1325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1825697951394154e-05,
+      "loss": 0.8189,
+      "step": 1326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1825299115595376e-05,
+      "loss": 0.6863,
+      "step": 1327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1824899828033255e-05,
+      "loss": 0.8448,
+      "step": 1328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1824500088726153e-05,
+      "loss": 0.821,
+      "step": 1329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1824099897692443e-05,
+      "loss": 0.6814,
+      "step": 1330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1823699254950537e-05,
+      "loss": 0.7136,
+      "step": 1331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1823298160518845e-05,
+      "loss": 0.7754,
+      "step": 1332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1822896614415816e-05,
+      "loss": 0.6487,
+      "step": 1333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182249461665991e-05,
+      "loss": 0.7361,
+      "step": 1334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1822092167269614e-05,
+      "loss": 0.8568,
+      "step": 1335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1821689266263426e-05,
+      "loss": 0.7633,
+      "step": 1336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1821285913659877e-05,
+      "loss": 0.7894,
+      "step": 1337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1820882109477515e-05,
+      "loss": 0.7075,
+      "step": 1338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1820477853734898e-05,
+      "loss": 0.7601,
+      "step": 1339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1820073146450622e-05,
+      "loss": 0.7837,
+      "step": 1340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181966798764329e-05,
+      "loss": 0.7895,
+      "step": 1341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1819262377331534e-05,
+      "loss": 0.757,
+      "step": 1342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1818856315534e-05,
+      "loss": 0.6975,
+      "step": 1343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1818449802269365e-05,
+      "loss": 0.7307,
+      "step": 1344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181804283755631e-05,
+      "loss": 0.7943,
+      "step": 1345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181763542141356e-05,
+      "loss": 0.7998,
+      "step": 1346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181722755385984e-05,
+      "loss": 0.7469,
+      "step": 1347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1816819234913905e-05,
+      "loss": 0.7635,
+      "step": 1348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1816410464594526e-05,
+      "loss": 0.7574,
+      "step": 1349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1816001242920506e-05,
+      "loss": 0.8011,
+      "step": 1350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1815591569910655e-05,
+      "loss": 0.7777,
+      "step": 1351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1815181445583812e-05,
+      "loss": 0.7449,
+      "step": 1352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181477086995883e-05,
+      "loss": 0.7603,
+      "step": 1353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1814359843054596e-05,
+      "loss": 0.6625,
+      "step": 1354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181394836489e-05,
+      "loss": 0.693,
+      "step": 1355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181353643548396e-05,
+      "loss": 0.7258,
+      "step": 1356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1813124054855428e-05,
+      "loss": 0.7708,
+      "step": 1357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1812711223023357e-05,
+      "loss": 0.8248,
+      "step": 1358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181229794000673e-05,
+      "loss": 0.703,
+      "step": 1359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1811884205824544e-05,
+      "loss": 0.8252,
+      "step": 1360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1811470020495833e-05,
+      "loss": 0.678,
+      "step": 1361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1811055384039638e-05,
+      "loss": 0.6635,
+      "step": 1362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181064029647502e-05,
+      "loss": 0.7104,
+      "step": 1363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1810224757821065e-05,
+      "loss": 0.8107,
+      "step": 1364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180980876809688e-05,
+      "loss": 0.7759,
+      "step": 1365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1809392327321598e-05,
+      "loss": 0.7155,
+      "step": 1366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180897543551436e-05,
+      "loss": 0.7478,
+      "step": 1367
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180855809269433e-05,
+      "loss": 0.6985,
+      "step": 1368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180814029888071e-05,
+      "loss": 0.7829,
+      "step": 1369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1807722054092705e-05,
+      "loss": 0.8243,
+      "step": 1370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180730335834954e-05,
+      "loss": 0.7648,
+      "step": 1371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180688421167047e-05,
+      "loss": 0.7465,
+      "step": 1372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1806464614074767e-05,
+      "loss": 0.6928,
+      "step": 1373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180604456558173e-05,
+      "loss": 0.7606,
+      "step": 1374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180562406621066e-05,
+      "loss": 0.778,
+      "step": 1375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1805203115980905e-05,
+      "loss": 0.7154,
+      "step": 1376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1804781714911813e-05,
+      "loss": 0.7882,
+      "step": 1377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1804359863022756e-05,
+      "loss": 0.7032,
+      "step": 1378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180393756033314e-05,
+      "loss": 0.7624,
+      "step": 1379
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1803514806862376e-05,
+      "loss": 0.6886,
+      "step": 1380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1803091602629905e-05,
+      "loss": 0.7874,
+      "step": 1381
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1802667947655183e-05,
+      "loss": 0.7474,
+      "step": 1382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1802243841957692e-05,
+      "loss": 0.7405,
+      "step": 1383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180181928555693e-05,
+      "loss": 0.7981,
+      "step": 1384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180139427847242e-05,
+      "loss": 0.7037,
+      "step": 1385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1800968820723703e-05,
+      "loss": 0.7214,
+      "step": 1386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180054291233034e-05,
+      "loss": 0.7138,
+      "step": 1387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1800116553311915e-05,
+      "loss": 0.75,
+      "step": 1388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1799689743688033e-05,
+      "loss": 0.736,
+      "step": 1389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1799262483478318e-05,
+      "loss": 0.7427,
+      "step": 1390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1798834772702413e-05,
+      "loss": 0.818,
+      "step": 1391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1798406611379987e-05,
+      "loss": 0.7813,
+      "step": 1392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1797977999530727e-05,
+      "loss": 0.7292,
+      "step": 1393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1797548937174336e-05,
+      "loss": 0.7305,
+      "step": 1394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1797119424330547e-05,
+      "loss": 0.6789,
+      "step": 1395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1796689461019105e-05,
+      "loss": 0.7069,
+      "step": 1396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1796259047259784e-05,
+      "loss": 0.7658,
+      "step": 1397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179582818307237e-05,
+      "loss": 0.7503,
+      "step": 1398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1795396868476674e-05,
+      "loss": 0.7356,
+      "step": 1399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1794965103492533e-05,
+      "loss": 0.8337,
+      "step": 1400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1794532888139792e-05,
+      "loss": 0.7691,
+      "step": 1401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1794100222438328e-05,
+      "loss": 0.7635,
+      "step": 1402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1793667106408037e-05,
+      "loss": 0.754,
+      "step": 1403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179323354006883e-05,
+      "loss": 0.7003,
+      "step": 1404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1792799523440642e-05,
+      "loss": 0.7456,
+      "step": 1405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1792365056543433e-05,
+      "loss": 0.8024,
+      "step": 1406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1791930139397177e-05,
+      "loss": 0.789,
+      "step": 1407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1791494772021868e-05,
+      "loss": 0.7299,
+      "step": 1408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179105895443753e-05,
+      "loss": 0.7514,
+      "step": 1409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1790622686664194e-05,
+      "loss": 0.7758,
+      "step": 1410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1790185968721928e-05,
+      "loss": 0.8009,
+      "step": 1411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1789748800630806e-05,
+      "loss": 0.7581,
+      "step": 1412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1789311182410936e-05,
+      "loss": 0.7375,
+      "step": 1413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178887311408243e-05,
+      "loss": 0.7454,
+      "step": 1414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1788434595665437e-05,
+      "loss": 0.7737,
+      "step": 1415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1787995627180116e-05,
+      "loss": 0.7188,
+      "step": 1416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1787556208646652e-05,
+      "loss": 0.8075,
+      "step": 1417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178711634008525e-05,
+      "loss": 0.6897,
+      "step": 1418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1786676021516138e-05,
+      "loss": 0.7316,
+      "step": 1419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1786235252959556e-05,
+      "loss": 0.7557,
+      "step": 1420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1785794034435773e-05,
+      "loss": 0.7551,
+      "step": 1421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1785352365965076e-05,
+      "loss": 0.7182,
+      "step": 1422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178491024756777e-05,
+      "loss": 0.7205,
+      "step": 1423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178446767926419e-05,
+      "loss": 0.694,
+      "step": 1424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1784024661074676e-05,
+      "loss": 0.7912,
+      "step": 1425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1783581193019605e-05,
+      "loss": 0.7336,
+      "step": 1426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1783137275119367e-05,
+      "loss": 0.7424,
+      "step": 1427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1782692907394374e-05,
+      "loss": 0.7598,
+      "step": 1428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178224808986505e-05,
+      "loss": 0.7432,
+      "step": 1429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1781802822551853e-05,
+      "loss": 0.7843,
+      "step": 1430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178135710547526e-05,
+      "loss": 0.6852,
+      "step": 1431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1780910938655757e-05,
+      "loss": 0.734,
+      "step": 1432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1780464322113864e-05,
+      "loss": 0.7831,
+      "step": 1433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1780017255870115e-05,
+      "loss": 0.7282,
+      "step": 1434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1779569739945068e-05,
+      "loss": 0.8119,
+      "step": 1435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1779121774359293e-05,
+      "loss": 0.7931,
+      "step": 1436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1778673359133396e-05,
+      "loss": 0.7793,
+      "step": 1437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1778224494287987e-05,
+      "loss": 0.7382,
+      "step": 1438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1777775179843707e-05,
+      "loss": 0.7366,
+      "step": 1439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177732541582122e-05,
+      "loss": 0.8251,
+      "step": 1440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17768752022412e-05,
+      "loss": 0.806,
+      "step": 1441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1776424539124354e-05,
+      "loss": 0.7582,
+      "step": 1442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1775973426491395e-05,
+      "loss": 0.7854,
+      "step": 1443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1775521864363075e-05,
+      "loss": 0.7615,
+      "step": 1444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1775069852760143e-05,
+      "loss": 0.746,
+      "step": 1445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1774617391703396e-05,
+      "loss": 0.7289,
+      "step": 1446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177416448121363e-05,
+      "loss": 0.7303,
+      "step": 1447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1773711121311674e-05,
+      "loss": 0.7152,
+      "step": 1448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1773257312018372e-05,
+      "loss": 0.7546,
+      "step": 1449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1772803053354586e-05,
+      "loss": 0.7703,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177234834534121e-05,
+      "loss": 0.812,
+      "step": 1451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177189318799914e-05,
+      "loss": 0.7114,
+      "step": 1452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1771437581349318e-05,
+      "loss": 0.6877,
+      "step": 1453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1770981525412684e-05,
+      "loss": 0.7506,
+      "step": 1454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1770525020210206e-05,
+      "loss": 0.8205,
+      "step": 1455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1770068065762882e-05,
+      "loss": 0.7327,
+      "step": 1456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1769610662091716e-05,
+      "loss": 0.7137,
+      "step": 1457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1769152809217738e-05,
+      "loss": 0.785,
+      "step": 1458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1768694507162005e-05,
+      "loss": 0.7967,
+      "step": 1459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1768235755945584e-05,
+      "loss": 0.7733,
+      "step": 1460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1767776555589574e-05,
+      "loss": 0.7059,
+      "step": 1461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1767316906115086e-05,
+      "loss": 0.7512,
+      "step": 1462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1766856807543254e-05,
+      "loss": 0.6796,
+      "step": 1463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1766396259895236e-05,
+      "loss": 0.7341,
+      "step": 1464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1765935263192202e-05,
+      "loss": 0.6902,
+      "step": 1465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1765473817455356e-05,
+      "loss": 0.7126,
+      "step": 1466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176501192270591e-05,
+      "loss": 0.7273,
+      "step": 1467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1764549578965104e-05,
+      "loss": 0.7569,
+      "step": 1468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1764086786254195e-05,
+      "loss": 0.7881,
+      "step": 1469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176362354459446e-05,
+      "loss": 0.8014,
+      "step": 1470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1763159854007203e-05,
+      "loss": 0.7123,
+      "step": 1471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1762695714513742e-05,
+      "loss": 0.7366,
+      "step": 1472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1762231126135418e-05,
+      "loss": 0.7659,
+      "step": 1473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1761766088893593e-05,
+      "loss": 0.6942,
+      "step": 1474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176130060280965e-05,
+      "loss": 0.805,
+      "step": 1475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176083466790499e-05,
+      "loss": 0.7286,
+      "step": 1476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1760368284201036e-05,
+      "loss": 0.7162,
+      "step": 1477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1759901451719236e-05,
+      "loss": 0.673,
+      "step": 1478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175943417048105e-05,
+      "loss": 0.8599,
+      "step": 1479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1758966440507968e-05,
+      "loss": 0.7545,
+      "step": 1480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1758498261821495e-05,
+      "loss": 0.6851,
+      "step": 1481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1758029634443155e-05,
+      "loss": 0.7375,
+      "step": 1482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1757560558394494e-05,
+      "loss": 0.7806,
+      "step": 1483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1757091033697086e-05,
+      "loss": 0.7235,
+      "step": 1484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1756621060372516e-05,
+      "loss": 0.7069,
+      "step": 1485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1756150638442393e-05,
+      "loss": 0.6985,
+      "step": 1486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175567976792835e-05,
+      "loss": 0.6637,
+      "step": 1487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175520844885203e-05,
+      "loss": 0.7131,
+      "step": 1488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175473668123511e-05,
+      "loss": 0.6881,
+      "step": 1489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175426446509928e-05,
+      "loss": 0.7064,
+      "step": 1490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1753791800466254e-05,
+      "loss": 0.7825,
+      "step": 1491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1753318687357763e-05,
+      "loss": 0.6792,
+      "step": 1492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1752845125795563e-05,
+      "loss": 0.8097,
+      "step": 1493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1752371115801428e-05,
+      "loss": 0.7716,
+      "step": 1494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1751896657397148e-05,
+      "loss": 0.8117,
+      "step": 1495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1751421750604542e-05,
+      "loss": 0.8635,
+      "step": 1496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1750946395445447e-05,
+      "loss": 0.7758,
+      "step": 1497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1750470591941717e-05,
+      "loss": 0.7409,
+      "step": 1498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1749994340115228e-05,
+      "loss": 0.7069,
+      "step": 1499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1749517639987885e-05,
+      "loss": 0.7741,
+      "step": 1500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17490404915816e-05,
+      "loss": 0.8349,
+      "step": 1501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1748562894918315e-05,
+      "loss": 0.8121,
+      "step": 1502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1748084850019988e-05,
+      "loss": 0.7699,
+      "step": 1503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17476063569086e-05,
+      "loss": 0.785,
+      "step": 1504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1747127415606153e-05,
+      "loss": 0.7395,
+      "step": 1505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1746648026134664e-05,
+      "loss": 0.7945,
+      "step": 1506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174616818851618e-05,
+      "loss": 0.7076,
+      "step": 1507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1745687902772765e-05,
+      "loss": 0.6722,
+      "step": 1508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1745207168926497e-05,
+      "loss": 0.8035,
+      "step": 1509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1744725986999484e-05,
+      "loss": 0.8255,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174424435701385e-05,
+      "loss": 0.7305,
+      "step": 1511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174376227899174e-05,
+      "loss": 0.785,
+      "step": 1512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1743279752955313e-05,
+      "loss": 0.8004,
+      "step": 1513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174279677892677e-05,
+      "loss": 0.7069,
+      "step": 1514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1742313356928302e-05,
+      "loss": 0.8353,
+      "step": 1515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1741829486982147e-05,
+      "loss": 0.6738,
+      "step": 1516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1741345169110553e-05,
+      "loss": 0.7305,
+      "step": 1517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1740860403335783e-05,
+      "loss": 0.8265,
+      "step": 1518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1740375189680128e-05,
+      "loss": 0.7163,
+      "step": 1519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1739889528165902e-05,
+      "loss": 0.7689,
+      "step": 1520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1739403418815433e-05,
+      "loss": 0.68,
+      "step": 1521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173891686165107e-05,
+      "loss": 0.7051,
+      "step": 1522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1738429856695185e-05,
+      "loss": 0.7941,
+      "step": 1523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1737942403970174e-05,
+      "loss": 0.7769,
+      "step": 1524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1737454503498448e-05,
+      "loss": 0.6551,
+      "step": 1525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173696615530244e-05,
+      "loss": 0.7097,
+      "step": 1526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1736477359404602e-05,
+      "loss": 0.7446,
+      "step": 1527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1735988115827412e-05,
+      "loss": 0.6728,
+      "step": 1528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1735498424593364e-05,
+      "loss": 0.817,
+      "step": 1529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1735008285724975e-05,
+      "loss": 0.7147,
+      "step": 1530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173451769924478e-05,
+      "loss": 0.7701,
+      "step": 1531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1734026665175336e-05,
+      "loss": 0.7366,
+      "step": 1532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1733535183539216e-05,
+      "loss": 0.7699,
+      "step": 1533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173304325435903e-05,
+      "loss": 0.8295,
+      "step": 1534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173255087765739e-05,
+      "loss": 0.7611,
+      "step": 1535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1732058053456927e-05,
+      "loss": 0.7647,
+      "step": 1536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1731564781780316e-05,
+      "loss": 0.6773,
+      "step": 1537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1731071062650226e-05,
+      "loss": 0.7779,
+      "step": 1538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1730576896089365e-05,
+      "loss": 0.6691,
+      "step": 1539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1730082282120452e-05,
+      "loss": 0.7171,
+      "step": 1540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1729587220766226e-05,
+      "loss": 0.7205,
+      "step": 1541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1729091712049455e-05,
+      "loss": 0.7904,
+      "step": 1542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1728595755992918e-05,
+      "loss": 0.771,
+      "step": 1543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1728099352619423e-05,
+      "loss": 0.7247,
+      "step": 1544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1727602501951793e-05,
+      "loss": 0.7416,
+      "step": 1545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172710520401287e-05,
+      "loss": 0.7218,
+      "step": 1546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1726607458825522e-05,
+      "loss": 0.7277,
+      "step": 1547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1726109266412634e-05,
+      "loss": 0.6664,
+      "step": 1548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1725610626797115e-05,
+      "loss": 0.7505,
+      "step": 1549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172511154000189e-05,
+      "loss": 0.7387,
+      "step": 1550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172461200604991e-05,
+      "loss": 0.6845,
+      "step": 1551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1724112024964138e-05,
+      "loss": 0.7655,
+      "step": 1552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172361159676757e-05,
+      "loss": 0.7395,
+      "step": 1553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172311072148321e-05,
+      "loss": 0.7474,
+      "step": 1554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1722609399134087e-05,
+      "loss": 0.7096,
+      "step": 1555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1722107629743252e-05,
+      "loss": 0.8613,
+      "step": 1556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1721605413333785e-05,
+      "loss": 0.7409,
+      "step": 1557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1721102749928765e-05,
+      "loss": 0.7039,
+      "step": 1558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1720599639551316e-05,
+      "loss": 0.7076,
+      "step": 1559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1720096082224563e-05,
+      "loss": 0.7209,
+      "step": 1560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171959207797166e-05,
+      "loss": 0.7452,
+      "step": 1561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1719087626815783e-05,
+      "loss": 0.8215,
+      "step": 1562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1718582728780127e-05,
+      "loss": 0.7135,
+      "step": 1563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17180773838879e-05,
+      "loss": 0.7091,
+      "step": 1564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171757159216235e-05,
+      "loss": 0.7605,
+      "step": 1565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1717065353626725e-05,
+      "loss": 0.8328,
+      "step": 1566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17165586683043e-05,
+      "loss": 0.7889,
+      "step": 1567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171605153621838e-05,
+      "loss": 0.7741,
+      "step": 1568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1715543957392277e-05,
+      "loss": 0.6961,
+      "step": 1569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1715035931849328e-05,
+      "loss": 0.7641,
+      "step": 1570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1714527459612895e-05,
+      "loss": 0.7445,
+      "step": 1571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1714018540706357e-05,
+      "loss": 0.7804,
+      "step": 1572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1713509175153114e-05,
+      "loss": 0.7975,
+      "step": 1573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1712999362976584e-05,
+      "loss": 0.795,
+      "step": 1574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171248910420021e-05,
+      "loss": 0.706,
+      "step": 1575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1711978398847454e-05,
+      "loss": 0.775,
+      "step": 1576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1711467246941797e-05,
+      "loss": 0.7378,
+      "step": 1577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1710955648506744e-05,
+      "loss": 0.8422,
+      "step": 1578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1710443603565814e-05,
+      "loss": 0.6933,
+      "step": 1579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1709931112142553e-05,
+      "loss": 0.7236,
+      "step": 1580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1709418174260524e-05,
+      "loss": 0.6985,
+      "step": 1581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1708904789943312e-05,
+      "loss": 0.6828,
+      "step": 1582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1708390959214522e-05,
+      "loss": 0.7475,
+      "step": 1583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1707876682097784e-05,
+      "loss": 0.7202,
+      "step": 1584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1707361958616736e-05,
+      "loss": 0.6408,
+      "step": 1585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.170684678879505e-05,
+      "loss": 0.7975,
+      "step": 1586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170633117265642e-05,
+      "loss": 0.6737,
+      "step": 1587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1705815110224535e-05,
+      "loss": 0.814,
+      "step": 1588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1705298601523143e-05,
+      "loss": 0.7099,
+      "step": 1589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1704781646575984e-05,
+      "loss": 0.7475,
+      "step": 1590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1704264245406824e-05,
+      "loss": 0.719,
+      "step": 1591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170374639803946e-05,
+      "loss": 0.7367,
+      "step": 1592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1703228104497698e-05,
+      "loss": 0.7802,
+      "step": 1593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170270936480537e-05,
+      "loss": 0.7789,
+      "step": 1594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1702190178986332e-05,
+      "loss": 0.7357,
+      "step": 1595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170167054706445e-05,
+      "loss": 0.6953,
+      "step": 1596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1701150469063615e-05,
+      "loss": 0.715,
+      "step": 1597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1700629945007743e-05,
+      "loss": 0.7305,
+      "step": 1598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170010897492077e-05,
+      "loss": 0.73,
+      "step": 1599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169958755882665e-05,
+      "loss": 0.7634,
+      "step": 1600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169906569674935e-05,
+      "loss": 0.7136,
+      "step": 1601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1698543388712872e-05,
+      "loss": 0.7407,
+      "step": 1602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169802063474123e-05,
+      "loss": 0.7805,
+      "step": 1603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1697497434858464e-05,
+      "loss": 0.7404,
+      "step": 1604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1696973789088618e-05,
+      "loss": 0.6864,
+      "step": 1605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1696449697455784e-05,
+      "loss": 0.7071,
+      "step": 1606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169592515998405e-05,
+      "loss": 0.6797,
+      "step": 1607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1695400176697536e-05,
+      "loss": 0.8059,
+      "step": 1608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1694874747620383e-05,
+      "loss": 0.8146,
+      "step": 1609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169434887277675e-05,
+      "loss": 0.6726,
+      "step": 1610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1693822552190813e-05,
+      "loss": 0.7232,
+      "step": 1611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169329578588677e-05,
+      "loss": 0.7839,
+      "step": 1612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169276857388885e-05,
+      "loss": 0.7965,
+      "step": 1613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169224091622129e-05,
+      "loss": 0.7367,
+      "step": 1614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169171281290835e-05,
+      "loss": 0.7468,
+      "step": 1615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1691184263974307e-05,
+      "loss": 0.7058,
+      "step": 1616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1690655269443477e-05,
+      "loss": 0.7603,
+      "step": 1617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1690125829340173e-05,
+      "loss": 0.8072,
+      "step": 1618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168959594368874e-05,
+      "loss": 0.7755,
+      "step": 1619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1689065612513542e-05,
+      "loss": 0.8404,
+      "step": 1620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1688534835838967e-05,
+      "loss": 0.7889,
+      "step": 1621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1688003613689415e-05,
+      "loss": 0.7377,
+      "step": 1622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1687471946089316e-05,
+      "loss": 0.7018,
+      "step": 1623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1686939833063108e-05,
+      "loss": 0.8096,
+      "step": 1624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1686407274635267e-05,
+      "loss": 0.8169,
+      "step": 1625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1685874270830275e-05,
+      "loss": 0.7823,
+      "step": 1626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1685340821672635e-05,
+      "loss": 0.7126,
+      "step": 1627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1684806927186885e-05,
+      "loss": 0.732,
+      "step": 1628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1684272587397564e-05,
+      "loss": 0.6526,
+      "step": 1629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1683737802329244e-05,
+      "loss": 0.759,
+      "step": 1630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168320257200652e-05,
+      "loss": 0.8361,
+      "step": 1631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168266689645399e-05,
+      "loss": 0.6207,
+      "step": 1632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1682130775696296e-05,
+      "loss": 0.7596,
+      "step": 1633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168159420975808e-05,
+      "loss": 0.7276,
+      "step": 1634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168105719866402e-05,
+      "loss": 0.6876,
+      "step": 1635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1680519742438797e-05,
+      "loss": 0.7392,
+      "step": 1636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1679981841107132e-05,
+      "loss": 0.7325,
+      "step": 1637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167944349469376e-05,
+      "loss": 0.6671,
+      "step": 1638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1678904703223423e-05,
+      "loss": 0.7736,
+      "step": 1639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1678365466720907e-05,
+      "loss": 0.6911,
+      "step": 1640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1677825785210993e-05,
+      "loss": 0.6982,
+      "step": 1641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1677285658718508e-05,
+      "loss": 0.8288,
+      "step": 1642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1676745087268278e-05,
+      "loss": 0.7412,
+      "step": 1643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1676204070885157e-05,
+      "loss": 0.7027,
+      "step": 1644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167566260959403e-05,
+      "loss": 0.7012,
+      "step": 1645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1675120703419785e-05,
+      "loss": 0.6851,
+      "step": 1646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1674578352387343e-05,
+      "loss": 0.7298,
+      "step": 1647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1674035556521638e-05,
+      "loss": 0.667,
+      "step": 1648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1673492315847632e-05,
+      "loss": 0.7018,
+      "step": 1649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1672948630390297e-05,
+      "loss": 0.8302,
+      "step": 1650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1672404500174633e-05,
+      "loss": 0.7645,
+      "step": 1651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1671859925225665e-05,
+      "loss": 0.7073,
+      "step": 1652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1671314905568428e-05,
+      "loss": 0.6903,
+      "step": 1653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1670769441227974e-05,
+      "loss": 0.7754,
+      "step": 1654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.16702235322294e-05,
+      "loss": 0.8356,
+      "step": 1655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1669677178597792e-05,
+      "loss": 0.7997,
+      "step": 1656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166913038035828e-05,
+      "loss": 0.7659,
+      "step": 1657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1668583137535997e-05,
+      "loss": 0.775,
+      "step": 1658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1668035450156113e-05,
+      "loss": 0.7596,
+      "step": 1659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1667487318243807e-05,
+      "loss": 0.7784,
+      "step": 1660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1666938741824284e-05,
+      "loss": 0.7832,
+      "step": 1661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1666389720922765e-05,
+      "loss": 0.8002,
+      "step": 1662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1665840255564495e-05,
+      "loss": 0.766,
+      "step": 1663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166529034577474e-05,
+      "loss": 0.7632,
+      "step": 1664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1664739991578777e-05,
+      "loss": 0.7427,
+      "step": 1665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166418919300192e-05,
+      "loss": 0.765,
+      "step": 1666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166363795006949e-05,
+      "loss": 0.7597,
+      "step": 1667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1663086262806833e-05,
+      "loss": 0.7541,
+      "step": 1668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166253413123932e-05,
+      "loss": 0.81,
+      "step": 1669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166198155539233e-05,
+      "loss": 0.7326,
+      "step": 1670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1661428535291272e-05,
+      "loss": 0.7233,
+      "step": 1671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166087507096158e-05,
+      "loss": 0.7945,
+      "step": 1672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.16603211624287e-05,
+      "loss": 0.7042,
+      "step": 1673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1659766809718092e-05,
+      "loss": 0.8058,
+      "step": 1674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1659212012855257e-05,
+      "loss": 0.6762,
+      "step": 1675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1658656771865698e-05,
+      "loss": 0.7248,
+      "step": 1676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1658101086774942e-05,
+      "loss": 0.756,
+      "step": 1677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1657544957608546e-05,
+      "loss": 0.7614,
+      "step": 1678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1656988384392076e-05,
+      "loss": 0.7888,
+      "step": 1679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1656431367151123e-05,
+      "loss": 0.7862,
+      "step": 1680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1655873905911302e-05,
+      "loss": 0.7505,
+      "step": 1681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1655316000698243e-05,
+      "loss": 0.7073,
+      "step": 1682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1654757651537594e-05,
+      "loss": 0.7485,
+      "step": 1683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1654198858455035e-05,
+      "loss": 0.6743,
+      "step": 1684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1653639621476255e-05,
+      "loss": 0.7984,
+      "step": 1685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1653079940626964e-05,
+      "loss": 0.8595,
+      "step": 1686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1652519815932905e-05,
+      "loss": 0.6621,
+      "step": 1687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1651959247419825e-05,
+      "loss": 0.7365,
+      "step": 1688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1651398235113502e-05,
+      "loss": 0.7791,
+      "step": 1689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165083677903973e-05,
+      "loss": 0.7354,
+      "step": 1690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1650274879224327e-05,
+      "loss": 0.7394,
+      "step": 1691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1649712535693123e-05,
+      "loss": 0.801,
+      "step": 1692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1649149748471977e-05,
+      "loss": 0.6792,
+      "step": 1693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164858651758677e-05,
+      "loss": 0.7471,
+      "step": 1694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1648022843063392e-05,
+      "loss": 0.7631,
+      "step": 1695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164745872492777e-05,
+      "loss": 0.7903,
+      "step": 1696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1646894163205832e-05,
+      "loss": 0.7996,
+      "step": 1697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1646329157923536e-05,
+      "loss": 0.7819,
+      "step": 1698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1645763709106873e-05,
+      "loss": 0.8009,
+      "step": 1699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164519781678183e-05,
+      "loss": 0.7344,
+      "step": 1700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1644631480974433e-05,
+      "loss": 0.8506,
+      "step": 1701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164406470171072e-05,
+      "loss": 0.7095,
+      "step": 1702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164349747901675e-05,
+      "loss": 0.7785,
+      "step": 1703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1642929812918606e-05,
+      "loss": 0.7391,
+      "step": 1704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1642361703442385e-05,
+      "loss": 0.7244,
+      "step": 1705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1641793150614212e-05,
+      "loss": 0.7239,
+      "step": 1706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164122415446023e-05,
+      "loss": 0.7048,
+      "step": 1707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1640654715006598e-05,
+      "loss": 0.7333,
+      "step": 1708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1640084832279503e-05,
+      "loss": 0.7528,
+      "step": 1709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1639514506305142e-05,
+      "loss": 0.8753,
+      "step": 1710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163894373710974e-05,
+      "loss": 0.6599,
+      "step": 1711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1638372524719543e-05,
+      "loss": 0.7753,
+      "step": 1712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1637800869160817e-05,
+      "loss": 0.7782,
+      "step": 1713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1637228770459837e-05,
+      "loss": 0.7732,
+      "step": 1714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163665622864292e-05,
+      "loss": 0.7388,
+      "step": 1715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163608324373638e-05,
+      "loss": 0.7883,
+      "step": 1716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1635509815766577e-05,
+      "loss": 0.7249,
+      "step": 1717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1634935944759862e-05,
+      "loss": 0.7651,
+      "step": 1718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163436163074263e-05,
+      "loss": 0.7533,
+      "step": 1719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1633786873741286e-05,
+      "loss": 0.7803,
+      "step": 1720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1633211673782253e-05,
+      "loss": 0.7326,
+      "step": 1721
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1632636030891985e-05,
+      "loss": 0.6816,
+      "step": 1722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1632059945096944e-05,
+      "loss": 0.7733,
+      "step": 1723
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1631483416423624e-05,
+      "loss": 0.7636,
+      "step": 1724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1630906444898532e-05,
+      "loss": 0.7738,
+      "step": 1725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163032903054819e-05,
+      "loss": 0.7918,
+      "step": 1726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1629751173399154e-05,
+      "loss": 0.7441,
+      "step": 1727
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1629172873477996e-05,
+      "loss": 0.7039,
+      "step": 1728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1628594130811298e-05,
+      "loss": 0.7839,
+      "step": 1729
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162801494542568e-05,
+      "loss": 0.8507,
+      "step": 1730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1627435317347763e-05,
+      "loss": 0.7663,
+      "step": 1731
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.16268552466042e-05,
+      "loss": 0.778,
+      "step": 1732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1626274733221676e-05,
+      "loss": 0.7551,
+      "step": 1733
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162569377722686e-05,
+      "loss": 0.6354,
+      "step": 1734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1625112378646482e-05,
+      "loss": 0.7186,
+      "step": 1735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162453053750727e-05,
+      "loss": 0.7973,
+      "step": 1736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1623948253835967e-05,
+      "loss": 0.6998,
+      "step": 1737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1623365527659362e-05,
+      "loss": 0.6866,
+      "step": 1738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1622782359004235e-05,
+      "loss": 0.7648,
+      "step": 1739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.162219874789741e-05,
+      "loss": 0.7693,
+      "step": 1740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1621614694365716e-05,
+      "loss": 0.7641,
+      "step": 1741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.162103019843601e-05,
+      "loss": 0.7142,
+      "step": 1742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1620445260135162e-05,
+      "loss": 0.7489,
+      "step": 1743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161985987949007e-05,
+      "loss": 0.7532,
+      "step": 1744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1619274056527652e-05,
+      "loss": 0.7739,
+      "step": 1745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1618687791274843e-05,
+      "loss": 0.7702,
+      "step": 1746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1618101083758597e-05,
+      "loss": 0.6496,
+      "step": 1747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1617513934005894e-05,
+      "loss": 0.7057,
+      "step": 1748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1616926342043726e-05,
+      "loss": 0.7391,
+      "step": 1749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1616338307899115e-05,
+      "loss": 0.7639,
+      "step": 1750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1615749831599093e-05,
+      "loss": 0.7222,
+      "step": 1751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1615160913170723e-05,
+      "loss": 0.7491,
+      "step": 1752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1614571552641085e-05,
+      "loss": 0.7096,
+      "step": 1753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1613981750037275e-05,
+      "loss": 0.7114,
+      "step": 1754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1613391505386407e-05,
+      "loss": 0.7396,
+      "step": 1755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1612800818715627e-05,
+      "loss": 0.7761,
+      "step": 1756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1612209690052088e-05,
+      "loss": 0.7281,
+      "step": 1757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1611618119422976e-05,
+      "loss": 0.7481,
+      "step": 1758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161102610685549e-05,
+      "loss": 0.7301,
+      "step": 1759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161043365237685e-05,
+      "loss": 0.7491,
+      "step": 1760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1609840756014294e-05,
+      "loss": 0.6814,
+      "step": 1761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1609247417795086e-05,
+      "loss": 0.7091,
+      "step": 1762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160865363774651e-05,
+      "loss": 0.785,
+      "step": 1763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160805941589586e-05,
+      "loss": 0.7655,
+      "step": 1764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1607464752270466e-05,
+      "loss": 0.7514,
+      "step": 1765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1606869646897666e-05,
+      "loss": 0.8048,
+      "step": 1766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1606274099804824e-05,
+      "loss": 0.8025,
+      "step": 1767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1605678111019323e-05,
+      "loss": 0.7739,
+      "step": 1768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1605081680568566e-05,
+      "loss": 0.8069,
+      "step": 1769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1604484808479978e-05,
+      "loss": 0.7977,
+      "step": 1770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1603887494781e-05,
+      "loss": 0.6806,
+      "step": 1771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1603289739499096e-05,
+      "loss": 0.7378,
+      "step": 1772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1602691542661755e-05,
+      "loss": 0.7048,
+      "step": 1773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160209290429648e-05,
+      "loss": 0.7725,
+      "step": 1774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1601493824430795e-05,
+      "loss": 0.815,
+      "step": 1775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1600894303092246e-05,
+      "loss": 0.6803,
+      "step": 1776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.16002943403084e-05,
+      "loss": 0.7257,
+      "step": 1777
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159969393610684e-05,
+      "loss": 0.7068,
+      "step": 1778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1599093090515176e-05,
+      "loss": 0.6657,
+      "step": 1779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1598491803561033e-05,
+      "loss": 0.7857,
+      "step": 1780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1597890075272055e-05,
+      "loss": 0.7343,
+      "step": 1781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1597287905675913e-05,
+      "loss": 0.7894,
+      "step": 1782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1596685294800295e-05,
+      "loss": 0.7379,
+      "step": 1783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1596082242672905e-05,
+      "loss": 0.7955,
+      "step": 1784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1595478749321476e-05,
+      "loss": 0.8227,
+      "step": 1785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1594874814773753e-05,
+      "loss": 0.7839,
+      "step": 1786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1594270439057504e-05,
+      "loss": 0.6827,
+      "step": 1787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1593665622200524e-05,
+      "loss": 0.7529,
+      "step": 1788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159306036423061e-05,
+      "loss": 0.7735,
+      "step": 1789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1592454665175605e-05,
+      "loss": 0.7968,
+      "step": 1790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159184852506335e-05,
+      "loss": 0.6772,
+      "step": 1791
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1591241943921722e-05,
+      "loss": 0.658,
+      "step": 1792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1590634921778603e-05,
+      "loss": 0.7739,
+      "step": 1793
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159002745866191e-05,
+      "loss": 0.636,
+      "step": 1794
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158941955459957e-05,
+      "loss": 0.6962,
+      "step": 1795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158881120961954e-05,
+      "loss": 0.7232,
+      "step": 1796
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1588202423749784e-05,
+      "loss": 0.6819,
+      "step": 1797
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15875931970183e-05,
+      "loss": 0.7834,
+      "step": 1798
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1586983529453097e-05,
+      "loss": 0.7558,
+      "step": 1799
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1586373421082206e-05,
+      "loss": 0.7477,
+      "step": 1800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1585762871933685e-05,
+      "loss": 0.7775,
+      "step": 1801
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15851518820356e-05,
+      "loss": 0.7835,
+      "step": 1802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1584540451416044e-05,
+      "loss": 0.6919,
+      "step": 1803
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1583928580103136e-05,
+      "loss": 0.7226,
+      "step": 1804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1583316268125007e-05,
+      "loss": 0.6942,
+      "step": 1805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1582703515509812e-05,
+      "loss": 0.7393,
+      "step": 1806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158209032228572e-05,
+      "loss": 0.7237,
+      "step": 1807
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1581476688480934e-05,
+      "loss": 0.7426,
+      "step": 1808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1580862614123666e-05,
+      "loss": 0.7751,
+      "step": 1809
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1580248099242143e-05,
+      "loss": 0.7354,
+      "step": 1810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157963314386463e-05,
+      "loss": 0.7168,
+      "step": 1811
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1579017748019395e-05,
+      "loss": 0.7318,
+      "step": 1812
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1578401911734737e-05,
+      "loss": 0.7499,
+      "step": 1813
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1577785635038975e-05,
+      "loss": 0.801,
+      "step": 1814
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1577168917960444e-05,
+      "loss": 0.7341,
+      "step": 1815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1576551760527494e-05,
+      "loss": 0.7244,
+      "step": 1816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157593416276851e-05,
+      "loss": 0.6872,
+      "step": 1817
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157531612471188e-05,
+      "loss": 0.7512,
+      "step": 1818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1574697646386028e-05,
+      "loss": 0.7136,
+      "step": 1819
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1574078727819393e-05,
+      "loss": 0.7862,
+      "step": 1820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1573459369040424e-05,
+      "loss": 0.6758,
+      "step": 1821
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1572839570077607e-05,
+      "loss": 0.746,
+      "step": 1822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157221933095944e-05,
+      "loss": 0.7358,
+      "step": 1823
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1571598651714434e-05,
+      "loss": 0.7681,
+      "step": 1824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1570977532371135e-05,
+      "loss": 0.7328,
+      "step": 1825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15703559729581e-05,
+      "loss": 0.6997,
+      "step": 1826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1569733973503902e-05,
+      "loss": 0.8019,
+      "step": 1827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.156911153403715e-05,
+      "loss": 0.7081,
+      "step": 1828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1568488654586458e-05,
+      "loss": 0.7166,
+      "step": 1829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1567865335180466e-05,
+      "loss": 0.7849,
+      "step": 1830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1567241575847832e-05,
+      "loss": 0.7583,
+      "step": 1831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1566617376617243e-05,
+      "loss": 0.677,
+      "step": 1832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1565992737517392e-05,
+      "loss": 0.7931,
+      "step": 1833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1565367658577005e-05,
+      "loss": 0.7217,
+      "step": 1834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.156474213982482e-05,
+      "loss": 0.7097,
+      "step": 1835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15641161812896e-05,
+      "loss": 0.6687,
+      "step": 1836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.156348978300012e-05,
+      "loss": 0.7226,
+      "step": 1837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1562862944985195e-05,
+      "loss": 0.7418,
+      "step": 1838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1562235667273634e-05,
+      "loss": 0.7946,
+      "step": 1839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1561607949894283e-05,
+      "loss": 0.7639,
+      "step": 1840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1560979792876007e-05,
+      "loss": 0.8068,
+      "step": 1841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1560351196247685e-05,
+      "loss": 0.7599,
+      "step": 1842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155972216003822e-05,
+      "loss": 0.7114,
+      "step": 1843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1559092684276535e-05,
+      "loss": 0.7809,
+      "step": 1844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1558462768991573e-05,
+      "loss": 0.6752,
+      "step": 1845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15578324142123e-05,
+      "loss": 0.7906,
+      "step": 1846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1557201619967694e-05,
+      "loss": 0.7138,
+      "step": 1847
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1556570386286768e-05,
+      "loss": 0.7682,
+      "step": 1848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1555938713198535e-05,
+      "loss": 0.7383,
+      "step": 1849
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1555306600732043e-05,
+      "loss": 0.8201,
+      "step": 1850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155467404891636e-05,
+      "loss": 0.7844,
+      "step": 1851
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155404105778057e-05,
+      "loss": 0.7395,
+      "step": 1852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155340762735377e-05,
+      "loss": 0.7841,
+      "step": 1853
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1552773757665094e-05,
+      "loss": 0.7041,
+      "step": 1854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1552139448743683e-05,
+      "loss": 0.7443,
+      "step": 1855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1551504700618697e-05,
+      "loss": 0.7525,
+      "step": 1856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1550869513319336e-05,
+      "loss": 0.8184,
+      "step": 1857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155023388687479e-05,
+      "loss": 0.6828,
+      "step": 1858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1549597821314292e-05,
+      "loss": 0.7128,
+      "step": 1859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1548961316667092e-05,
+      "loss": 0.7538,
+      "step": 1860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1548324372962448e-05,
+      "loss": 0.8105,
+      "step": 1861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1547686990229653e-05,
+      "loss": 0.7725,
+      "step": 1862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.154704916849801e-05,
+      "loss": 0.7294,
+      "step": 1863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1546410907796846e-05,
+      "loss": 0.6837,
+      "step": 1864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1545772208155507e-05,
+      "loss": 0.7346,
+      "step": 1865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1545133069603365e-05,
+      "loss": 0.6884,
+      "step": 1866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1544493492169803e-05,
+      "loss": 0.7444,
+      "step": 1867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1543853475884228e-05,
+      "loss": 0.6907,
+      "step": 1868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.154321302077607e-05,
+      "loss": 0.8312,
+      "step": 1869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1542572126874778e-05,
+      "loss": 0.7134,
+      "step": 1870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1541930794209818e-05,
+      "loss": 0.7269,
+      "step": 1871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1541289022810675e-05,
+      "loss": 0.678,
+      "step": 1872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1540646812706863e-05,
+      "loss": 0.7746,
+      "step": 1873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.154000416392791e-05,
+      "loss": 0.6846,
+      "step": 1874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153936107650336e-05,
+      "loss": 0.7242,
+      "step": 1875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1538717550462786e-05,
+      "loss": 0.8406,
+      "step": 1876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1538073585835778e-05,
+      "loss": 0.6975,
+      "step": 1877
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1537429182651943e-05,
+      "loss": 0.7836,
+      "step": 1878
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153678434094091e-05,
+      "loss": 0.6858,
+      "step": 1879
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153613906073233e-05,
+      "loss": 0.7302,
+      "step": 1880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153549334205587e-05,
+      "loss": 0.7058,
+      "step": 1881
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1534847184941222e-05,
+      "loss": 0.7737,
+      "step": 1882
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15342005894181e-05,
+      "loss": 0.7374,
+      "step": 1883
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1533553555516226e-05,
+      "loss": 0.74,
+      "step": 1884
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1532906083265356e-05,
+      "loss": 0.73,
+      "step": 1885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1532258172695258e-05,
+      "loss": 0.6477,
+      "step": 1886
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1531609823835727e-05,
+      "loss": 0.7125,
+      "step": 1887
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153096103671657e-05,
+      "loss": 0.7996,
+      "step": 1888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1530311811367615e-05,
+      "loss": 0.7309,
+      "step": 1889
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1529662147818716e-05,
+      "loss": 0.6822,
+      "step": 1890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1529012046099747e-05,
+      "loss": 0.8387,
+      "step": 1891
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.15283615062406e-05,
+      "loss": 0.7947,
+      "step": 1892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152771052827118e-05,
+      "loss": 0.7051,
+      "step": 1893
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1527059112221427e-05,
+      "loss": 0.6801,
+      "step": 1894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1526407258121287e-05,
+      "loss": 0.7153,
+      "step": 1895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1525754966000728e-05,
+      "loss": 0.7565,
+      "step": 1896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1525102235889752e-05,
+      "loss": 0.7307,
+      "step": 1897
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1524449067818368e-05,
+      "loss": 0.7248,
+      "step": 1898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152379546181661e-05,
+      "loss": 0.6647,
+      "step": 1899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152314141791452e-05,
+      "loss": 0.7996,
+      "step": 1900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1522486936142183e-05,
+      "loss": 0.7884,
+      "step": 1901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1521832016529687e-05,
+      "loss": 0.7746,
+      "step": 1902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1521176659107142e-05,
+      "loss": 0.7518,
+      "step": 1903
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1520520863904688e-05,
+      "loss": 0.7643,
+      "step": 1904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1519864630952476e-05,
+      "loss": 0.8102,
+      "step": 1905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1519207960280674e-05,
+      "loss": 0.7152,
+      "step": 1906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151855085191948e-05,
+      "loss": 0.7666,
+      "step": 1907
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1517893305899112e-05,
+      "loss": 0.7239,
+      "step": 1908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1517235322249792e-05,
+      "loss": 0.7242,
+      "step": 1909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151657690100178e-05,
+      "loss": 0.6917,
+      "step": 1910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1515918042185356e-05,
+      "loss": 0.7117,
+      "step": 1911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1515258745830805e-05,
+      "loss": 0.755,
+      "step": 1912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1514599011968443e-05,
+      "loss": 0.7533,
+      "step": 1913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151393884062861e-05,
+      "loss": 0.7632,
+      "step": 1914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1513278231841655e-05,
+      "loss": 0.7226,
+      "step": 1915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1512617185637954e-05,
+      "loss": 0.7161,
+      "step": 1916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.15119557020479e-05,
+      "loss": 0.7561,
+      "step": 1917
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1511293781101908e-05,
+      "loss": 0.7314,
+      "step": 1918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151063142283042e-05,
+      "loss": 0.7243,
+      "step": 1919
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1509968627263878e-05,
+      "loss": 0.7523,
+      "step": 1920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150930539443277e-05,
+      "loss": 0.7075,
+      "step": 1921
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1508641724367582e-05,
+      "loss": 0.7702,
+      "step": 1922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1507977617098832e-05,
+      "loss": 0.7394,
+      "step": 1923
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150731307265706e-05,
+      "loss": 0.743,
+      "step": 1924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1506648091072813e-05,
+      "loss": 0.6761,
+      "step": 1925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1505982672376672e-05,
+      "loss": 0.7025,
+      "step": 1926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1505316816599232e-05,
+      "loss": 0.6504,
+      "step": 1927
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150465052377111e-05,
+      "loss": 0.7372,
+      "step": 1928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150398379392294e-05,
+      "loss": 0.7477,
+      "step": 1929
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150331662708538e-05,
+      "loss": 0.6886,
+      "step": 1930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1502649023289104e-05,
+      "loss": 0.7191,
+      "step": 1931
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1501980982564806e-05,
+      "loss": 0.7311,
+      "step": 1932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1501312504943204e-05,
+      "loss": 0.7977,
+      "step": 1933
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150064359045504e-05,
+      "loss": 0.7205,
+      "step": 1934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149997423913106e-05,
+      "loss": 0.7583,
+      "step": 1935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1499304451002053e-05,
+      "loss": 0.7365,
+      "step": 1936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1498634226098804e-05,
+      "loss": 0.8054,
+      "step": 1937
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1497963564452136e-05,
+      "loss": 0.794,
+      "step": 1938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1497292466092883e-05,
+      "loss": 0.7358,
+      "step": 1939
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1496620931051902e-05,
+      "loss": 0.7552,
+      "step": 1940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1495948959360073e-05,
+      "loss": 0.7168,
+      "step": 1941
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149527655104829e-05,
+      "loss": 0.7321,
+      "step": 1942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1494603706147472e-05,
+      "loss": 0.739,
+      "step": 1943
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1493930424688557e-05,
+      "loss": 0.6741,
+      "step": 1944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1493256706702494e-05,
+      "loss": 0.6637,
+      "step": 1945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1492582552220273e-05,
+      "loss": 0.6794,
+      "step": 1946
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149190796127288e-05,
+      "loss": 0.6922,
+      "step": 1947
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1491232933891344e-05,
+      "loss": 0.763,
+      "step": 1948
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149055747010669e-05,
+      "loss": 0.7098,
+      "step": 1949
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148988156994998e-05,
+      "loss": 0.6867,
+      "step": 1950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.14892052334523e-05,
+      "loss": 0.7257,
+      "step": 1951
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1488528460644736e-05,
+      "loss": 0.7858,
+      "step": 1952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148785125155841e-05,
+      "loss": 0.7321,
+      "step": 1953
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148717360622446e-05,
+      "loss": 0.7087,
+      "step": 1954
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1486495524674052e-05,
+      "loss": 0.7275,
+      "step": 1955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148581700693835e-05,
+      "loss": 0.7674,
+      "step": 1956
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1485138053048555e-05,
+      "loss": 0.8018,
+      "step": 1957
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1484458663035893e-05,
+      "loss": 0.7629,
+      "step": 1958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1483778836931594e-05,
+      "loss": 0.7098,
+      "step": 1959
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148309857476692e-05,
+      "loss": 0.7347,
+      "step": 1960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148241787657315e-05,
+      "loss": 0.7361,
+      "step": 1961
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148173674238158e-05,
+      "loss": 0.7931,
+      "step": 1962
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1481055172223527e-05,
+      "loss": 0.7276,
+      "step": 1963
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1480373166130335e-05,
+      "loss": 0.728,
+      "step": 1964
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1479690724133355e-05,
+      "loss": 0.6118,
+      "step": 1965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147900784626397e-05,
+      "loss": 0.7128,
+      "step": 1966
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147832453255358e-05,
+      "loss": 0.7514,
+      "step": 1967
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.14776407830336e-05,
+      "loss": 0.7332,
+      "step": 1968
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1476956597735468e-05,
+      "loss": 0.7496,
+      "step": 1969
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1476271976690646e-05,
+      "loss": 0.8179,
+      "step": 1970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147558691993061e-05,
+      "loss": 0.6655,
+      "step": 1971
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147490142748686e-05,
+      "loss": 0.7297,
+      "step": 1972
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1474215499390913e-05,
+      "loss": 0.706,
+      "step": 1973
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147352913567431e-05,
+      "loss": 0.8128,
+      "step": 1974
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1472842336368605e-05,
+      "loss": 0.6734,
+      "step": 1975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1472155101505384e-05,
+      "loss": 0.6902,
+      "step": 1976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1471467431116242e-05,
+      "loss": 0.7573,
+      "step": 1977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1470779325232795e-05,
+      "loss": 0.7038,
+      "step": 1978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1470090783886687e-05,
+      "loss": 0.7645,
+      "step": 1979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146940180710957e-05,
+      "loss": 0.6895,
+      "step": 1980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1468712394933136e-05,
+      "loss": 0.7202,
+      "step": 1981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146802254738907e-05,
+      "loss": 0.732,
+      "step": 1982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1467332264509095e-05,
+      "loss": 0.7871,
+      "step": 1983
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1466641546324956e-05,
+      "loss": 0.7012,
+      "step": 1984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.14659503928684e-05,
+      "loss": 0.6915,
+      "step": 1985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146525880417122e-05,
+      "loss": 0.7803,
+      "step": 1986
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1464566780265204e-05,
+      "loss": 0.756,
+      "step": 1987
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1463874321182176e-05,
+      "loss": 0.7895,
+      "step": 1988
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1463181426953977e-05,
+      "loss": 0.7737,
+      "step": 1989
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146248809761246e-05,
+      "loss": 0.7863,
+      "step": 1990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1461794333189506e-05,
+      "loss": 0.7986,
+      "step": 1991
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1461100133717016e-05,
+      "loss": 0.7683,
+      "step": 1992
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1460405499226907e-05,
+      "loss": 0.7366,
+      "step": 1993
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1459710429751122e-05,
+      "loss": 0.7014,
+      "step": 1994
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1459014925321618e-05,
+      "loss": 0.7265,
+      "step": 1995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1458318985970366e-05,
+      "loss": 0.7256,
+      "step": 1996
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145762261172938e-05,
+      "loss": 0.7161,
+      "step": 1997
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1456925802630667e-05,
+      "loss": 0.6577,
+      "step": 1998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145622855870627e-05,
+      "loss": 0.727,
+      "step": 1999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1455530879988248e-05,
+      "loss": 0.8036,
+      "step": 2000
+    },
+    {
+      "epoch": 0.13,
+      "eval_text_loss": 0.6650489568710327,
+      "eval_text_runtime": 1.8687,
+      "eval_text_samples_per_second": 39.599,
+      "eval_text_steps_per_second": 1.07,
+      "step": 2000
+    },
+    {
+      "epoch": 0.13,
+      "eval_svit_loss": 0.9054301977157593,
+      "eval_svit_runtime": 6.5757,
+      "eval_svit_samples_per_second": 124.093,
+      "eval_svit_steps_per_second": 1.977,
+      "step": 2000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145483276650868e-05,
+      "loss": 0.7144,
+      "step": 2001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1454134218299667e-05,
+      "loss": 0.7338,
+      "step": 2002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1453435235393325e-05,
+      "loss": 0.7783,
+      "step": 2003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1452735817821794e-05,
+      "loss": 0.7086,
+      "step": 2004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1452035965617234e-05,
+      "loss": 0.6683,
+      "step": 2005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1451335678811823e-05,
+      "loss": 0.7251,
+      "step": 2006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145063495743776e-05,
+      "loss": 0.7537,
+      "step": 2007
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144993380152726e-05,
+      "loss": 0.6351,
+      "step": 2008
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144923221111257e-05,
+      "loss": 0.7588,
+      "step": 2009
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144853018622594e-05,
+      "loss": 0.7244,
+      "step": 2010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1447827726899656e-05,
+      "loss": 0.7182,
+      "step": 2011
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1447124833166017e-05,
+      "loss": 0.7311,
+      "step": 2012
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1446421505057336e-05,
+      "loss": 0.7846,
+      "step": 2013
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1445717742605952e-05,
+      "loss": 0.7833,
+      "step": 2014
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1445013545844228e-05,
+      "loss": 0.7215,
+      "step": 2015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1444308914804544e-05,
+      "loss": 0.7853,
+      "step": 2016
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144360384951929e-05,
+      "loss": 0.7457,
+      "step": 2017
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1442898350020894e-05,
+      "loss": 0.6898,
+      "step": 2018
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144219241634179e-05,
+      "loss": 0.763,
+      "step": 2019
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1441486048514435e-05,
+      "loss": 0.6818,
+      "step": 2020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1440779246571317e-05,
+      "loss": 0.7484,
+      "step": 2021
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144007201054492e-05,
+      "loss": 0.7412,
+      "step": 2022
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1439364340467775e-05,
+      "loss": 0.7782,
+      "step": 2023
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143865623637241e-05,
+      "loss": 0.6759,
+      "step": 2024
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1437947698291392e-05,
+      "loss": 0.6906,
+      "step": 2025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1437238726257293e-05,
+      "loss": 0.8019,
+      "step": 2026
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1436529320302717e-05,
+      "loss": 0.7606,
+      "step": 2027
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1435819480460277e-05,
+      "loss": 0.7961,
+      "step": 2028
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1435109206762615e-05,
+      "loss": 0.7609,
+      "step": 2029
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143439849924239e-05,
+      "loss": 0.7339,
+      "step": 2030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1433687357932273e-05,
+      "loss": 0.7354,
+      "step": 2031
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1432975782864968e-05,
+      "loss": 0.6956,
+      "step": 2032
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1432263774073192e-05,
+      "loss": 0.7335,
+      "step": 2033
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143155133158968e-05,
+      "loss": 0.7909,
+      "step": 2034
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1430838455447195e-05,
+      "loss": 0.7138,
+      "step": 2035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143012514567851e-05,
+      "loss": 0.6849,
+      "step": 2036
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1429411402316427e-05,
+      "loss": 0.72,
+      "step": 2037
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1428697225393762e-05,
+      "loss": 0.7751,
+      "step": 2038
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1427982614943348e-05,
+      "loss": 0.7552,
+      "step": 2039
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142726757099805e-05,
+      "loss": 0.794,
+      "step": 2040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142655209359074e-05,
+      "loss": 0.7459,
+      "step": 2041
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1425836182754317e-05,
+      "loss": 0.7456,
+      "step": 2042
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.14251198385217e-05,
+      "loss": 0.7369,
+      "step": 2043
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1424403060925826e-05,
+      "loss": 0.6967,
+      "step": 2044
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1423685849999645e-05,
+      "loss": 0.7645,
+      "step": 2045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1422968205776145e-05,
+      "loss": 0.7624,
+      "step": 2046
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1422250128288317e-05,
+      "loss": 0.6893,
+      "step": 2047
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1421531617569177e-05,
+      "loss": 0.7567,
+      "step": 2048
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1420812673651764e-05,
+      "loss": 0.6872,
+      "step": 2049
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1420093296569138e-05,
+      "loss": 0.8271,
+      "step": 2050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1419373486354365e-05,
+      "loss": 0.7736,
+      "step": 2051
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1418653243040553e-05,
+      "loss": 0.7245,
+      "step": 2052
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1417932566660818e-05,
+      "loss": 0.7499,
+      "step": 2053
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1417211457248286e-05,
+      "loss": 0.7552,
+      "step": 2054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1416489914836123e-05,
+      "loss": 0.691,
+      "step": 2055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1415767939457502e-05,
+      "loss": 0.7105,
+      "step": 2056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141504553114562e-05,
+      "loss": 0.6953,
+      "step": 2057
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1414322689933692e-05,
+      "loss": 0.701,
+      "step": 2058
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1413599415854954e-05,
+      "loss": 0.7083,
+      "step": 2059
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1412875708942658e-05,
+      "loss": 0.7804,
+      "step": 2060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1412151569230092e-05,
+      "loss": 0.8136,
+      "step": 2061
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1411426996750538e-05,
+      "loss": 0.7527,
+      "step": 2062
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1410701991537318e-05,
+      "loss": 0.7684,
+      "step": 2063
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1409976553623768e-05,
+      "loss": 0.8011,
+      "step": 2064
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140925068304324e-05,
+      "loss": 0.7393,
+      "step": 2065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1408524379829112e-05,
+      "loss": 0.7837,
+      "step": 2066
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140779764401478e-05,
+      "loss": 0.7814,
+      "step": 2067
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1407070475633654e-05,
+      "loss": 0.6661,
+      "step": 2068
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1406342874719175e-05,
+      "loss": 0.7045,
+      "step": 2069
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1405614841304794e-05,
+      "loss": 0.713,
+      "step": 2070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1404886375423983e-05,
+      "loss": 0.7548,
+      "step": 2071
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1404157477110247e-05,
+      "loss": 0.7412,
+      "step": 2072
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140342814639709e-05,
+      "loss": 0.7396,
+      "step": 2073
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140269838331805e-05,
+      "loss": 0.6795,
+      "step": 2074
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1401968187906682e-05,
+      "loss": 0.7884,
+      "step": 2075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1401237560196558e-05,
+      "loss": 0.7411,
+      "step": 2076
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1400506500221276e-05,
+      "loss": 0.7264,
+      "step": 2077
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1399775008014446e-05,
+      "loss": 0.8299,
+      "step": 2078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.13990430836097e-05,
+      "loss": 0.7491,
+      "step": 2079
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.13983107270407e-05,
+      "loss": 0.8049,
+      "step": 2080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139757793834111e-05,
+      "loss": 0.7544,
+      "step": 2081
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139684471754463e-05,
+      "loss": 0.7325,
+      "step": 2082
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139611106468497e-05,
+      "loss": 0.6625,
+      "step": 2083
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139537697979586e-05,
+      "loss": 0.8803,
+      "step": 2084
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139464246291106e-05,
+      "loss": 0.6829,
+      "step": 2085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1393907514064337e-05,
+      "loss": 0.7903,
+      "step": 2086
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1393172133289488e-05,
+      "loss": 0.7737,
+      "step": 2087
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1392436320620325e-05,
+      "loss": 0.7352,
+      "step": 2088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1391700076090674e-05,
+      "loss": 0.7273,
+      "step": 2089
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1390963399734395e-05,
+      "loss": 0.746,
+      "step": 2090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1390226291585358e-05,
+      "loss": 0.7464,
+      "step": 2091
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138948875167745e-05,
+      "loss": 0.743,
+      "step": 2092
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1388750780044592e-05,
+      "loss": 0.8072,
+      "step": 2093
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138801237672071e-05,
+      "loss": 0.6223,
+      "step": 2094
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1387273541739753e-05,
+      "loss": 0.7042,
+      "step": 2095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1386534275135697e-05,
+      "loss": 0.699,
+      "step": 2096
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1385794576942534e-05,
+      "loss": 0.6789,
+      "step": 2097
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1385054447194267e-05,
+      "loss": 0.7698,
+      "step": 2098
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1384313885924936e-05,
+      "loss": 0.7461,
+      "step": 2099
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1383572893168585e-05,
+      "loss": 0.7679,
+      "step": 2100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1382831468959293e-05,
+      "loss": 0.8199,
+      "step": 2101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1382089613331143e-05,
+      "loss": 0.6845,
+      "step": 2102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1381347326318247e-05,
+      "loss": 0.7658,
+      "step": 2103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1380604607954734e-05,
+      "loss": 0.7096,
+      "step": 2104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1379861458274754e-05,
+      "loss": 0.84,
+      "step": 2105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.137911787731248e-05,
+      "loss": 0.696,
+      "step": 2106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.13783738651021e-05,
+      "loss": 0.7753,
+      "step": 2107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1377629421677822e-05,
+      "loss": 0.7574,
+      "step": 2108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1376884547073876e-05,
+      "loss": 0.7721,
+      "step": 2109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.137613924132451e-05,
+      "loss": 0.8185,
+      "step": 2110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1375393504464e-05,
+      "loss": 0.7147,
+      "step": 2111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1374647336526622e-05,
+      "loss": 0.7377,
+      "step": 2112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1373900737546697e-05,
+      "loss": 0.7603,
+      "step": 2113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1373153707558543e-05,
+      "loss": 0.7932,
+      "step": 2114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1372406246596514e-05,
+      "loss": 0.7516,
+      "step": 2115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1371658354694978e-05,
+      "loss": 0.786,
+      "step": 2116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1370910031888324e-05,
+      "loss": 0.7552,
+      "step": 2117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.137016127821095e-05,
+      "loss": 0.7312,
+      "step": 2118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1369412093697296e-05,
+      "loss": 0.7396,
+      "step": 2119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.13686624783818e-05,
+      "loss": 0.7377,
+      "step": 2120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1367912432298933e-05,
+      "loss": 0.7261,
+      "step": 2121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1367161955483184e-05,
+      "loss": 0.6735,
+      "step": 2122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1366411047969054e-05,
+      "loss": 0.7458,
+      "step": 2123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1365659709791074e-05,
+      "loss": 0.687,
+      "step": 2124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.136490794098379e-05,
+      "loss": 0.6851,
+      "step": 2125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1364155741581765e-05,
+      "loss": 0.7401,
+      "step": 2126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1363403111619587e-05,
+      "loss": 0.7397,
+      "step": 2127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1362650051131863e-05,
+      "loss": 0.7868,
+      "step": 2128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1361896560153217e-05,
+      "loss": 0.7334,
+      "step": 2129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.136114263871829e-05,
+      "loss": 0.7681,
+      "step": 2130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1360388286861755e-05,
+      "loss": 0.7029,
+      "step": 2131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1359633504618287e-05,
+      "loss": 0.7614,
+      "step": 2132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.13588782920226e-05,
+      "loss": 0.8066,
+      "step": 2133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1358122649109416e-05,
+      "loss": 0.7179,
+      "step": 2134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1357366575913476e-05,
+      "loss": 0.7132,
+      "step": 2135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1356610072469546e-05,
+      "loss": 0.776,
+      "step": 2136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1355853138812408e-05,
+      "loss": 0.7644,
+      "step": 2137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.135509577497687e-05,
+      "loss": 0.8082,
+      "step": 2138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.135433798099775e-05,
+      "loss": 0.6595,
+      "step": 2139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1353579756909893e-05,
+      "loss": 0.7794,
+      "step": 2140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1352821102748167e-05,
+      "loss": 0.7394,
+      "step": 2141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1352062018547444e-05,
+      "loss": 0.821,
+      "step": 2142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1351302504342633e-05,
+      "loss": 0.7432,
+      "step": 2143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1350542560168658e-05,
+      "loss": 0.7873,
+      "step": 2144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1349782186060457e-05,
+      "loss": 0.7401,
+      "step": 2145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134902138205299e-05,
+      "loss": 0.7486,
+      "step": 2146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1348260148181246e-05,
+      "loss": 0.7388,
+      "step": 2147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134749848448022e-05,
+      "loss": 0.7047,
+      "step": 2148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1346736390984935e-05,
+      "loss": 0.6877,
+      "step": 2149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1345973867730433e-05,
+      "loss": 0.7403,
+      "step": 2150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134521091475177e-05,
+      "loss": 0.857,
+      "step": 2151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1344447532084028e-05,
+      "loss": 0.7525,
+      "step": 2152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134368371976231e-05,
+      "loss": 0.7471,
+      "step": 2153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1342919477821735e-05,
+      "loss": 0.6634,
+      "step": 2154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1342154806297443e-05,
+      "loss": 0.8594,
+      "step": 2155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1341389705224587e-05,
+      "loss": 0.6662,
+      "step": 2156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1340624174638355e-05,
+      "loss": 0.7552,
+      "step": 2157
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133985821457394e-05,
+      "loss": 0.7403,
+      "step": 2158
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1339091825066563e-05,
+      "loss": 0.7681,
+      "step": 2159
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1338325006151462e-05,
+      "loss": 0.7418,
+      "step": 2160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1337557757863894e-05,
+      "loss": 0.7132,
+      "step": 2161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1336790080239134e-05,
+      "loss": 0.7271,
+      "step": 2162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1336021973312486e-05,
+      "loss": 0.7153,
+      "step": 2163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1335253437119264e-05,
+      "loss": 0.6868,
+      "step": 2164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1334484471694803e-05,
+      "loss": 0.7938,
+      "step": 2165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1333715077074465e-05,
+      "loss": 0.713,
+      "step": 2166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1332945253293622e-05,
+      "loss": 0.7797,
+      "step": 2167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133217500038767e-05,
+      "loss": 0.6628,
+      "step": 2168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133140431839203e-05,
+      "loss": 0.789,
+      "step": 2169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133063320734213e-05,
+      "loss": 0.7429,
+      "step": 2170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.132986166727343e-05,
+      "loss": 0.7194,
+      "step": 2171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1329089698221407e-05,
+      "loss": 0.7492,
+      "step": 2172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1328317300221552e-05,
+      "loss": 0.8018,
+      "step": 2173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.132754447330938e-05,
+      "loss": 0.7584,
+      "step": 2174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1326771217520427e-05,
+      "loss": 0.72,
+      "step": 2175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1325997532890244e-05,
+      "loss": 0.7647,
+      "step": 2176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1325223419454412e-05,
+      "loss": 0.7058,
+      "step": 2177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1324448877248517e-05,
+      "loss": 0.7015,
+      "step": 2178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.132367390630817e-05,
+      "loss": 0.8216,
+      "step": 2179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1322898506669014e-05,
+      "loss": 0.7748,
+      "step": 2180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1322122678366692e-05,
+      "loss": 0.6948,
+      "step": 2181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1321346421436884e-05,
+      "loss": 0.7729,
+      "step": 2182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1320569735915275e-05,
+      "loss": 0.7832,
+      "step": 2183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.131979262183758e-05,
+      "loss": 0.752,
+      "step": 2184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.131901507923953e-05,
+      "loss": 0.7044,
+      "step": 2185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1318237108156878e-05,
+      "loss": 0.7269,
+      "step": 2186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.131745870862539e-05,
+      "loss": 0.7203,
+      "step": 2187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1316679880680865e-05,
+      "loss": 0.7163,
+      "step": 2188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1315900624359106e-05,
+      "loss": 0.7552,
+      "step": 2189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.131512093969594e-05,
+      "loss": 0.7576,
+      "step": 2190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1314340826727226e-05,
+      "loss": 0.7413,
+      "step": 2191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.131356028548883e-05,
+      "loss": 0.7005,
+      "step": 2192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1312779316016632e-05,
+      "loss": 0.7118,
+      "step": 2193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1311997918346552e-05,
+      "loss": 0.7436,
+      "step": 2194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1311216092514518e-05,
+      "loss": 0.6837,
+      "step": 2195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.131043383855647e-05,
+      "loss": 0.7556,
+      "step": 2196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1309651156508385e-05,
+      "loss": 0.6953,
+      "step": 2197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.130886804640624e-05,
+      "loss": 0.7677,
+      "step": 2198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1308084508286053e-05,
+      "loss": 0.7246,
+      "step": 2199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1307300542183843e-05,
+      "loss": 0.6778,
+      "step": 2200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1306516148135658e-05,
+      "loss": 0.7362,
+      "step": 2201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1305731326177568e-05,
+      "loss": 0.6436,
+      "step": 2202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1304946076345654e-05,
+      "loss": 0.7397,
+      "step": 2203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1304160398676025e-05,
+      "loss": 0.785,
+      "step": 2204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.13033742932048e-05,
+      "loss": 0.7522,
+      "step": 2205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1302587759968134e-05,
+      "loss": 0.7578,
+      "step": 2206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1301800799002184e-05,
+      "loss": 0.7775,
+      "step": 2207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1301013410343136e-05,
+      "loss": 0.7446,
+      "step": 2208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1300225594027192e-05,
+      "loss": 0.813,
+      "step": 2209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129943735009058e-05,
+      "loss": 0.6775,
+      "step": 2210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129864867856954e-05,
+      "loss": 0.8223,
+      "step": 2211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1297859579500338e-05,
+      "loss": 0.7017,
+      "step": 2212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129707005291925e-05,
+      "loss": 0.7025,
+      "step": 2213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129628009886258e-05,
+      "loss": 0.6326,
+      "step": 2214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1295489717366653e-05,
+      "loss": 0.7458,
+      "step": 2215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1294698908467812e-05,
+      "loss": 0.6824,
+      "step": 2216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1293907672202415e-05,
+      "loss": 0.7681,
+      "step": 2217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129311600860684e-05,
+      "loss": 0.8073,
+      "step": 2218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129232391771749e-05,
+      "loss": 0.7307,
+      "step": 2219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1291531399570787e-05,
+      "loss": 0.7002,
+      "step": 2220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129073845420317e-05,
+      "loss": 0.719,
+      "step": 2221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1289945081651093e-05,
+      "loss": 0.6588,
+      "step": 2222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1289151281951046e-05,
+      "loss": 0.7611,
+      "step": 2223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1288357055139514e-05,
+      "loss": 0.7812,
+      "step": 2224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1287562401253024e-05,
+      "loss": 0.799,
+      "step": 2225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1286767320328115e-05,
+      "loss": 0.6827,
+      "step": 2226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128597181240134e-05,
+      "loss": 0.7565,
+      "step": 2227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1285175877509276e-05,
+      "loss": 0.7192,
+      "step": 2228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128437951568852e-05,
+      "loss": 0.7428,
+      "step": 2229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1283582726975695e-05,
+      "loss": 0.6736,
+      "step": 2230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128278551140743e-05,
+      "loss": 0.7342,
+      "step": 2231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128198786902038e-05,
+      "loss": 0.766,
+      "step": 2232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1281189799851222e-05,
+      "loss": 0.8152,
+      "step": 2233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1280391303936654e-05,
+      "loss": 0.7606,
+      "step": 2234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1279592381313386e-05,
+      "loss": 0.7985,
+      "step": 2235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1278793032018155e-05,
+      "loss": 0.7615,
+      "step": 2236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1277993256087716e-05,
+      "loss": 0.7075,
+      "step": 2237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127719305355884e-05,
+      "loss": 0.7344,
+      "step": 2238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127639242446832e-05,
+      "loss": 0.7408,
+      "step": 2239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1275591368852967e-05,
+      "loss": 0.676,
+      "step": 2240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1274789886749616e-05,
+      "loss": 0.7836,
+      "step": 2241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127398797819512e-05,
+      "loss": 0.7777,
+      "step": 2242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1273185643226345e-05,
+      "loss": 0.722,
+      "step": 2243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1272382881880188e-05,
+      "loss": 0.7489,
+      "step": 2244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1271579694193554e-05,
+      "loss": 0.7391,
+      "step": 2245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1270776080203382e-05,
+      "loss": 0.7607,
+      "step": 2246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1269972039946614e-05,
+      "loss": 0.6937,
+      "step": 2247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.126916757346022e-05,
+      "loss": 0.7271,
+      "step": 2248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1268362680781194e-05,
+      "loss": 0.7123,
+      "step": 2249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1267557361946538e-05,
+      "loss": 0.7087,
+      "step": 2250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1266751616993283e-05,
+      "loss": 0.7016,
+      "step": 2251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.126594544595848e-05,
+      "loss": 0.78,
+      "step": 2252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1265138848879192e-05,
+      "loss": 0.7346,
+      "step": 2253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1264331825792512e-05,
+      "loss": 0.6489,
+      "step": 2254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1263524376735538e-05,
+      "loss": 0.7804,
+      "step": 2255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1262716501745404e-05,
+      "loss": 0.6973,
+      "step": 2256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.126190820085925e-05,
+      "loss": 0.6857,
+      "step": 2257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1261099474114247e-05,
+      "loss": 0.7151,
+      "step": 2258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.126029032154757e-05,
+      "loss": 0.741,
+      "step": 2259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1259480743196436e-05,
+      "loss": 0.6956,
+      "step": 2260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1258670739098065e-05,
+      "loss": 0.6621,
+      "step": 2261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1257860309289696e-05,
+      "loss": 0.7754,
+      "step": 2262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1257049453808592e-05,
+      "loss": 0.7614,
+      "step": 2263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1256238172692046e-05,
+      "loss": 0.8164,
+      "step": 2264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125542646597735e-05,
+      "loss": 0.6596,
+      "step": 2265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1254614333701826e-05,
+      "loss": 0.7606,
+      "step": 2266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1253801775902823e-05,
+      "loss": 0.8014,
+      "step": 2267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.12529887926177e-05,
+      "loss": 0.8071,
+      "step": 2268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125217538388383e-05,
+      "loss": 0.7069,
+      "step": 2269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125136154973862e-05,
+      "loss": 0.7499,
+      "step": 2270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1250547290219493e-05,
+      "loss": 0.6594,
+      "step": 2271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1249732605363885e-05,
+      "loss": 0.7696,
+      "step": 2272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124891749520925e-05,
+      "loss": 0.671,
+      "step": 2273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1248101959793067e-05,
+      "loss": 0.6891,
+      "step": 2274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1247285999152842e-05,
+      "loss": 0.7116,
+      "step": 2275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1246469613326087e-05,
+      "loss": 0.6624,
+      "step": 2276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124565280235034e-05,
+      "loss": 0.7471,
+      "step": 2277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1244835566263158e-05,
+      "loss": 0.7554,
+      "step": 2278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1244017905102113e-05,
+      "loss": 0.6535,
+      "step": 2279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124319981890481e-05,
+      "loss": 0.7541,
+      "step": 2280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1242381307708855e-05,
+      "loss": 0.7498,
+      "step": 2281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124156237155189e-05,
+      "loss": 0.7724,
+      "step": 2282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1240743010471562e-05,
+      "loss": 0.717,
+      "step": 2283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123992322450555e-05,
+      "loss": 0.6466,
+      "step": 2284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123910301369155e-05,
+      "loss": 0.7292,
+      "step": 2285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1238282378067264e-05,
+      "loss": 0.8106,
+      "step": 2286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1237461317670438e-05,
+      "loss": 0.7131,
+      "step": 2287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1236639832538817e-05,
+      "loss": 0.7429,
+      "step": 2288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123581792271017e-05,
+      "loss": 0.6931,
+      "step": 2289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1234995588222293e-05,
+      "loss": 0.6696,
+      "step": 2290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1234172829112997e-05,
+      "loss": 0.6754,
+      "step": 2291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1233349645420108e-05,
+      "loss": 0.7366,
+      "step": 2292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123252603718148e-05,
+      "loss": 0.7484,
+      "step": 2293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123170200443498e-05,
+      "loss": 0.8319,
+      "step": 2294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1230877547218494e-05,
+      "loss": 0.7641,
+      "step": 2295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1230052665569935e-05,
+      "loss": 0.6581,
+      "step": 2296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1229227359527228e-05,
+      "loss": 0.7376,
+      "step": 2297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.122840162912832e-05,
+      "loss": 0.7238,
+      "step": 2298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1227575474411178e-05,
+      "loss": 0.6646,
+      "step": 2299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.122674889541379e-05,
+      "loss": 0.7114,
+      "step": 2300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1225921892174162e-05,
+      "loss": 0.7789,
+      "step": 2301
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1225094464730317e-05,
+      "loss": 0.8211,
+      "step": 2302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.12242666131203e-05,
+      "loss": 0.7248,
+      "step": 2303
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.122343833738218e-05,
+      "loss": 0.7711,
+      "step": 2304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1222609637554032e-05,
+      "loss": 0.7936,
+      "step": 2305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1221780513673965e-05,
+      "loss": 0.7141,
+      "step": 2306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.12209509657801e-05,
+      "loss": 0.7574,
+      "step": 2307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1220120993910583e-05,
+      "loss": 0.7497,
+      "step": 2308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1219290598103575e-05,
+      "loss": 0.7336,
+      "step": 2309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121845977839725e-05,
+      "loss": 0.745,
+      "step": 2310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1217628534829815e-05,
+      "loss": 0.7564,
+      "step": 2311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121679686743949e-05,
+      "loss": 0.6815,
+      "step": 2312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1215964776264516e-05,
+      "loss": 0.701,
+      "step": 2313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1215132261343146e-05,
+      "loss": 0.6723,
+      "step": 2314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1214299322713666e-05,
+      "loss": 0.7552,
+      "step": 2315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1213465960414372e-05,
+      "loss": 0.7415,
+      "step": 2316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121263217448358e-05,
+      "loss": 0.7485,
+      "step": 2317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1211797964959626e-05,
+      "loss": 0.7445,
+      "step": 2318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121096333188087e-05,
+      "loss": 0.7893,
+      "step": 2319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1210128275285686e-05,
+      "loss": 0.7415,
+      "step": 2320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120929279521247e-05,
+      "loss": 0.6963,
+      "step": 2321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120845689169964e-05,
+      "loss": 0.7773,
+      "step": 2322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1207620564785623e-05,
+      "loss": 0.6955,
+      "step": 2323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1206783814508883e-05,
+      "loss": 0.7495,
+      "step": 2324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120594664090788e-05,
+      "loss": 0.7726,
+      "step": 2325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120510904402112e-05,
+      "loss": 0.7033,
+      "step": 2326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1204271023887115e-05,
+      "loss": 0.7015,
+      "step": 2327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120343258054439e-05,
+      "loss": 0.726,
+      "step": 2328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1202593714031495e-05,
+      "loss": 0.8025,
+      "step": 2329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1201754424387005e-05,
+      "loss": 0.6388,
+      "step": 2330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120091471164951e-05,
+      "loss": 0.7011,
+      "step": 2331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120007457585762e-05,
+      "loss": 0.6929,
+      "step": 2332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.119923401704996e-05,
+      "loss": 0.7937,
+      "step": 2333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1198393035265188e-05,
+      "loss": 0.7466,
+      "step": 2334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1197551630541962e-05,
+      "loss": 0.7225,
+      "step": 2335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1196709802918974e-05,
+      "loss": 0.7724,
+      "step": 2336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.119586755243493e-05,
+      "loss": 0.6938,
+      "step": 2337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.119502487912856e-05,
+      "loss": 0.7903,
+      "step": 2338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.11941817830386e-05,
+      "loss": 0.7024,
+      "step": 2339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1193338264203827e-05,
+      "loss": 0.6605,
+      "step": 2340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1192494322663016e-05,
+      "loss": 0.7087,
+      "step": 2341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1191649958454978e-05,
+      "loss": 0.6731,
+      "step": 2342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1190805171618532e-05,
+      "loss": 0.7156,
+      "step": 2343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1189959962192524e-05,
+      "loss": 0.7505,
+      "step": 2344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1189114330215817e-05,
+      "loss": 0.7602,
+      "step": 2345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1188268275727287e-05,
+      "loss": 0.741,
+      "step": 2346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.118742179876584e-05,
+      "loss": 0.7951,
+      "step": 2347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11865748993704e-05,
+      "loss": 0.7724,
+      "step": 2348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1185727577579898e-05,
+      "loss": 0.8135,
+      "step": 2349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11848798334333e-05,
+      "loss": 0.6655,
+      "step": 2350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1184031666969586e-05,
+      "loss": 0.7364,
+      "step": 2351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.118318307822775e-05,
+      "loss": 0.6773,
+      "step": 2352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.118233406724681e-05,
+      "loss": 0.6872,
+      "step": 2353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1181484634065808e-05,
+      "loss": 0.8085,
+      "step": 2354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1180634778723797e-05,
+      "loss": 0.7721,
+      "step": 2355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1179784501259854e-05,
+      "loss": 0.7056,
+      "step": 2356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117893380171307e-05,
+      "loss": 0.6793,
+      "step": 2357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1178082680122568e-05,
+      "loss": 0.7259,
+      "step": 2358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1177231136527477e-05,
+      "loss": 0.7256,
+      "step": 2359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1176379170966948e-05,
+      "loss": 0.7839,
+      "step": 2360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117552678348016e-05,
+      "loss": 0.8141,
+      "step": 2361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1174673974106302e-05,
+      "loss": 0.6709,
+      "step": 2362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117382074288459e-05,
+      "loss": 0.7617,
+      "step": 2363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1172967089854246e-05,
+      "loss": 0.7088,
+      "step": 2364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1172113015054532e-05,
+      "loss": 0.7879,
+      "step": 2365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1171258518524714e-05,
+      "loss": 0.7118,
+      "step": 2366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1170403600304076e-05,
+      "loss": 0.7632,
+      "step": 2367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1169548260431934e-05,
+      "loss": 0.7467,
+      "step": 2368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1168692498947608e-05,
+      "loss": 0.6265,
+      "step": 2369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1167836315890454e-05,
+      "loss": 0.7267,
+      "step": 2370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.116697971129984e-05,
+      "loss": 0.7114,
+      "step": 2371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.116612268521514e-05,
+      "loss": 0.7472,
+      "step": 2372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1165265237675774e-05,
+      "loss": 0.7409,
+      "step": 2373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1164407368721156e-05,
+      "loss": 0.6414,
+      "step": 2374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.116354907839074e-05,
+      "loss": 0.6667,
+      "step": 2375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.116269036672398e-05,
+      "loss": 0.6726,
+      "step": 2376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.116183123376037e-05,
+      "loss": 0.7981,
+      "step": 2377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1160971679539402e-05,
+      "loss": 0.698,
+      "step": 2378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1160111704100607e-05,
+      "loss": 0.7243,
+      "step": 2379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115925130748352e-05,
+      "loss": 0.7303,
+      "step": 2380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11583904897277e-05,
+      "loss": 0.7802,
+      "step": 2381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1157529250872736e-05,
+      "loss": 0.6813,
+      "step": 2382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1156667590958224e-05,
+      "loss": 0.7048,
+      "step": 2383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1155805510023778e-05,
+      "loss": 0.6485,
+      "step": 2384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1154943008109037e-05,
+      "loss": 0.81,
+      "step": 2385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1154080085253666e-05,
+      "loss": 0.782,
+      "step": 2386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1153216741497334e-05,
+      "loss": 0.8404,
+      "step": 2387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115235297687974e-05,
+      "loss": 0.7938,
+      "step": 2388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11514887914406e-05,
+      "loss": 0.7346,
+      "step": 2389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1150624185219647e-05,
+      "loss": 0.8117,
+      "step": 2390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114975915825664e-05,
+      "loss": 0.7459,
+      "step": 2391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1148893710591343e-05,
+      "loss": 0.7567,
+      "step": 2392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114802784226356e-05,
+      "loss": 0.7009,
+      "step": 2393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11471615533131e-05,
+      "loss": 0.752,
+      "step": 2394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114629484377979e-05,
+      "loss": 0.7257,
+      "step": 2395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1145427713703483e-05,
+      "loss": 0.7119,
+      "step": 2396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114456016312405e-05,
+      "loss": 0.7218,
+      "step": 2397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1143692192081384e-05,
+      "loss": 0.697,
+      "step": 2398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114282380061539e-05,
+      "loss": 0.7778,
+      "step": 2399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1141954988765994e-05,
+      "loss": 0.7779,
+      "step": 2400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1141085756573148e-05,
+      "loss": 0.7284,
+      "step": 2401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114021610407682e-05,
+      "loss": 0.6691,
+      "step": 2402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1139346031316992e-05,
+      "loss": 0.767,
+      "step": 2403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1138475538333673e-05,
+      "loss": 0.7156,
+      "step": 2404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1137604625166887e-05,
+      "loss": 0.73,
+      "step": 2405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1136733291856675e-05,
+      "loss": 0.7358,
+      "step": 2406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1135861538443105e-05,
+      "loss": 0.7205,
+      "step": 2407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1134989364966256e-05,
+      "loss": 0.7718,
+      "step": 2408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1134116771466236e-05,
+      "loss": 0.6748,
+      "step": 2409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.113324375798316e-05,
+      "loss": 0.7869,
+      "step": 2410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1132370324557176e-05,
+      "loss": 0.7981,
+      "step": 2411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1131496471228435e-05,
+      "loss": 0.6995,
+      "step": 2412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1130622198037124e-05,
+      "loss": 0.673,
+      "step": 2413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.112974750502344e-05,
+      "loss": 0.7524,
+      "step": 2414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1128872392227595e-05,
+      "loss": 0.7026,
+      "step": 2415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1127996859689836e-05,
+      "loss": 0.7761,
+      "step": 2416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1127120907450413e-05,
+      "loss": 0.7351,
+      "step": 2417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1126244535549603e-05,
+      "loss": 0.7061,
+      "step": 2418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1125367744027706e-05,
+      "loss": 0.6584,
+      "step": 2419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1124490532925026e-05,
+      "loss": 0.7234,
+      "step": 2420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.112361290228191e-05,
+      "loss": 0.6889,
+      "step": 2421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1122734852138703e-05,
+      "loss": 0.7832,
+      "step": 2422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1121856382535778e-05,
+      "loss": 0.7274,
+      "step": 2423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.112097749351353e-05,
+      "loss": 0.7971,
+      "step": 2424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.112009818511236e-05,
+      "loss": 0.7299,
+      "step": 2425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1119218457372715e-05,
+      "loss": 0.7614,
+      "step": 2426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.111833831033503e-05,
+      "loss": 0.7882,
+      "step": 2427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1117457744039783e-05,
+      "loss": 0.6997,
+      "step": 2428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1116576758527455e-05,
+      "loss": 0.7287,
+      "step": 2429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.111569535383856e-05,
+      "loss": 0.664,
+      "step": 2430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.111481353001362e-05,
+      "loss": 0.704,
+      "step": 2431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.111393128709318e-05,
+      "loss": 0.7119,
+      "step": 2432
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1113048625117807e-05,
+      "loss": 0.7856,
+      "step": 2433
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.111216554412809e-05,
+      "loss": 0.7289,
+      "step": 2434
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1111282044164626e-05,
+      "loss": 0.7863,
+      "step": 2435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.111039812526804e-05,
+      "loss": 0.7236,
+      "step": 2436
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1109513787478976e-05,
+      "loss": 0.6912,
+      "step": 2437
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1108629030838097e-05,
+      "loss": 0.6459,
+      "step": 2438
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1107743855386077e-05,
+      "loss": 0.7301,
+      "step": 2439
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.110685826116362e-05,
+      "loss": 0.7884,
+      "step": 2440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1105972248211446e-05,
+      "loss": 0.7257,
+      "step": 2441
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1105085816570296e-05,
+      "loss": 0.6759,
+      "step": 2442
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.110419896628092e-05,
+      "loss": 0.7584,
+      "step": 2443
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11033116973841e-05,
+      "loss": 0.7026,
+      "step": 2444
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1102424009920635e-05,
+      "loss": 0.7736,
+      "step": 2445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1101535903931335e-05,
+      "loss": 0.6923,
+      "step": 2446
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1100647379457043e-05,
+      "loss": 0.7793,
+      "step": 2447
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.10997584365386e-05,
+      "loss": 0.8073,
+      "step": 2448
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.109886907521689e-05,
+      "loss": 0.7609,
+      "step": 2449
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.10979792955328e-05,
+      "loss": 0.7472,
+      "step": 2450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1097089097527243e-05,
+      "loss": 0.658,
+      "step": 2451
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1096198481241153e-05,
+      "loss": 0.7238,
+      "step": 2452
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1095307446715476e-05,
+      "loss": 0.7159,
+      "step": 2453
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1094415993991182e-05,
+      "loss": 0.747,
+      "step": 2454
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1093524123109266e-05,
+      "loss": 0.7945,
+      "step": 2455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1092631834110725e-05,
+      "loss": 0.7614,
+      "step": 2456
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1091739127036595e-05,
+      "loss": 0.7608,
+      "step": 2457
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.109084600192792e-05,
+      "loss": 0.7851,
+      "step": 2458
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1089952458825762e-05,
+      "loss": 0.7092,
+      "step": 2459
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.108905849777121e-05,
+      "loss": 0.743,
+      "step": 2460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1088164118805364e-05,
+      "loss": 0.6531,
+      "step": 2461
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1087269321969352e-05,
+      "loss": 0.7036,
+      "step": 2462
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1086374107304313e-05,
+      "loss": 0.8006,
+      "step": 2463
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.108547847485141e-05,
+      "loss": 0.7158,
+      "step": 2464
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1084582424651825e-05,
+      "loss": 0.7837,
+      "step": 2465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1083685956746755e-05,
+      "loss": 0.6184,
+      "step": 2466
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.108278907117742e-05,
+      "loss": 0.7627,
+      "step": 2467
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.108189176798506e-05,
+      "loss": 0.7274,
+      "step": 2468
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.108099404721093e-05,
+      "loss": 0.755,
+      "step": 2469
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1080095908896308e-05,
+      "loss": 0.7665,
+      "step": 2470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1079197353082494e-05,
+      "loss": 0.7479,
+      "step": 2471
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.10782983798108e-05,
+      "loss": 0.8097,
+      "step": 2472
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1077398989122555e-05,
+      "loss": 0.7503,
+      "step": 2473
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.107649918105912e-05,
+      "loss": 0.7443,
+      "step": 2474
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.107559895566187e-05,
+      "loss": 0.6781,
+      "step": 2475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.107469831297219e-05,
+      "loss": 0.8096,
+      "step": 2476
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1073797253031496e-05,
+      "loss": 0.708,
+      "step": 2477
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1072895775881213e-05,
+      "loss": 0.7156,
+      "step": 2478
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1071993881562796e-05,
+      "loss": 0.6922,
+      "step": 2479
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.107109157011771e-05,
+      "loss": 0.679,
+      "step": 2480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1070188841587444e-05,
+      "loss": 0.8055,
+      "step": 2481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106928569601351e-05,
+      "loss": 0.6892,
+      "step": 2482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1068382133437424e-05,
+      "loss": 0.7018,
+      "step": 2483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106747815390074e-05,
+      "loss": 0.7559,
+      "step": 2484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1066573757445022e-05,
+      "loss": 0.7862,
+      "step": 2485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106566894411185e-05,
+      "loss": 0.6596,
+      "step": 2486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106476371394283e-05,
+      "loss": 0.756,
+      "step": 2487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106385806697958e-05,
+      "loss": 0.7176,
+      "step": 2488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1062952003263743e-05,
+      "loss": 0.7188,
+      "step": 2489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1062045522836984e-05,
+      "loss": 0.7505,
+      "step": 2490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106113862574098e-05,
+      "loss": 0.7109,
+      "step": 2491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1060231312017427e-05,
+      "loss": 0.7446,
+      "step": 2492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1059323581708045e-05,
+      "loss": 0.7615,
+      "step": 2493
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105841543485457e-05,
+      "loss": 0.7326,
+      "step": 2494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105750687149876e-05,
+      "loss": 0.7232,
+      "step": 2495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1056597891682395e-05,
+      "loss": 0.6954,
+      "step": 2496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1055688495447256e-05,
+      "loss": 0.6978,
+      "step": 2497
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105477868283517e-05,
+      "loss": 0.708,
+      "step": 2498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1053868453887962e-05,
+      "loss": 0.7569,
+      "step": 2499
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105295780864749e-05,
+      "loss": 0.7253,
+      "step": 2500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1052046747155617e-05,
+      "loss": 0.7289,
+      "step": 2501
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105113526945424e-05,
+      "loss": 0.7038,
+      "step": 2502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1050223375585264e-05,
+      "loss": 0.7905,
+      "step": 2503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104931106559062e-05,
+      "loss": 0.7652,
+      "step": 2504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104839833951226e-05,
+      "loss": 0.6823,
+      "step": 2505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1047485197392145e-05,
+      "loss": 0.7099,
+      "step": 2506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1046571639272262e-05,
+      "loss": 0.7062,
+      "step": 2507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1045657665194614e-05,
+      "loss": 0.7242,
+      "step": 2508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1044743275201228e-05,
+      "loss": 0.7435,
+      "step": 2509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104382846933415e-05,
+      "loss": 0.814,
+      "step": 2510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1042913247635436e-05,
+      "loss": 0.6528,
+      "step": 2511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1041997610147168e-05,
+      "loss": 0.7185,
+      "step": 2512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1041081556911453e-05,
+      "loss": 0.6558,
+      "step": 2513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1040165087970406e-05,
+      "loss": 0.7034,
+      "step": 2514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1039248203366168e-05,
+      "loss": 0.7515,
+      "step": 2515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.103833090314089e-05,
+      "loss": 0.7998,
+      "step": 2516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1037413187336764e-05,
+      "loss": 0.6704,
+      "step": 2517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1036495055995973e-05,
+      "loss": 0.7214,
+      "step": 2518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1035576509160734e-05,
+      "loss": 0.7263,
+      "step": 2519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1034657546873286e-05,
+      "loss": 0.6452,
+      "step": 2520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1033738169175884e-05,
+      "loss": 0.7907,
+      "step": 2521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1032818376110793e-05,
+      "loss": 0.7658,
+      "step": 2522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.103189816772031e-05,
+      "loss": 0.7314,
+      "step": 2523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1030977544046744e-05,
+      "loss": 0.7286,
+      "step": 2524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1030056505132425e-05,
+      "loss": 0.8075,
+      "step": 2525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1029135051019702e-05,
+      "loss": 0.7439,
+      "step": 2526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1028213181750945e-05,
+      "loss": 0.672,
+      "step": 2527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.102729089736854e-05,
+      "loss": 0.7719,
+      "step": 2528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1026368197914894e-05,
+      "loss": 0.7081,
+      "step": 2529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.102544508343243e-05,
+      "loss": 0.7309,
+      "step": 2530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1024521553963592e-05,
+      "loss": 0.7881,
+      "step": 2531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1023597609550847e-05,
+      "loss": 0.7372,
+      "step": 2532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1022673250236674e-05,
+      "loss": 0.7046,
+      "step": 2533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1021748476063582e-05,
+      "loss": 0.6664,
+      "step": 2534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1020823287074082e-05,
+      "loss": 0.7086,
+      "step": 2535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1019897683310718e-05,
+      "loss": 0.7989,
+      "step": 2536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1018971664816052e-05,
+      "loss": 0.7932,
+      "step": 2537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1018045231632656e-05,
+      "loss": 0.7007,
+      "step": 2538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1017118383803132e-05,
+      "loss": 0.7637,
+      "step": 2539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1016191121370093e-05,
+      "loss": 0.7467,
+      "step": 2540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.101526344437618e-05,
+      "loss": 0.7321,
+      "step": 2541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1014335352864035e-05,
+      "loss": 0.7542,
+      "step": 2542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1013406846876342e-05,
+      "loss": 0.7647,
+      "step": 2543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1012477926455794e-05,
+      "loss": 0.7799,
+      "step": 2544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1011548591645095e-05,
+      "loss": 0.7581,
+      "step": 2545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1010618842486977e-05,
+      "loss": 0.7306,
+      "step": 2546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1009688679024193e-05,
+      "loss": 0.7317,
+      "step": 2547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.100875810129951e-05,
+      "loss": 0.7159,
+      "step": 2548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1007827109355717e-05,
+      "loss": 0.7475,
+      "step": 2549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.100689570323562e-05,
+      "loss": 0.7105,
+      "step": 2550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1005963882982043e-05,
+      "loss": 0.7468,
+      "step": 2551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.100503164863783e-05,
+      "loss": 0.6786,
+      "step": 2552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1004099000245848e-05,
+      "loss": 0.7091,
+      "step": 2553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1003165937848977e-05,
+      "loss": 0.7436,
+      "step": 2554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.100223246149012e-05,
+      "loss": 0.7407,
+      "step": 2555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.10012985712122e-05,
+      "loss": 0.7052,
+      "step": 2556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1000364267058156e-05,
+      "loss": 0.7765,
+      "step": 2557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.099942954907094e-05,
+      "loss": 0.776,
+      "step": 2558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.099849441729354e-05,
+      "loss": 0.7418,
+      "step": 2559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0997558871768947e-05,
+      "loss": 0.6351,
+      "step": 2560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0996622912540183e-05,
+      "loss": 0.7143,
+      "step": 2561
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0995686539650272e-05,
+      "loss": 0.7622,
+      "step": 2562
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.099474975314228e-05,
+      "loss": 0.7808,
+      "step": 2563
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0993812553059272e-05,
+      "loss": 0.7054,
+      "step": 2564
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0992874939444346e-05,
+      "loss": 0.723,
+      "step": 2565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0991936912340607e-05,
+      "loss": 0.7383,
+      "step": 2566
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.099099847179119e-05,
+      "loss": 0.7935,
+      "step": 2567
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.099005961783924e-05,
+      "loss": 0.7078,
+      "step": 2568
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0989120350527933e-05,
+      "loss": 0.7179,
+      "step": 2569
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0988180669900444e-05,
+      "loss": 0.7062,
+      "step": 2570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098724057599999e-05,
+      "loss": 0.7034,
+      "step": 2571
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0986300068869792e-05,
+      "loss": 0.6444,
+      "step": 2572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098535914855309e-05,
+      "loss": 0.7871,
+      "step": 2573
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0984417815093153e-05,
+      "loss": 0.7171,
+      "step": 2574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0983476068533264e-05,
+      "loss": 0.6926,
+      "step": 2575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098253390891672e-05,
+      "loss": 0.7084,
+      "step": 2576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098159133628684e-05,
+      "loss": 0.787,
+      "step": 2577
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098064835068697e-05,
+      "loss": 0.7741,
+      "step": 2578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.097970495216046e-05,
+      "loss": 0.7503,
+      "step": 2579
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0978761140750692e-05,
+      "loss": 0.7098,
+      "step": 2580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.097781691650106e-05,
+      "loss": 0.7642,
+      "step": 2581
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0976872279454983e-05,
+      "loss": 0.7195,
+      "step": 2582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0975927229655888e-05,
+      "loss": 0.8103,
+      "step": 2583
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0974981767147237e-05,
+      "loss": 0.745,
+      "step": 2584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0974035891972492e-05,
+      "loss": 0.7856,
+      "step": 2585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0973089604175155e-05,
+      "loss": 0.6581,
+      "step": 2586
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0972142903798724e-05,
+      "loss": 0.7889,
+      "step": 2587
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0971195790886734e-05,
+      "loss": 0.694,
+      "step": 2588
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0970248265482737e-05,
+      "loss": 0.7927,
+      "step": 2589
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0969300327630293e-05,
+      "loss": 0.7135,
+      "step": 2590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.096835197737299e-05,
+      "loss": 0.648,
+      "step": 2591
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0967403214754433e-05,
+      "loss": 0.6594,
+      "step": 2592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0966454039818247e-05,
+      "loss": 0.7604,
+      "step": 2593
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0965504452608073e-05,
+      "loss": 0.6775,
+      "step": 2594
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0964554453167577e-05,
+      "loss": 0.6548,
+      "step": 2595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0963604041540428e-05,
+      "loss": 0.7709,
+      "step": 2596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0962653217770338e-05,
+      "loss": 0.7448,
+      "step": 2597
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0961701981901022e-05,
+      "loss": 0.6593,
+      "step": 2598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0960750333976215e-05,
+      "loss": 0.7095,
+      "step": 2599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095979827403967e-05,
+      "loss": 0.697,
+      "step": 2600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0958845802135173e-05,
+      "loss": 0.7266,
+      "step": 2601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095789291830651e-05,
+      "loss": 0.7175,
+      "step": 2602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0956939622597496e-05,
+      "loss": 0.7535,
+      "step": 2603
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095598591505197e-05,
+      "loss": 0.754,
+      "step": 2604
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095503179571377e-05,
+      "loss": 0.6832,
+      "step": 2605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095407726462677e-05,
+      "loss": 0.8169,
+      "step": 2606
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095312232183487e-05,
+      "loss": 0.7042,
+      "step": 2607
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0952166967381963e-05,
+      "loss": 0.829,
+      "step": 2608
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0951211201311986e-05,
+      "loss": 0.7324,
+      "step": 2609
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0950255023668877e-05,
+      "loss": 0.7433,
+      "step": 2610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.094929843449661e-05,
+      "loss": 0.731,
+      "step": 2611
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0948341433839162e-05,
+      "loss": 0.7594,
+      "step": 2612
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0947384021740535e-05,
+      "loss": 0.8136,
+      "step": 2613
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0946426198244756e-05,
+      "loss": 0.7329,
+      "step": 2614
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0945467963395858e-05,
+      "loss": 0.7244,
+      "step": 2615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0944509317237905e-05,
+      "loss": 0.784,
+      "step": 2616
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0943550259814977e-05,
+      "loss": 0.7526,
+      "step": 2617
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0942590791171166e-05,
+      "loss": 0.7419,
+      "step": 2618
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0941630911350586e-05,
+      "loss": 0.7749,
+      "step": 2619
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0940670620397382e-05,
+      "loss": 0.7139,
+      "step": 2620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.09397099183557e-05,
+      "loss": 0.7199,
+      "step": 2621
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0938748805269715e-05,
+      "loss": 0.6786,
+      "step": 2622
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0937787281183618e-05,
+      "loss": 0.715,
+      "step": 2623
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.093682534614162e-05,
+      "loss": 0.7532,
+      "step": 2624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.093586300018795e-05,
+      "loss": 0.7239,
+      "step": 2625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0934900243366855e-05,
+      "loss": 0.7728,
+      "step": 2626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.09339370757226e-05,
+      "loss": 0.7586,
+      "step": 2627
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0932973497299485e-05,
+      "loss": 0.7618,
+      "step": 2628
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0932009508141797e-05,
+      "loss": 0.7857,
+      "step": 2629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0931045108293867e-05,
+      "loss": 0.724,
+      "step": 2630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.093008029780004e-05,
+      "loss": 0.7841,
+      "step": 2631
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0929115076704678e-05,
+      "loss": 0.7091,
+      "step": 2632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0928149445052155e-05,
+      "loss": 0.7196,
+      "step": 2633
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092718340288688e-05,
+      "loss": 0.7184,
+      "step": 2634
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092621695025326e-05,
+      "loss": 0.748,
+      "step": 2635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092525008719574e-05,
+      "loss": 0.7268,
+      "step": 2636
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0924282813758775e-05,
+      "loss": 0.7066,
+      "step": 2637
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092331512998684e-05,
+      "loss": 0.7739,
+      "step": 2638
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0922347035924422e-05,
+      "loss": 0.7418,
+      "step": 2639
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0921378531616044e-05,
+      "loss": 0.6272,
+      "step": 2640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092040961710623e-05,
+      "loss": 0.7417,
+      "step": 2641
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0919440292439532e-05,
+      "loss": 0.7645,
+      "step": 2642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0918470557660523e-05,
+      "loss": 0.7592,
+      "step": 2643
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0917500412813786e-05,
+      "loss": 0.7825,
+      "step": 2644
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0916529857943932e-05,
+      "loss": 0.7201,
+      "step": 2645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.091555889309558e-05,
+      "loss": 0.7392,
+      "step": 2646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0914587518313385e-05,
+      "loss": 0.7624,
+      "step": 2647
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0913615733642e-05,
+      "loss": 0.7672,
+      "step": 2648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0912643539126115e-05,
+      "loss": 0.72,
+      "step": 2649
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0911670934810426e-05,
+      "loss": 0.6915,
+      "step": 2650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.091069792073965e-05,
+      "loss": 0.8128,
+      "step": 2651
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.090972449695854e-05,
+      "loss": 0.7293,
+      "step": 2652
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.090875066351184e-05,
+      "loss": 0.7112,
+      "step": 2653
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.090777642044433e-05,
+      "loss": 0.779,
+      "step": 2654
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0906801767800808e-05,
+      "loss": 0.7273,
+      "step": 2655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0905826705626086e-05,
+      "loss": 0.6748,
+      "step": 2656
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0904851233964994e-05,
+      "loss": 0.665,
+      "step": 2657
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0903875352862392e-05,
+      "loss": 0.7024,
+      "step": 2658
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0902899062363143e-05,
+      "loss": 0.7646,
+      "step": 2659
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.090192236251214e-05,
+      "loss": 0.7662,
+      "step": 2660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0900945253354288e-05,
+      "loss": 0.7707,
+      "step": 2661
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.089996773493452e-05,
+      "loss": 0.767,
+      "step": 2662
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0898989807297775e-05,
+      "loss": 0.7353,
+      "step": 2663
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0898011470489023e-05,
+      "loss": 0.7414,
+      "step": 2664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0897032724553244e-05,
+      "loss": 0.7674,
+      "step": 2665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0896053569535443e-05,
+      "loss": 0.7871,
+      "step": 2666
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.089507400548064e-05,
+      "loss": 0.6833,
+      "step": 2667
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0894094032433878e-05,
+      "loss": 0.6821,
+      "step": 2668
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.089311365044021e-05,
+      "loss": 0.6888,
+      "step": 2669
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.089213285954472e-05,
+      "loss": 0.7675,
+      "step": 2670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0891151659792493e-05,
+      "loss": 0.7168,
+      "step": 2671
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0890170051228658e-05,
+      "loss": 0.6297,
+      "step": 2672
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.088918803389834e-05,
+      "loss": 0.6889,
+      "step": 2673
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0888205607846697e-05,
+      "loss": 0.6826,
+      "step": 2674
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0887222773118896e-05,
+      "loss": 0.6811,
+      "step": 2675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0886239529760135e-05,
+      "loss": 0.7427,
+      "step": 2676
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0885255877815614e-05,
+      "loss": 0.685,
+      "step": 2677
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.088427181733056e-05,
+      "loss": 0.77,
+      "step": 2678
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0883287348350233e-05,
+      "loss": 0.7778,
+      "step": 2679
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0882302470919887e-05,
+      "loss": 0.6788,
+      "step": 2680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0881317185084806e-05,
+      "loss": 0.7258,
+      "step": 2681
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08803314908903e-05,
+      "loss": 0.7265,
+      "step": 2682
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0879345388381687e-05,
+      "loss": 0.8239,
+      "step": 2683
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0878358877604303e-05,
+      "loss": 0.8004,
+      "step": 2684
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0877371958603516e-05,
+      "loss": 0.7479,
+      "step": 2685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08763846314247e-05,
+      "loss": 0.7401,
+      "step": 2686
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.087539689611325e-05,
+      "loss": 0.7853,
+      "step": 2687
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0874408752714585e-05,
+      "loss": 0.7311,
+      "step": 2688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0873420201274138e-05,
+      "loss": 0.7068,
+      "step": 2689
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.087243124183736e-05,
+      "loss": 0.7059,
+      "step": 2690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0871441874449734e-05,
+      "loss": 0.7159,
+      "step": 2691
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0870452099156734e-05,
+      "loss": 0.7311,
+      "step": 2692
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.086946191600388e-05,
+      "loss": 0.6877,
+      "step": 2693
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0868471325036695e-05,
+      "loss": 0.7874,
+      "step": 2694
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0867480326300735e-05,
+      "loss": 0.7957,
+      "step": 2695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0866488919841553e-05,
+      "loss": 0.7618,
+      "step": 2696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0865497105704742e-05,
+      "loss": 0.7445,
+      "step": 2697
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0864504883935907e-05,
+      "loss": 0.698,
+      "step": 2698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0863512254580662e-05,
+      "loss": 0.7972,
+      "step": 2699
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0862519217684653e-05,
+      "loss": 0.7389,
+      "step": 2700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.086152577329354e-05,
+      "loss": 0.6812,
+      "step": 2701
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0860531921452994e-05,
+      "loss": 0.7695,
+      "step": 2702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0859537662208722e-05,
+      "loss": 0.7091,
+      "step": 2703
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0858542995606435e-05,
+      "loss": 0.7733,
+      "step": 2704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0857547921691868e-05,
+      "loss": 0.6891,
+      "step": 2705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.085655244051077e-05,
+      "loss": 0.7106,
+      "step": 2706
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.085555655210892e-05,
+      "loss": 0.7407,
+      "step": 2707
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08545602565321e-05,
+      "loss": 0.7329,
+      "step": 2708
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0853563553826128e-05,
+      "loss": 0.7362,
+      "step": 2709
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0852566444036827e-05,
+      "loss": 0.7406,
+      "step": 2710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0851568927210048e-05,
+      "loss": 0.713,
+      "step": 2711
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.085057100339165e-05,
+      "loss": 0.7888,
+      "step": 2712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0849572672627523e-05,
+      "loss": 0.6885,
+      "step": 2713
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084857393496356e-05,
+      "loss": 0.7357,
+      "step": 2714
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0847574790445697e-05,
+      "loss": 0.6907,
+      "step": 2715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0846575239119867e-05,
+      "loss": 0.7403,
+      "step": 2716
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0845575281032028e-05,
+      "loss": 0.7329,
+      "step": 2717
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0844574916228155e-05,
+      "loss": 0.7652,
+      "step": 2718
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084357414475425e-05,
+      "loss": 0.7262,
+      "step": 2719
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084257296665633e-05,
+      "loss": 0.6164,
+      "step": 2720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0841571381980422e-05,
+      "loss": 0.6248,
+      "step": 2721
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0840569390772582e-05,
+      "loss": 0.7717,
+      "step": 2722
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083956699307888e-05,
+      "loss": 0.7836,
+      "step": 2723
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0838564188945406e-05,
+      "loss": 0.7786,
+      "step": 2724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083756097841827e-05,
+      "loss": 0.735,
+      "step": 2725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0836557361543598e-05,
+      "loss": 0.7357,
+      "step": 2726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0835553338367533e-05,
+      "loss": 0.7539,
+      "step": 2727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083454890893625e-05,
+      "loss": 0.6722,
+      "step": 2728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083354407329592e-05,
+      "loss": 0.742,
+      "step": 2729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083253883149275e-05,
+      "loss": 0.7779,
+      "step": 2730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0831533183572964e-05,
+      "loss": 0.7508,
+      "step": 2731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083052712958279e-05,
+      "loss": 0.7423,
+      "step": 2732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0829520669568502e-05,
+      "loss": 0.7632,
+      "step": 2733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0828513803576363e-05,
+      "loss": 0.6715,
+      "step": 2734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0827506531652677e-05,
+      "loss": 0.756,
+      "step": 2735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0826498853843752e-05,
+      "loss": 0.7047,
+      "step": 2736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0825490770195924e-05,
+      "loss": 0.7095,
+      "step": 2737
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0824482280755546e-05,
+      "loss": 0.732,
+      "step": 2738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0823473385568982e-05,
+      "loss": 0.7649,
+      "step": 2739
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0822464084682628e-05,
+      "loss": 0.8123,
+      "step": 2740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0821454378142885e-05,
+      "loss": 0.7294,
+      "step": 2741
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.082044426599618e-05,
+      "loss": 0.7862,
+      "step": 2742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.081943374828896e-05,
+      "loss": 0.7623,
+      "step": 2743
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0818422825067686e-05,
+      "loss": 0.8124,
+      "step": 2744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0817411496378846e-05,
+      "loss": 0.6781,
+      "step": 2745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.081639976226893e-05,
+      "loss": 0.7474,
+      "step": 2746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0815387622784463e-05,
+      "loss": 0.7367,
+      "step": 2747
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0814375077971985e-05,
+      "loss": 0.783,
+      "step": 2748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0813362127878047e-05,
+      "loss": 0.6625,
+      "step": 2749
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.081234877254923e-05,
+      "loss": 0.6476,
+      "step": 2750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0811335012032117e-05,
+      "loss": 0.7251,
+      "step": 2751
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0810320846373333e-05,
+      "loss": 0.6801,
+      "step": 2752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0809306275619504e-05,
+      "loss": 0.726,
+      "step": 2753
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.080829129981728e-05,
+      "loss": 0.6759,
+      "step": 2754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0807275919013324e-05,
+      "loss": 0.66,
+      "step": 2755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.080626013325433e-05,
+      "loss": 0.7543,
+      "step": 2756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0805243942587e-05,
+      "loss": 0.8122,
+      "step": 2757
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.080422734705806e-05,
+      "loss": 0.7275,
+      "step": 2758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0803210346714253e-05,
+      "loss": 0.7576,
+      "step": 2759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0802192941602336e-05,
+      "loss": 0.7165,
+      "step": 2760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0801175131769092e-05,
+      "loss": 0.6632,
+      "step": 2761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0800156917261317e-05,
+      "loss": 0.6925,
+      "step": 2762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0799138298125832e-05,
+      "loss": 0.7581,
+      "step": 2763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.079811927440947e-05,
+      "loss": 0.7104,
+      "step": 2764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0797099846159085e-05,
+      "loss": 0.6817,
+      "step": 2765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0796080013421553e-05,
+      "loss": 0.6979,
+      "step": 2766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0795059776243762e-05,
+      "loss": 0.655,
+      "step": 2767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0794039134672626e-05,
+      "loss": 0.7519,
+      "step": 2768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0793018088755068e-05,
+      "loss": 0.7146,
+      "step": 2769
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.079199663853804e-05,
+      "loss": 0.7619,
+      "step": 2770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0790974784068506e-05,
+      "loss": 0.7099,
+      "step": 2771
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.078995252539345e-05,
+      "loss": 0.7539,
+      "step": 2772
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0788929862559877e-05,
+      "loss": 0.7419,
+      "step": 2773
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.078790679561481e-05,
+      "loss": 0.7592,
+      "step": 2774
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0786883324605285e-05,
+      "loss": 0.7699,
+      "step": 2775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.078585944957836e-05,
+      "loss": 0.7117,
+      "step": 2776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.078483517058112e-05,
+      "loss": 0.813,
+      "step": 2777
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.078381048766065e-05,
+      "loss": 0.7165,
+      "step": 2778
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0782785400864074e-05,
+      "loss": 0.7171,
+      "step": 2779
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0781759910238518e-05,
+      "loss": 0.7668,
+      "step": 2780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0780734015831143e-05,
+      "loss": 0.7988,
+      "step": 2781
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0779707717689108e-05,
+      "loss": 0.661,
+      "step": 2782
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.077868101585961e-05,
+      "loss": 0.7751,
+      "step": 2783
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.077765391038985e-05,
+      "loss": 0.732,
+      "step": 2784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.077662640132706e-05,
+      "loss": 0.6966,
+      "step": 2785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0775598488718484e-05,
+      "loss": 0.6994,
+      "step": 2786
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.077457017261138e-05,
+      "loss": 0.7601,
+      "step": 2787
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.077354145305303e-05,
+      "loss": 0.7232,
+      "step": 2788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.077251233009074e-05,
+      "loss": 0.7186,
+      "step": 2789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0771482803771827e-05,
+      "loss": 0.772,
+      "step": 2790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.077045287414362e-05,
+      "loss": 0.8089,
+      "step": 2791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0769422541253486e-05,
+      "loss": 0.6585,
+      "step": 2792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0768391805148796e-05,
+      "loss": 0.6988,
+      "step": 2793
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0767360665876934e-05,
+      "loss": 0.7182,
+      "step": 2794
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0766329123485326e-05,
+      "loss": 0.7299,
+      "step": 2795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.076529717802139e-05,
+      "loss": 0.7514,
+      "step": 2796
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.076426482953258e-05,
+      "loss": 0.6891,
+      "step": 2797
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0763232078066364e-05,
+      "loss": 0.7471,
+      "step": 2798
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0762198923670225e-05,
+      "loss": 0.714,
+      "step": 2799
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0761165366391665e-05,
+      "loss": 0.7821,
+      "step": 2800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.076013140627821e-05,
+      "loss": 0.7316,
+      "step": 2801
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0759097043377402e-05,
+      "loss": 0.7518,
+      "step": 2802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0758062277736797e-05,
+      "loss": 0.6913,
+      "step": 2803
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0757027109403973e-05,
+      "loss": 0.7648,
+      "step": 2804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0755991538426533e-05,
+      "loss": 0.7407,
+      "step": 2805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0754955564852083e-05,
+      "loss": 0.7621,
+      "step": 2806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0753919188728265e-05,
+      "loss": 0.6859,
+      "step": 2807
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0752882410102724e-05,
+      "loss": 0.6809,
+      "step": 2808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0751845229023136e-05,
+      "loss": 0.7547,
+      "step": 2809
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.075080764553719e-05,
+      "loss": 0.7503,
+      "step": 2810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0749769659692588e-05,
+      "loss": 0.7168,
+      "step": 2811
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0748731271537063e-05,
+      "loss": 0.7495,
+      "step": 2812
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0747692481118356e-05,
+      "loss": 0.7491,
+      "step": 2813
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074665328848423e-05,
+      "loss": 0.7261,
+      "step": 2814
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074561369368247e-05,
+      "loss": 0.7925,
+      "step": 2815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074457369676087e-05,
+      "loss": 0.7391,
+      "step": 2816
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0743533297767256e-05,
+      "loss": 0.7629,
+      "step": 2817
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0742492496749456e-05,
+      "loss": 0.7332,
+      "step": 2818
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0741451293755336e-05,
+      "loss": 0.7032,
+      "step": 2819
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0740409688832763e-05,
+      "loss": 0.6693,
+      "step": 2820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0739367682029635e-05,
+      "loss": 0.7577,
+      "step": 2821
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.073832527339386e-05,
+      "loss": 0.6535,
+      "step": 2822
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0737282462973365e-05,
+      "loss": 0.6753,
+      "step": 2823
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0736239250816098e-05,
+      "loss": 0.7524,
+      "step": 2824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.073519563697003e-05,
+      "loss": 0.7841,
+      "step": 2825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0734151621483145e-05,
+      "loss": 0.7878,
+      "step": 2826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0733107204403445e-05,
+      "loss": 0.7126,
+      "step": 2827
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.073206238577895e-05,
+      "loss": 0.8086,
+      "step": 2828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0731017165657708e-05,
+      "loss": 0.6913,
+      "step": 2829
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.072997154408777e-05,
+      "loss": 0.8012,
+      "step": 2830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0728925521117213e-05,
+      "loss": 0.7463,
+      "step": 2831
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0727879096794136e-05,
+      "loss": 0.6263,
+      "step": 2832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.072683227116665e-05,
+      "loss": 0.692,
+      "step": 2833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0725785044282896e-05,
+      "loss": 0.7352,
+      "step": 2834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0724737416191014e-05,
+      "loss": 0.6879,
+      "step": 2835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.072368938693918e-05,
+      "loss": 0.728,
+      "step": 2836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0722640956575585e-05,
+      "loss": 0.7161,
+      "step": 2837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0721592125148427e-05,
+      "loss": 0.72,
+      "step": 2838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0720542892705934e-05,
+      "loss": 0.7101,
+      "step": 2839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071949325929635e-05,
+      "loss": 0.6743,
+      "step": 2840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0718443224967938e-05,
+      "loss": 0.7358,
+      "step": 2841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071739278976898e-05,
+      "loss": 0.7831,
+      "step": 2842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0716341953747768e-05,
+      "loss": 0.7239,
+      "step": 2843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0715290716952624e-05,
+      "loss": 0.7739,
+      "step": 2844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0714239079431882e-05,
+      "loss": 0.7125,
+      "step": 2845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0713187041233897e-05,
+      "loss": 0.7172,
+      "step": 2846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071213460240704e-05,
+      "loss": 0.6931,
+      "step": 2847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.07110817629997e-05,
+      "loss": 0.6855,
+      "step": 2848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071002852306029e-05,
+      "loss": 0.6999,
+      "step": 2849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0708974882637235e-05,
+      "loss": 0.6598,
+      "step": 2850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.070792084177898e-05,
+      "loss": 0.7659,
+      "step": 2851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0706866400533997e-05,
+      "loss": 0.7734,
+      "step": 2852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0705811558950758e-05,
+      "loss": 0.6968,
+      "step": 2853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.070475631707777e-05,
+      "loss": 0.6783,
+      "step": 2854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.070370067496355e-05,
+      "loss": 0.8156,
+      "step": 2855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.070264463265664e-05,
+      "loss": 0.6715,
+      "step": 2856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.070158819020559e-05,
+      "loss": 0.7666,
+      "step": 2857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0700531347658985e-05,
+      "loss": 0.683,
+      "step": 2858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0699474105065407e-05,
+      "loss": 0.7384,
+      "step": 2859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0698416462473476e-05,
+      "loss": 0.7514,
+      "step": 2860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.069735841993182e-05,
+      "loss": 0.7278,
+      "step": 2861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0696299977489083e-05,
+      "loss": 0.7035,
+      "step": 2862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0695241135193936e-05,
+      "loss": 0.7302,
+      "step": 2863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.069418189309506e-05,
+      "loss": 0.6821,
+      "step": 2864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0693122251241163e-05,
+      "loss": 0.7747,
+      "step": 2865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0692062209680967e-05,
+      "loss": 0.6982,
+      "step": 2866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.069100176846321e-05,
+      "loss": 0.797,
+      "step": 2867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0689940927636653e-05,
+      "loss": 0.7399,
+      "step": 2868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.068887968725007e-05,
+      "loss": 0.728,
+      "step": 2869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0687818047352257e-05,
+      "loss": 0.6934,
+      "step": 2870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0686756007992026e-05,
+      "loss": 0.6918,
+      "step": 2871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0685693569218218e-05,
+      "loss": 0.8146,
+      "step": 2872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0684630731079674e-05,
+      "loss": 0.7436,
+      "step": 2873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0683567493625265e-05,
+      "loss": 0.8202,
+      "step": 2874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0682503856903875e-05,
+      "loss": 0.816,
+      "step": 2875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0681439820964422e-05,
+      "loss": 0.8725,
+      "step": 2876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0680375385855818e-05,
+      "loss": 0.8006,
+      "step": 2877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0679310551627013e-05,
+      "loss": 0.698,
+      "step": 2878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.067824531832696e-05,
+      "loss": 0.7296,
+      "step": 2879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.067717968600464e-05,
+      "loss": 0.6857,
+      "step": 2880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0676113654709057e-05,
+      "loss": 0.714,
+      "step": 2881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.067504722448922e-05,
+      "loss": 0.7505,
+      "step": 2882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0673980395394164e-05,
+      "loss": 0.7622,
+      "step": 2883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0672913167472944e-05,
+      "loss": 0.7724,
+      "step": 2884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0671845540774626e-05,
+      "loss": 0.8084,
+      "step": 2885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0670777515348305e-05,
+      "loss": 0.7385,
+      "step": 2886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0669709091243085e-05,
+      "loss": 0.7952,
+      "step": 2887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.066864026850809e-05,
+      "loss": 0.7149,
+      "step": 2888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.066757104719247e-05,
+      "loss": 0.6908,
+      "step": 2889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.066650142734538e-05,
+      "loss": 0.7729,
+      "step": 2890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0665431409016007e-05,
+      "loss": 0.6722,
+      "step": 2891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0664360992253543e-05,
+      "loss": 0.8252,
+      "step": 2892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0663290177107215e-05,
+      "loss": 0.697,
+      "step": 2893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.066221896362625e-05,
+      "loss": 0.7716,
+      "step": 2894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0661147351859905e-05,
+      "loss": 0.6789,
+      "step": 2895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0660075341857453e-05,
+      "loss": 0.6926,
+      "step": 2896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0659002933668184e-05,
+      "loss": 0.7795,
+      "step": 2897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0657930127341408e-05,
+      "loss": 0.7473,
+      "step": 2898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0656856922926454e-05,
+      "loss": 0.7759,
+      "step": 2899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.065578332047266e-05,
+      "loss": 0.7297,
+      "step": 2900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0654709320029395e-05,
+      "loss": 0.8355,
+      "step": 2901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0653634921646045e-05,
+      "loss": 0.6798,
+      "step": 2902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0652560125372e-05,
+      "loss": 0.7683,
+      "step": 2903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0651484931256686e-05,
+      "loss": 0.7071,
+      "step": 2904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.065040933934954e-05,
+      "loss": 0.7307,
+      "step": 2905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0649333349700018e-05,
+      "loss": 0.6519,
+      "step": 2906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0648256962357587e-05,
+      "loss": 0.648,
+      "step": 2907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0647180177371748e-05,
+      "loss": 0.69,
+      "step": 2908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0646102994792008e-05,
+      "loss": 0.6415,
+      "step": 2909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0645025414667888e-05,
+      "loss": 0.7131,
+      "step": 2910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0643947437048946e-05,
+      "loss": 0.7512,
+      "step": 2911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.064286906198474e-05,
+      "loss": 0.7499,
+      "step": 2912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0641790289524852e-05,
+      "loss": 0.7902,
+      "step": 2913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0640711119718892e-05,
+      "loss": 0.741,
+      "step": 2914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.063963155261647e-05,
+      "loss": 0.7227,
+      "step": 2915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0638551588267233e-05,
+      "loss": 0.7286,
+      "step": 2916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0637471226720827e-05,
+      "loss": 0.7497,
+      "step": 2917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0636390468026937e-05,
+      "loss": 0.703,
+      "step": 2918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0635309312235247e-05,
+      "loss": 0.7283,
+      "step": 2919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0634227759395473e-05,
+      "loss": 0.7087,
+      "step": 2920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0633145809557347e-05,
+      "loss": 0.6973,
+      "step": 2921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0632063462770614e-05,
+      "loss": 0.6984,
+      "step": 2922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0630980719085034e-05,
+      "loss": 0.7063,
+      "step": 2923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.06298975785504e-05,
+      "loss": 0.773,
+      "step": 2924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0628814041216508e-05,
+      "loss": 0.781,
+      "step": 2925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0627730107133182e-05,
+      "loss": 0.704,
+      "step": 2926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0626645776350263e-05,
+      "loss": 0.7679,
+      "step": 2927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0625561048917604e-05,
+      "loss": 0.7276,
+      "step": 2928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0624475924885077e-05,
+      "loss": 0.7542,
+      "step": 2929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0623390404302587e-05,
+      "loss": 0.7277,
+      "step": 2930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0622304487220038e-05,
+      "loss": 0.6793,
+      "step": 2931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.062121817368736e-05,
+      "loss": 0.7312,
+      "step": 2932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0620131463754497e-05,
+      "loss": 0.7473,
+      "step": 2933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0619044357471425e-05,
+      "loss": 0.7811,
+      "step": 2934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0617956854888125e-05,
+      "loss": 0.7535,
+      "step": 2935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0616868956054597e-05,
+      "loss": 0.8318,
+      "step": 2936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.061578066102087e-05,
+      "loss": 0.7438,
+      "step": 2937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.061469196983697e-05,
+      "loss": 0.7441,
+      "step": 2938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0613602882552966e-05,
+      "loss": 0.7297,
+      "step": 2939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0612513399218932e-05,
+      "loss": 0.7001,
+      "step": 2940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.061142351988496e-05,
+      "loss": 0.7473,
+      "step": 2941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.061033324460116e-05,
+      "loss": 0.6882,
+      "step": 2942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0609242573417667e-05,
+      "loss": 0.7585,
+      "step": 2943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0608151506384624e-05,
+      "loss": 0.6955,
+      "step": 2944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.060706004355221e-05,
+      "loss": 0.7629,
+      "step": 2945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0605968184970594e-05,
+      "loss": 0.7228,
+      "step": 2946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0604875930689985e-05,
+      "loss": 0.7054,
+      "step": 2947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.060378328076061e-05,
+      "loss": 0.6793,
+      "step": 2948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0602690235232706e-05,
+      "loss": 0.7516,
+      "step": 2949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0601596794156528e-05,
+      "loss": 0.7239,
+      "step": 2950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0600502957582354e-05,
+      "loss": 0.6972,
+      "step": 2951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0599408725560477e-05,
+      "loss": 0.7643,
+      "step": 2952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.059831409814121e-05,
+      "loss": 0.7079,
+      "step": 2953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0597219075374882e-05,
+      "loss": 0.8222,
+      "step": 2954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0596123657311845e-05,
+      "loss": 0.652,
+      "step": 2955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0595027844002462e-05,
+      "loss": 0.6589,
+      "step": 2956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0593931635497122e-05,
+      "loss": 0.649,
+      "step": 2957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0592835031846226e-05,
+      "loss": 0.7225,
+      "step": 2958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0591738033100196e-05,
+      "loss": 0.7957,
+      "step": 2959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0590640639309472e-05,
+      "loss": 0.6692,
+      "step": 2960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.058954285052451e-05,
+      "loss": 0.6866,
+      "step": 2961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0588444666795786e-05,
+      "loss": 0.7602,
+      "step": 2962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.05873460881738e-05,
+      "loss": 0.7558,
+      "step": 2963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0586247114709055e-05,
+      "loss": 0.6768,
+      "step": 2964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0585147746452087e-05,
+      "loss": 0.758,
+      "step": 2965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.058404798345344e-05,
+      "loss": 0.7503,
+      "step": 2966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0582947825763687e-05,
+      "loss": 0.7061,
+      "step": 2967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.058184727343341e-05,
+      "loss": 0.691,
+      "step": 2968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.058074632651321e-05,
+      "loss": 0.7439,
+      "step": 2969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.057964498505371e-05,
+      "loss": 0.6253,
+      "step": 2970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.057854324910555e-05,
+      "loss": 0.7351,
+      "step": 2971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.057744111871938e-05,
+      "loss": 0.7673,
+      "step": 2972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0576338593945888e-05,
+      "loss": 0.7094,
+      "step": 2973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0575235674835762e-05,
+      "loss": 0.7411,
+      "step": 2974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.057413236143971e-05,
+      "loss": 0.7204,
+      "step": 2975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0573028653808467e-05,
+      "loss": 0.7633,
+      "step": 2976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0571924551992782e-05,
+      "loss": 0.6509,
+      "step": 2977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0570820056043414e-05,
+      "loss": 0.7202,
+      "step": 2978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.056971516601115e-05,
+      "loss": 0.7459,
+      "step": 2979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0568609881946798e-05,
+      "loss": 0.7073,
+      "step": 2980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0567504203901176e-05,
+      "loss": 0.7316,
+      "step": 2981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.056639813192512e-05,
+      "loss": 0.7622,
+      "step": 2982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0565291666069488e-05,
+      "loss": 0.729,
+      "step": 2983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.056418480638515e-05,
+      "loss": 0.7375,
+      "step": 2984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0563077552923006e-05,
+      "loss": 0.6856,
+      "step": 2985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0561969905733966e-05,
+      "loss": 0.6444,
+      "step": 2986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.056086186486896e-05,
+      "loss": 0.7137,
+      "step": 2987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.055975343037893e-05,
+      "loss": 0.717,
+      "step": 2988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0558644602314846e-05,
+      "loss": 0.7531,
+      "step": 2989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.055753538072769e-05,
+      "loss": 0.7458,
+      "step": 2990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0556425765668464e-05,
+      "loss": 0.7816,
+      "step": 2991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0555315757188187e-05,
+      "loss": 0.7802,
+      "step": 2992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0554205355337898e-05,
+      "loss": 0.7771,
+      "step": 2993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.055309456016865e-05,
+      "loss": 0.7499,
+      "step": 2994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.055198337173152e-05,
+      "loss": 0.734,
+      "step": 2995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.05508717900776e-05,
+      "loss": 0.6733,
+      "step": 2996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0549759815257996e-05,
+      "loss": 0.729,
+      "step": 2997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0548647447323843e-05,
+      "loss": 0.6859,
+      "step": 2998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0547534686326283e-05,
+      "loss": 0.7108,
+      "step": 2999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0546421532316474e-05,
+      "loss": 0.6929,
+      "step": 3000
+    },
+    {
+      "epoch": 0.2,
+      "eval_text_loss": 0.6714286804199219,
+      "eval_text_runtime": 1.843,
+      "eval_text_samples_per_second": 40.152,
+      "eval_text_steps_per_second": 1.085,
+      "step": 3000
+    },
+    {
+      "epoch": 0.2,
+      "eval_svit_loss": 0.8842295408248901,
+      "eval_svit_runtime": 6.6109,
+      "eval_svit_samples_per_second": 123.432,
+      "eval_svit_steps_per_second": 1.966,
+      "step": 3000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.054530798534561e-05,
+      "loss": 0.7524,
+      "step": 3001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.054419404546489e-05,
+      "loss": 0.6882,
+      "step": 3002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0543079712725525e-05,
+      "loss": 0.674,
+      "step": 3003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0541964987178755e-05,
+      "loss": 0.7448,
+      "step": 3004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0540849868875835e-05,
+      "loss": 0.7582,
+      "step": 3005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.053973435786804e-05,
+      "loss": 0.701,
+      "step": 3006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.053861845420666e-05,
+      "loss": 0.7497,
+      "step": 3007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0537502157942998e-05,
+      "loss": 0.6188,
+      "step": 3008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0536385469128386e-05,
+      "loss": 0.7196,
+      "step": 3009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.053526838781417e-05,
+      "loss": 0.7283,
+      "step": 3010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0534150914051713e-05,
+      "loss": 0.7791,
+      "step": 3011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0533033047892393e-05,
+      "loss": 0.8078,
+      "step": 3012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0531914789387606e-05,
+      "loss": 0.826,
+      "step": 3013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.053079613858878e-05,
+      "loss": 0.8068,
+      "step": 3014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.052967709554734e-05,
+      "loss": 0.7714,
+      "step": 3015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.052855766031474e-05,
+      "loss": 0.7048,
+      "step": 3016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0527437832942456e-05,
+      "loss": 0.8079,
+      "step": 3017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0526317613481976e-05,
+      "loss": 0.7439,
+      "step": 3018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0525197001984805e-05,
+      "loss": 0.7665,
+      "step": 3019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.052407599850247e-05,
+      "loss": 0.6895,
+      "step": 3020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0522954603086514e-05,
+      "loss": 0.7198,
+      "step": 3021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.05218328157885e-05,
+      "loss": 0.7444,
+      "step": 3022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.052071063666e-05,
+      "loss": 0.7222,
+      "step": 3023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0519588065752623e-05,
+      "loss": 0.6775,
+      "step": 3024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.051846510311797e-05,
+      "loss": 0.617,
+      "step": 3025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0517341748807693e-05,
+      "loss": 0.7289,
+      "step": 3026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0516218002873425e-05,
+      "loss": 0.686,
+      "step": 3027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0515093865366847e-05,
+      "loss": 0.6977,
+      "step": 3028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0513969336339642e-05,
+      "loss": 0.7497,
+      "step": 3029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0512844415843515e-05,
+      "loss": 0.7427,
+      "step": 3030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.051171910393019e-05,
+      "loss": 0.8007,
+      "step": 3031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.051059340065141e-05,
+      "loss": 0.7667,
+      "step": 3032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0509467306058934e-05,
+      "loss": 0.748,
+      "step": 3033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0508340820204538e-05,
+      "loss": 0.6898,
+      "step": 3034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0507213943140017e-05,
+      "loss": 0.8026,
+      "step": 3035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.050608667491719e-05,
+      "loss": 0.7435,
+      "step": 3036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.050495901558788e-05,
+      "loss": 0.8353,
+      "step": 3037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0503830965203937e-05,
+      "loss": 0.751,
+      "step": 3038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0502702523817234e-05,
+      "loss": 0.6392,
+      "step": 3039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0501573691479656e-05,
+      "loss": 0.7429,
+      "step": 3040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.05004444682431e-05,
+      "loss": 0.6813,
+      "step": 3041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0499314854159493e-05,
+      "loss": 0.7624,
+      "step": 3042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0498184849280772e-05,
+      "loss": 0.7058,
+      "step": 3043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0497054453658895e-05,
+      "loss": 0.7287,
+      "step": 3044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.049592366734584e-05,
+      "loss": 0.6715,
+      "step": 3045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.049479249039359e-05,
+      "loss": 0.751,
+      "step": 3046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0493660922854167e-05,
+      "loss": 0.7776,
+      "step": 3047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0492528964779594e-05,
+      "loss": 0.7287,
+      "step": 3048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.049139661622192e-05,
+      "loss": 0.7236,
+      "step": 3049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0490263877233214e-05,
+      "loss": 0.6637,
+      "step": 3050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.048913074786555e-05,
+      "loss": 0.772,
+      "step": 3051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0487997228171034e-05,
+      "loss": 0.7489,
+      "step": 3052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0486863318201785e-05,
+      "loss": 0.7389,
+      "step": 3053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0485729018009942e-05,
+      "loss": 0.673,
+      "step": 3054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.048459432764765e-05,
+      "loss": 0.7559,
+      "step": 3055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0483459247167093e-05,
+      "loss": 0.767,
+      "step": 3056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0482323776620455e-05,
+      "loss": 0.7148,
+      "step": 3057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0481187916059947e-05,
+      "loss": 0.7634,
+      "step": 3058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0480051665537795e-05,
+      "loss": 0.813,
+      "step": 3059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0478915025106242e-05,
+      "loss": 0.7544,
+      "step": 3060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0477777994817556e-05,
+      "loss": 0.7003,
+      "step": 3061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0476640574724007e-05,
+      "loss": 0.7638,
+      "step": 3062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.04755027648779e-05,
+      "loss": 0.7269,
+      "step": 3063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.047436456533155e-05,
+      "loss": 0.7417,
+      "step": 3064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.047322597613729e-05,
+      "loss": 0.7133,
+      "step": 3065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0472086997347473e-05,
+      "loss": 0.7389,
+      "step": 3066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0470947629014472e-05,
+      "loss": 0.7685,
+      "step": 3067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0469807871190667e-05,
+      "loss": 0.7547,
+      "step": 3068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0468667723928468e-05,
+      "loss": 0.7957,
+      "step": 3069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.04675271872803e-05,
+      "loss": 0.816,
+      "step": 3070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.04663862612986e-05,
+      "loss": 0.7519,
+      "step": 3071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0465244946035834e-05,
+      "loss": 0.7836,
+      "step": 3072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0464103241544473e-05,
+      "loss": 0.7728,
+      "step": 3073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.046296114787701e-05,
+      "loss": 0.7889,
+      "step": 3074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0461818665085968e-05,
+      "loss": 0.7373,
+      "step": 3075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.046067579322387e-05,
+      "loss": 0.7181,
+      "step": 3076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0459532532343267e-05,
+      "loss": 0.692,
+      "step": 3077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0458388882496727e-05,
+      "loss": 0.7641,
+      "step": 3078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0457244843736832e-05,
+      "loss": 0.8598,
+      "step": 3079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0456100416116186e-05,
+      "loss": 0.7279,
+      "step": 3080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0454955599687407e-05,
+      "loss": 0.7109,
+      "step": 3081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0453810394503137e-05,
+      "loss": 0.8264,
+      "step": 3082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0452664800616033e-05,
+      "loss": 0.748,
+      "step": 3083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0451518818078764e-05,
+      "loss": 0.6903,
+      "step": 3084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045037244694402e-05,
+      "loss": 0.7636,
+      "step": 3085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0449225687264518e-05,
+      "loss": 0.7652,
+      "step": 3086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0448078539092986e-05,
+      "loss": 0.8107,
+      "step": 3087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0446931002482163e-05,
+      "loss": 0.8147,
+      "step": 3088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.044578307748481e-05,
+      "loss": 0.7585,
+      "step": 3089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0444634764153718e-05,
+      "loss": 0.6961,
+      "step": 3090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.044348606254168e-05,
+      "loss": 0.7898,
+      "step": 3091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0442336972701518e-05,
+      "loss": 0.7217,
+      "step": 3092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0441187494686057e-05,
+      "loss": 0.644,
+      "step": 3093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0440037628548156e-05,
+      "loss": 0.7009,
+      "step": 3094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0438887374340685e-05,
+      "loss": 0.7444,
+      "step": 3095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0437736732116534e-05,
+      "loss": 0.6604,
+      "step": 3096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0436585701928606e-05,
+      "loss": 0.7501,
+      "step": 3097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0435434283829825e-05,
+      "loss": 0.7234,
+      "step": 3098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0434282477873135e-05,
+      "loss": 0.7352,
+      "step": 3099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.043313028411149e-05,
+      "loss": 0.7643,
+      "step": 3100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.043197770259788e-05,
+      "loss": 0.6295,
+      "step": 3101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0430824733385283e-05,
+      "loss": 0.7493,
+      "step": 3102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0429671376526724e-05,
+      "loss": 0.6714,
+      "step": 3103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.042851763207523e-05,
+      "loss": 0.7683,
+      "step": 3104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.042736350008385e-05,
+      "loss": 0.7274,
+      "step": 3105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0426208980605655e-05,
+      "loss": 0.8048,
+      "step": 3106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0425054073693723e-05,
+      "loss": 0.7014,
+      "step": 3107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0423898779401155e-05,
+      "loss": 0.7471,
+      "step": 3108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0422743097781078e-05,
+      "loss": 0.7478,
+      "step": 3109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0421587028886628e-05,
+      "loss": 0.7134,
+      "step": 3110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0420430572770954e-05,
+      "loss": 0.6477,
+      "step": 3111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.041927372948724e-05,
+      "loss": 0.6823,
+      "step": 3112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0418116499088666e-05,
+      "loss": 0.7431,
+      "step": 3113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0416958881628446e-05,
+      "loss": 0.748,
+      "step": 3114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.041580087715981e-05,
+      "loss": 0.6487,
+      "step": 3115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0414642485735998e-05,
+      "loss": 0.7067,
+      "step": 3116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.041348370741028e-05,
+      "loss": 0.7796,
+      "step": 3117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0412324542235924e-05,
+      "loss": 0.6851,
+      "step": 3118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0411164990266236e-05,
+      "loss": 0.7938,
+      "step": 3119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.041000505155453e-05,
+      "loss": 0.7786,
+      "step": 3120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.040884472615414e-05,
+      "loss": 0.7497,
+      "step": 3121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0407684014118417e-05,
+      "loss": 0.7412,
+      "step": 3122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0406522915500726e-05,
+      "loss": 0.7183,
+      "step": 3123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.040536143035446e-05,
+      "loss": 0.7385,
+      "step": 3124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0404199558733023e-05,
+      "loss": 0.6723,
+      "step": 3125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0403037300689838e-05,
+      "loss": 0.7581,
+      "step": 3126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0401874656278337e-05,
+      "loss": 0.7118,
+      "step": 3127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0400711625551988e-05,
+      "loss": 0.7739,
+      "step": 3128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0399548208564263e-05,
+      "loss": 0.686,
+      "step": 3129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0398384405368654e-05,
+      "loss": 0.7568,
+      "step": 3130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039722021601867e-05,
+      "loss": 0.7046,
+      "step": 3131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039605564056785e-05,
+      "loss": 0.7348,
+      "step": 3132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039489067906973e-05,
+      "loss": 0.723,
+      "step": 3133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0393725331577877e-05,
+      "loss": 0.6894,
+      "step": 3134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039255959814587e-05,
+      "loss": 0.7075,
+      "step": 3135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0391393478827322e-05,
+      "loss": 0.7609,
+      "step": 3136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039022697367584e-05,
+      "loss": 0.6802,
+      "step": 3137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0389060082745058e-05,
+      "loss": 0.6452,
+      "step": 3138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0387892806088638e-05,
+      "loss": 0.7291,
+      "step": 3139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0386725143760242e-05,
+      "loss": 0.698,
+      "step": 3140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0385557095813564e-05,
+      "loss": 0.7451,
+      "step": 3141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.038438866230231e-05,
+      "loss": 0.7478,
+      "step": 3142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0383219843280202e-05,
+      "loss": 0.7106,
+      "step": 3143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.038205063880098e-05,
+      "loss": 0.7995,
+      "step": 3144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0380881048918407e-05,
+      "loss": 0.7706,
+      "step": 3145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0379711073686263e-05,
+      "loss": 0.6962,
+      "step": 3146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0378540713158337e-05,
+      "loss": 0.7473,
+      "step": 3147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0377369967388446e-05,
+      "loss": 0.7081,
+      "step": 3148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0376198836430417e-05,
+      "loss": 0.742,
+      "step": 3149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.03750273203381e-05,
+      "loss": 0.7001,
+      "step": 3150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0373855419165362e-05,
+      "loss": 0.755,
+      "step": 3151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0372683132966083e-05,
+      "loss": 0.717,
+      "step": 3152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.037151046179417e-05,
+      "loss": 0.7748,
+      "step": 3153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0370337405703535e-05,
+      "loss": 0.7334,
+      "step": 3154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0369163964748118e-05,
+      "loss": 0.728,
+      "step": 3155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0367990138981876e-05,
+      "loss": 0.6993,
+      "step": 3156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0366815928458777e-05,
+      "loss": 0.6913,
+      "step": 3157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0365641333232813e-05,
+      "loss": 0.7523,
+      "step": 3158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.036446635335799e-05,
+      "loss": 0.7053,
+      "step": 3159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0363290988888335e-05,
+      "loss": 0.7377,
+      "step": 3160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0362115239877885e-05,
+      "loss": 0.715,
+      "step": 3161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.036093910638071e-05,
+      "loss": 0.6991,
+      "step": 3162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0359762588450882e-05,
+      "loss": 0.7663,
+      "step": 3163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0358585686142496e-05,
+      "loss": 0.8193,
+      "step": 3164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.035740839950967e-05,
+      "loss": 0.7826,
+      "step": 3165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0356230728606532e-05,
+      "loss": 0.7171,
+      "step": 3166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.035505267348723e-05,
+      "loss": 0.7162,
+      "step": 3167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0353874234205935e-05,
+      "loss": 0.7901,
+      "step": 3168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0352695410816823e-05,
+      "loss": 0.8028,
+      "step": 3169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0351516203374105e-05,
+      "loss": 0.7319,
+      "step": 3170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0350336611931996e-05,
+      "loss": 0.7076,
+      "step": 3171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0349156636544733e-05,
+      "loss": 0.6925,
+      "step": 3172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0347976277266572e-05,
+      "loss": 0.6645,
+      "step": 3173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0346795534151783e-05,
+      "loss": 0.7454,
+      "step": 3174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.034561440725466e-05,
+      "loss": 0.7856,
+      "step": 3175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0344432896629507e-05,
+      "loss": 0.7857,
+      "step": 3176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0343251002330653e-05,
+      "loss": 0.6447,
+      "step": 3177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0342068724412438e-05,
+      "loss": 0.7163,
+      "step": 3178
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0340886062929224e-05,
+      "loss": 0.752,
+      "step": 3179
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0339703017935387e-05,
+      "loss": 0.6826,
+      "step": 3180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033851958948533e-05,
+      "loss": 0.7242,
+      "step": 3181
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033733577763346e-05,
+      "loss": 0.7995,
+      "step": 3182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0336151582434206e-05,
+      "loss": 0.6697,
+      "step": 3183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033496700394202e-05,
+      "loss": 0.7225,
+      "step": 3184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0333782042211378e-05,
+      "loss": 0.7128,
+      "step": 3185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0332596697296745e-05,
+      "loss": 0.7664,
+      "step": 3186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033141096925264e-05,
+      "loss": 0.7146,
+      "step": 3187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033022485813357e-05,
+      "loss": 0.7597,
+      "step": 3188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032903836399408e-05,
+      "loss": 0.7404,
+      "step": 3189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032785148688872e-05,
+      "loss": 0.7261,
+      "step": 3190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0326664226872064e-05,
+      "loss": 0.7704,
+      "step": 3191
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0325476583998705e-05,
+      "loss": 0.706,
+      "step": 3192
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0324288558323243e-05,
+      "loss": 0.6961,
+      "step": 3193
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032310014990031e-05,
+      "loss": 0.6956,
+      "step": 3194
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032191135878454e-05,
+      "loss": 0.724,
+      "step": 3195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0320722185030604e-05,
+      "loss": 0.6697,
+      "step": 3196
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.031953262869317e-05,
+      "loss": 0.6971,
+      "step": 3197
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.031834268982694e-05,
+      "loss": 0.7241,
+      "step": 3198
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.031715236848662e-05,
+      "loss": 0.7214,
+      "step": 3199
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0315961664726954e-05,
+      "loss": 0.7114,
+      "step": 3200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0314770578602672e-05,
+      "loss": 0.7307,
+      "step": 3201
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.031357911016855e-05,
+      "loss": 0.6265,
+      "step": 3202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0312387259479375e-05,
+      "loss": 0.7222,
+      "step": 3203
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.031119502658994e-05,
+      "loss": 0.7212,
+      "step": 3204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0310002411555063e-05,
+      "loss": 0.6936,
+      "step": 3205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0308809414429588e-05,
+      "loss": 0.78,
+      "step": 3206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.030761603526836e-05,
+      "loss": 0.6079,
+      "step": 3207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0306422274126255e-05,
+      "loss": 0.757,
+      "step": 3208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0305228131058162e-05,
+      "loss": 0.637,
+      "step": 3209
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0304033606118984e-05,
+      "loss": 0.7161,
+      "step": 3210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0302838699363648e-05,
+      "loss": 0.7835,
+      "step": 3211
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.030164341084709e-05,
+      "loss": 0.7036,
+      "step": 3212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.030044774062428e-05,
+      "loss": 0.71,
+      "step": 3213
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0299251688750177e-05,
+      "loss": 0.7178,
+      "step": 3214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.029805525527979e-05,
+      "loss": 0.7456,
+      "step": 3215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0296858440268126e-05,
+      "loss": 0.7243,
+      "step": 3216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0295661243770217e-05,
+      "loss": 0.765,
+      "step": 3217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.02944636658411e-05,
+      "loss": 0.7528,
+      "step": 3218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0293265706535845e-05,
+      "loss": 0.6863,
+      "step": 3219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.029206736590954e-05,
+      "loss": 0.7754,
+      "step": 3220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0290868644017275e-05,
+      "loss": 0.7501,
+      "step": 3221
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.028966954091417e-05,
+      "loss": 0.7167,
+      "step": 3222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0288470056655363e-05,
+      "loss": 0.7607,
+      "step": 3223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0287270191295997e-05,
+      "loss": 0.7689,
+      "step": 3224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0286069944891247e-05,
+      "loss": 0.7701,
+      "step": 3225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0284869317496305e-05,
+      "loss": 0.7365,
+      "step": 3226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0283668309166364e-05,
+      "loss": 0.7558,
+      "step": 3227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0282466919956654e-05,
+      "loss": 0.7084,
+      "step": 3228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0281265149922412e-05,
+      "loss": 0.749,
+      "step": 3229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0280062999118894e-05,
+      "loss": 0.7654,
+      "step": 3230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0278860467601375e-05,
+      "loss": 0.7303,
+      "step": 3231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0277657555425148e-05,
+      "loss": 0.6881,
+      "step": 3232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0276454262645522e-05,
+      "loss": 0.7526,
+      "step": 3233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0275250589317825e-05,
+      "loss": 0.7525,
+      "step": 3234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0274046535497396e-05,
+      "loss": 0.7086,
+      "step": 3235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0272842101239606e-05,
+      "loss": 0.7239,
+      "step": 3236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0271637286599824e-05,
+      "loss": 0.684,
+      "step": 3237
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0270432091633457e-05,
+      "loss": 0.7733,
+      "step": 3238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0269226516395913e-05,
+      "loss": 0.6241,
+      "step": 3239
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0268020560942624e-05,
+      "loss": 0.681,
+      "step": 3240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0266814225329042e-05,
+      "loss": 0.6973,
+      "step": 3241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026560750961063e-05,
+      "loss": 0.7796,
+      "step": 3242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026440041384288e-05,
+      "loss": 0.8243,
+      "step": 3243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026319293808129e-05,
+      "loss": 0.6507,
+      "step": 3244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0261985082381373e-05,
+      "loss": 0.7695,
+      "step": 3245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026077684679867e-05,
+      "loss": 0.8496,
+      "step": 3246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0259568231388738e-05,
+      "loss": 0.7054,
+      "step": 3247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.025835923620715e-05,
+      "loss": 0.769,
+      "step": 3248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0257149861309488e-05,
+      "loss": 0.7518,
+      "step": 3249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0255940106751362e-05,
+      "loss": 0.704,
+      "step": 3250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0254729972588397e-05,
+      "loss": 0.7552,
+      "step": 3251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0253519458876237e-05,
+      "loss": 0.7654,
+      "step": 3252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.025230856567053e-05,
+      "loss": 0.756,
+      "step": 3253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0251097293026967e-05,
+      "loss": 0.7988,
+      "step": 3254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0249885641001232e-05,
+      "loss": 0.6931,
+      "step": 3255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0248673609649044e-05,
+      "loss": 0.6611,
+      "step": 3256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.024746119902612e-05,
+      "loss": 0.7839,
+      "step": 3257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0246248409188217e-05,
+      "loss": 0.7967,
+      "step": 3258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0245035240191096e-05,
+      "loss": 0.759,
+      "step": 3259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0243821692090535e-05,
+      "loss": 0.7019,
+      "step": 3260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0242607764942333e-05,
+      "loss": 0.6832,
+      "step": 3261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0241393458802308e-05,
+      "loss": 0.7171,
+      "step": 3262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0240178773726294e-05,
+      "loss": 0.7265,
+      "step": 3263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.023896370977014e-05,
+      "loss": 0.6995,
+      "step": 3264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.023774826698971e-05,
+      "loss": 0.7525,
+      "step": 3265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0236532445440898e-05,
+      "loss": 0.7067,
+      "step": 3266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0235316245179602e-05,
+      "loss": 0.7593,
+      "step": 3267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0234099666261744e-05,
+      "loss": 0.6983,
+      "step": 3268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.023288270874326e-05,
+      "loss": 0.7715,
+      "step": 3269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.023166537268011e-05,
+      "loss": 0.765,
+      "step": 3270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0230447658128256e-05,
+      "loss": 0.7472,
+      "step": 3271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0229229565143698e-05,
+      "loss": 0.7718,
+      "step": 3272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.022801109378244e-05,
+      "loss": 0.7115,
+      "step": 3273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0226792244100512e-05,
+      "loss": 0.7537,
+      "step": 3274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0225573016153946e-05,
+      "loss": 0.7384,
+      "step": 3275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.022435340999881e-05,
+      "loss": 0.6719,
+      "step": 3276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0223133425691176e-05,
+      "loss": 0.7387,
+      "step": 3277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0221913063287143e-05,
+      "loss": 0.753,
+      "step": 3278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.022069232284282e-05,
+      "loss": 0.7417,
+      "step": 3279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.021947120441433e-05,
+      "loss": 0.7503,
+      "step": 3280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0218249708057836e-05,
+      "loss": 0.7902,
+      "step": 3281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0217027833829486e-05,
+      "loss": 0.8301,
+      "step": 3282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.021580558178547e-05,
+      "loss": 0.7255,
+      "step": 3283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0214582951981978e-05,
+      "loss": 0.669,
+      "step": 3284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0213359944475238e-05,
+      "loss": 0.6916,
+      "step": 3285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0212136559321474e-05,
+      "loss": 0.7125,
+      "step": 3286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0210912796576942e-05,
+      "loss": 0.578,
+      "step": 3287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0209688656297903e-05,
+      "loss": 0.6667,
+      "step": 3288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0208464138540654e-05,
+      "loss": 0.8224,
+      "step": 3289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0207239243361492e-05,
+      "loss": 0.6963,
+      "step": 3290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0206013970816734e-05,
+      "loss": 0.7413,
+      "step": 3291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.020478832096272e-05,
+      "loss": 0.755,
+      "step": 3292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0203562293855804e-05,
+      "loss": 0.7326,
+      "step": 3293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.020233588955236e-05,
+      "loss": 0.6661,
+      "step": 3294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0201109108108777e-05,
+      "loss": 0.7482,
+      "step": 3295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0199881949581462e-05,
+      "loss": 0.7628,
+      "step": 3296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.019865441402684e-05,
+      "loss": 0.7101,
+      "step": 3297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0197426501501352e-05,
+      "loss": 0.7629,
+      "step": 3298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0196198212061454e-05,
+      "loss": 0.7211,
+      "step": 3299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0194969545763627e-05,
+      "loss": 0.7209,
+      "step": 3300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0193740502664363e-05,
+      "loss": 0.6704,
+      "step": 3301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0192511082820175e-05,
+      "loss": 0.6451,
+      "step": 3302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.019128128628758e-05,
+      "loss": 0.7469,
+      "step": 3303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0190051113123142e-05,
+      "loss": 0.7297,
+      "step": 3304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0188820563383406e-05,
+      "loss": 0.8334,
+      "step": 3305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0187589637124965e-05,
+      "loss": 0.6937,
+      "step": 3306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.018635833440441e-05,
+      "loss": 0.6252,
+      "step": 3307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0185126655278358e-05,
+      "loss": 0.6953,
+      "step": 3308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.018389459980344e-05,
+      "loss": 0.6986,
+      "step": 3309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0182662168036305e-05,
+      "loss": 0.7508,
+      "step": 3310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0181429360033623e-05,
+      "loss": 0.7431,
+      "step": 3311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0180196175852076e-05,
+      "loss": 0.6851,
+      "step": 3312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0178962615548364e-05,
+      "loss": 0.7292,
+      "step": 3313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0177728679179208e-05,
+      "loss": 0.799,
+      "step": 3314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.017649436680134e-05,
+      "loss": 0.8305,
+      "step": 3315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0175259678471518e-05,
+      "loss": 0.7676,
+      "step": 3316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.017402461424651e-05,
+      "loss": 0.7315,
+      "step": 3317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0172789174183104e-05,
+      "loss": 0.6847,
+      "step": 3318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0171553358338105e-05,
+      "loss": 0.7334,
+      "step": 3319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0170317166768335e-05,
+      "loss": 0.6786,
+      "step": 3320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0169080599530633e-05,
+      "loss": 0.8057,
+      "step": 3321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.016784365668186e-05,
+      "loss": 0.7628,
+      "step": 3322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0166606338278885e-05,
+      "loss": 0.7542,
+      "step": 3323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0165368644378605e-05,
+      "loss": 0.6944,
+      "step": 3324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.016413057503792e-05,
+      "loss": 0.6762,
+      "step": 3325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.016289213031376e-05,
+      "loss": 0.7856,
+      "step": 3326
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0161653310263075e-05,
+      "loss": 0.78,
+      "step": 3327
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0160414114942817e-05,
+      "loss": 0.7203,
+      "step": 3328
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0159174544409967e-05,
+      "loss": 0.7056,
+      "step": 3329
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.015793459872152e-05,
+      "loss": 0.7492,
+      "step": 3330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0156694277934485e-05,
+      "loss": 0.7517,
+      "step": 3331
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0155453582105895e-05,
+      "loss": 0.7283,
+      "step": 3332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0154212511292798e-05,
+      "loss": 0.6498,
+      "step": 3333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.015297106555225e-05,
+      "loss": 0.7634,
+      "step": 3334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0151729244941344e-05,
+      "loss": 0.7894,
+      "step": 3335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0150487049517174e-05,
+      "loss": 0.7168,
+      "step": 3336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0149244479336847e-05,
+      "loss": 0.6771,
+      "step": 3337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0148001534457505e-05,
+      "loss": 0.7021,
+      "step": 3338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0146758214936298e-05,
+      "loss": 0.7433,
+      "step": 3339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0145514520830392e-05,
+      "loss": 0.6577,
+      "step": 3340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0144270452196966e-05,
+      "loss": 0.7299,
+      "step": 3341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0143026009093232e-05,
+      "loss": 0.671,
+      "step": 3342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.01417811915764e-05,
+      "loss": 0.8116,
+      "step": 3343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0140535999703708e-05,
+      "loss": 0.7761,
+      "step": 3344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0139290433532416e-05,
+      "loss": 0.7082,
+      "step": 3345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0138044493119786e-05,
+      "loss": 0.7296,
+      "step": 3346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0136798178523115e-05,
+      "loss": 0.7736,
+      "step": 3347
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.01355514897997e-05,
+      "loss": 0.7314,
+      "step": 3348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0134304427006865e-05,
+      "loss": 0.6798,
+      "step": 3349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.013305699020195e-05,
+      "loss": 0.7122,
+      "step": 3350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0131809179442314e-05,
+      "loss": 0.7517,
+      "step": 3351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0130560994785326e-05,
+      "loss": 0.77,
+      "step": 3352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0129312436288385e-05,
+      "loss": 0.726,
+      "step": 3353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0128063504008893e-05,
+      "loss": 0.7454,
+      "step": 3354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0126814198004272e-05,
+      "loss": 0.7561,
+      "step": 3355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.012556451833197e-05,
+      "loss": 0.6655,
+      "step": 3356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0124314465049448e-05,
+      "loss": 0.7132,
+      "step": 3357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0123064038214185e-05,
+      "loss": 0.7385,
+      "step": 3358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0121813237883667e-05,
+      "loss": 0.7209,
+      "step": 3359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0120562064115408e-05,
+      "loss": 0.7196,
+      "step": 3360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0119310516966942e-05,
+      "loss": 0.6967,
+      "step": 3361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.011805859649581e-05,
+      "loss": 0.7178,
+      "step": 3362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.011680630275957e-05,
+      "loss": 0.7479,
+      "step": 3363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0115553635815808e-05,
+      "loss": 0.7834,
+      "step": 3364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0114300595722126e-05,
+      "loss": 0.7473,
+      "step": 3365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.011304718253613e-05,
+      "loss": 0.6687,
+      "step": 3366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.011179339631545e-05,
+      "loss": 0.7118,
+      "step": 3367
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0110539237117746e-05,
+      "loss": 0.7855,
+      "step": 3368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0109284705000674e-05,
+      "loss": 0.6843,
+      "step": 3369
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0108029800021918e-05,
+      "loss": 0.6467,
+      "step": 3370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0106774522239178e-05,
+      "loss": 0.7847,
+      "step": 3371
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0105518871710177e-05,
+      "loss": 0.7674,
+      "step": 3372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0104262848492642e-05,
+      "loss": 0.7988,
+      "step": 3373
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0103006452644328e-05,
+      "loss": 0.7062,
+      "step": 3374
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0101749684222998e-05,
+      "loss": 0.7313,
+      "step": 3375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0100492543286448e-05,
+      "loss": 0.6968,
+      "step": 3376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0099235029892475e-05,
+      "loss": 0.7059,
+      "step": 3377
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.00979771440989e-05,
+      "loss": 0.7813,
+      "step": 3378
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0096718885963557e-05,
+      "loss": 0.681,
+      "step": 3379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0095460255544302e-05,
+      "loss": 0.8107,
+      "step": 3380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.009420125289901e-05,
+      "loss": 0.7142,
+      "step": 3381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0092941878085566e-05,
+      "loss": 0.7273,
+      "step": 3382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0091682131161873e-05,
+      "loss": 0.7004,
+      "step": 3383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.009042201218586e-05,
+      "loss": 0.7257,
+      "step": 3384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0089161521215464e-05,
+      "loss": 0.7602,
+      "step": 3385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0087900658308638e-05,
+      "loss": 0.7285,
+      "step": 3386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.008663942352336e-05,
+      "loss": 0.7602,
+      "step": 3387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0085377816917622e-05,
+      "loss": 0.7021,
+      "step": 3388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0084115838549432e-05,
+      "loss": 0.6903,
+      "step": 3389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0082853488476808e-05,
+      "loss": 0.7547,
+      "step": 3390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0081590766757806e-05,
+      "loss": 0.7563,
+      "step": 3391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.008032767345047e-05,
+      "loss": 0.767,
+      "step": 3392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.007906420861289e-05,
+      "loss": 0.7633,
+      "step": 3393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.007780037230315e-05,
+      "loss": 0.781,
+      "step": 3394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0076536164579367e-05,
+      "loss": 0.7332,
+      "step": 3395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0075271585499664e-05,
+      "loss": 0.7495,
+      "step": 3396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0074006635122188e-05,
+      "loss": 0.7478,
+      "step": 3397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0072741313505102e-05,
+      "loss": 0.7346,
+      "step": 3398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0071475620706585e-05,
+      "loss": 0.6852,
+      "step": 3399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.007020955678483e-05,
+      "loss": 0.6861,
+      "step": 3400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0068943121798054e-05,
+      "loss": 0.7281,
+      "step": 3401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0067676315804483e-05,
+      "loss": 0.7295,
+      "step": 3402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.006640913886237e-05,
+      "loss": 0.6834,
+      "step": 3403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0065141591029972e-05,
+      "loss": 0.8341,
+      "step": 3404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0063873672365577e-05,
+      "loss": 0.7706,
+      "step": 3405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.006260538292748e-05,
+      "loss": 0.6663,
+      "step": 3406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0061336722773992e-05,
+      "loss": 0.7138,
+      "step": 3407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.006006769196346e-05,
+      "loss": 0.718,
+      "step": 3408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0058798290554215e-05,
+      "loss": 0.7734,
+      "step": 3409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.005752851860464e-05,
+      "loss": 0.7496,
+      "step": 3410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0056258376173103e-05,
+      "loss": 0.6927,
+      "step": 3411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0054987863318018e-05,
+      "loss": 0.7058,
+      "step": 3412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0053716980097796e-05,
+      "loss": 0.7707,
+      "step": 3413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0052445726570874e-05,
+      "loss": 0.6779,
+      "step": 3414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0051174102795706e-05,
+      "loss": 0.7559,
+      "step": 3415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0049902108830755e-05,
+      "loss": 0.7395,
+      "step": 3416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0048629744734507e-05,
+      "loss": 0.6309,
+      "step": 3417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.004735701056547e-05,
+      "loss": 0.6351,
+      "step": 3418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0046083906382162e-05,
+      "loss": 0.761,
+      "step": 3419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0044810432243114e-05,
+      "loss": 0.7882,
+      "step": 3420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.004353658820689e-05,
+      "loss": 0.6936,
+      "step": 3421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.004226237433205e-05,
+      "loss": 0.7331,
+      "step": 3422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0040987790677194e-05,
+      "loss": 0.7201,
+      "step": 3423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0039712837300918e-05,
+      "loss": 0.6756,
+      "step": 3424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0038437514261846e-05,
+      "loss": 0.6863,
+      "step": 3425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0037161821618617e-05,
+      "loss": 0.7836,
+      "step": 3426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0035885759429884e-05,
+      "loss": 0.7033,
+      "step": 3427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0034609327754324e-05,
+      "loss": 0.6987,
+      "step": 3428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0033332526650627e-05,
+      "loss": 0.7374,
+      "step": 3429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0032055356177498e-05,
+      "loss": 0.6949,
+      "step": 3430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.003077781639366e-05,
+      "loss": 0.673,
+      "step": 3431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0029499907357857e-05,
+      "loss": 0.6844,
+      "step": 3432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0028221629128843e-05,
+      "loss": 0.7331,
+      "step": 3433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0026942981765394e-05,
+      "loss": 0.6391,
+      "step": 3434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.00256639653263e-05,
+      "loss": 0.6348,
+      "step": 3435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0024384579870377e-05,
+      "loss": 0.744,
+      "step": 3436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.002310482545644e-05,
+      "loss": 0.7139,
+      "step": 3437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.002182470214334e-05,
+      "loss": 0.7641,
+      "step": 3438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0020544209989933e-05,
+      "loss": 0.6646,
+      "step": 3439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0019263349055097e-05,
+      "loss": 0.6899,
+      "step": 3440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.001798211939772e-05,
+      "loss": 0.7339,
+      "step": 3441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.001670052107672e-05,
+      "loss": 0.7791,
+      "step": 3442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0015418554151025e-05,
+      "loss": 0.7824,
+      "step": 3443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0014136218679567e-05,
+      "loss": 0.6889,
+      "step": 3444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0012853514721323e-05,
+      "loss": 0.713,
+      "step": 3445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0011570442335263e-05,
+      "loss": 0.6971,
+      "step": 3446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0010287001580382e-05,
+      "loss": 0.7589,
+      "step": 3447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.000900319251569e-05,
+      "loss": 0.7767,
+      "step": 3448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0007719015200227e-05,
+      "loss": 0.7746,
+      "step": 3449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0006434469693025e-05,
+      "loss": 0.7161,
+      "step": 3450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0005149556053152e-05,
+      "loss": 0.7015,
+      "step": 3451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0003864274339696e-05,
+      "loss": 0.7254,
+      "step": 3452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.000257862461174e-05,
+      "loss": 0.7054,
+      "step": 3453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.000129260692841e-05,
+      "loss": 0.8143,
+      "step": 3454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.000000622134883e-05,
+      "loss": 0.6907,
+      "step": 3455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9998719467932145e-05,
+      "loss": 0.7503,
+      "step": 3456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9997432346737525e-05,
+      "loss": 0.6745,
+      "step": 3457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9996144857824147e-05,
+      "loss": 0.7174,
+      "step": 3458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9994857001251216e-05,
+      "loss": 0.7192,
+      "step": 3459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9993568777077936e-05,
+      "loss": 0.7748,
+      "step": 3460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9992280185363552e-05,
+      "loss": 0.7724,
+      "step": 3461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9990991226167305e-05,
+      "loss": 0.7302,
+      "step": 3462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9989701899548464e-05,
+      "loss": 0.69,
+      "step": 3463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988412205566303e-05,
+      "loss": 0.7009,
+      "step": 3464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987122144280137e-05,
+      "loss": 0.7617,
+      "step": 3465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998583171574927e-05,
+      "loss": 0.7482,
+      "step": 3466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9984540920033043e-05,
+      "loss": 0.6687,
+      "step": 3467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99832497571908e-05,
+      "loss": 0.7841,
+      "step": 3468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998195822728191e-05,
+      "loss": 0.7089,
+      "step": 3469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998066633036576e-05,
+      "loss": 0.7382,
+      "step": 3470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.997937406650175e-05,
+      "loss": 0.7198,
+      "step": 3471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9978081435749295e-05,
+      "loss": 0.7682,
+      "step": 3472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.997678843816783e-05,
+      "loss": 0.7258,
+      "step": 3473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.997549507381681e-05,
+      "loss": 0.7452,
+      "step": 3474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9974201342755703e-05,
+      "loss": 0.7054,
+      "step": 3475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9972907245043988e-05,
+      "loss": 0.6888,
+      "step": 3476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9971612780741177e-05,
+      "loss": 0.649,
+      "step": 3477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.997031794990678e-05,
+      "loss": 0.7146,
+      "step": 3478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9969022752600336e-05,
+      "loss": 0.7047,
+      "step": 3479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99677271888814e-05,
+      "loss": 0.7298,
+      "step": 3480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996643125880954e-05,
+      "loss": 0.7298,
+      "step": 3481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996513496244434e-05,
+      "loss": 0.6843,
+      "step": 3482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9963838299845405e-05,
+      "loss": 0.7553,
+      "step": 3483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996254127107236e-05,
+      "loss": 0.7608,
+      "step": 3484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9961243876184832e-05,
+      "loss": 0.76,
+      "step": 3485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.995994611524248e-05,
+      "loss": 0.7378,
+      "step": 3486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9958647988304976e-05,
+      "loss": 0.7725,
+      "step": 3487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9957349495432004e-05,
+      "loss": 0.7407,
+      "step": 3488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.995605063668327e-05,
+      "loss": 0.7739,
+      "step": 3489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9954751412118497e-05,
+      "loss": 0.7348,
+      "step": 3490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.995345182179742e-05,
+      "loss": 0.6957,
+      "step": 3491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9952151865779793e-05,
+      "loss": 0.7335,
+      "step": 3492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9950851544125388e-05,
+      "loss": 0.6911,
+      "step": 3493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9949550856893996e-05,
+      "loss": 0.7159,
+      "step": 3494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9948249804145417e-05,
+      "loss": 0.7187,
+      "step": 3495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9946948385939478e-05,
+      "loss": 0.74,
+      "step": 3496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9945646602336015e-05,
+      "loss": 0.7039,
+      "step": 3497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9944344453394885e-05,
+      "loss": 0.8316,
+      "step": 3498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.994304193917596e-05,
+      "loss": 0.7542,
+      "step": 3499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.994173905973913e-05,
+      "loss": 0.7112,
+      "step": 3500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9940435815144297e-05,
+      "loss": 0.6995,
+      "step": 3501
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9939132205451386e-05,
+      "loss": 0.7227,
+      "step": 3502
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9937828230720337e-05,
+      "loss": 0.6425,
+      "step": 3503
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9936523891011105e-05,
+      "loss": 0.7112,
+      "step": 3504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9935219186383665e-05,
+      "loss": 0.6636,
+      "step": 3505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9933914116898005e-05,
+      "loss": 0.7184,
+      "step": 3506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9932608682614133e-05,
+      "loss": 0.7033,
+      "step": 3507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.993130288359207e-05,
+      "loss": 0.7248,
+      "step": 3508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9929996719891858e-05,
+      "loss": 0.7785,
+      "step": 3509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9928690191573554e-05,
+      "loss": 0.7449,
+      "step": 3510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992738329869723e-05,
+      "loss": 0.707,
+      "step": 3511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992607604132298e-05,
+      "loss": 0.68,
+      "step": 3512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924768419510908e-05,
+      "loss": 0.7033,
+      "step": 3513
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9923460433321138e-05,
+      "loss": 0.7359,
+      "step": 3514
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992215208281381e-05,
+      "loss": 0.6916,
+      "step": 3515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9920843368049085e-05,
+      "loss": 0.6903,
+      "step": 3516
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9919534289087135e-05,
+      "loss": 0.7281,
+      "step": 3517
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.991822484598815e-05,
+      "loss": 0.7133,
+      "step": 3518
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.991691503881234e-05,
+      "loss": 0.6451,
+      "step": 3519
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9915604867619924e-05,
+      "loss": 0.6847,
+      "step": 3520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9914294332471152e-05,
+      "loss": 0.6946,
+      "step": 3521
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9912983433426272e-05,
+      "loss": 0.7868,
+      "step": 3522
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.991167217054557e-05,
+      "loss": 0.7019,
+      "step": 3523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9910360543889328e-05,
+      "loss": 0.6764,
+      "step": 3524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9909048553517854e-05,
+      "loss": 0.8072,
+      "step": 3525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990773619949148e-05,
+      "loss": 0.6979,
+      "step": 3526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990642348187054e-05,
+      "loss": 0.7161,
+      "step": 3527
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99051104007154e-05,
+      "loss": 0.7414,
+      "step": 3528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990379695608643e-05,
+      "loss": 0.7665,
+      "step": 3529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990248314804402e-05,
+      "loss": 0.722,
+      "step": 3530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990116897664858e-05,
+      "loss": 0.7886,
+      "step": 3531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9899854441960538e-05,
+      "loss": 0.7086,
+      "step": 3532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989853954404033e-05,
+      "loss": 0.7241,
+      "step": 3533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989722428294842e-05,
+      "loss": 0.6418,
+      "step": 3534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9895908658745284e-05,
+      "loss": 0.7989,
+      "step": 3535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989459267149141e-05,
+      "loss": 0.6901,
+      "step": 3536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893276321247304e-05,
+      "loss": 0.7562,
+      "step": 3537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891959608073497e-05,
+      "loss": 0.8179,
+      "step": 3538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989064253203053e-05,
+      "loss": 0.7298,
+      "step": 3539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889325093178955e-05,
+      "loss": 0.7424,
+      "step": 3540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888007291579358e-05,
+      "loss": 0.7477,
+      "step": 3541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886689127292325e-05,
+      "loss": 0.7095,
+      "step": 3542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885370600378463e-05,
+      "loss": 0.7853,
+      "step": 3543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884051710898402e-05,
+      "loss": 0.725,
+      "step": 3544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988273245891278e-05,
+      "loss": 0.6759,
+      "step": 3545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881412844482262e-05,
+      "loss": 0.6661,
+      "step": 3546
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880092867667515e-05,
+      "loss": 0.6869,
+      "step": 3547
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878772528529233e-05,
+      "loss": 0.6748,
+      "step": 3548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9877451827128127e-05,
+      "loss": 0.6883,
+      "step": 3549
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876130763524928e-05,
+      "loss": 0.7027,
+      "step": 3550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874809337780364e-05,
+      "loss": 0.6808,
+      "step": 3551
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873487549955206e-05,
+      "loss": 0.7451,
+      "step": 3552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872165400110223e-05,
+      "loss": 0.6767,
+      "step": 3553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987084288830621e-05,
+      "loss": 0.7293,
+      "step": 3554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869520014603974e-05,
+      "loss": 0.7165,
+      "step": 3555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986819677906434e-05,
+      "loss": 0.6993,
+      "step": 3556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866873181748148e-05,
+      "loss": 0.7109,
+      "step": 3557
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986554922271626e-05,
+      "loss": 0.6947,
+      "step": 3558
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864224902029547e-05,
+      "loss": 0.6867,
+      "step": 3559
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986290021974891e-05,
+      "loss": 0.6698,
+      "step": 3560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861575175935246e-05,
+      "loss": 0.7162,
+      "step": 3561
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860249770649485e-05,
+      "loss": 0.7721,
+      "step": 3562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9858924003952574e-05,
+      "loss": 0.7056,
+      "step": 3563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985759787590546e-05,
+      "loss": 0.7135,
+      "step": 3564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9856271386569126e-05,
+      "loss": 0.7656,
+      "step": 3565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985494453600456e-05,
+      "loss": 0.684,
+      "step": 3566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985361732427277e-05,
+      "loss": 0.6927,
+      "step": 3567
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9852289751434782e-05,
+      "loss": 0.7524,
+      "step": 3568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9850961817551637e-05,
+      "loss": 0.7346,
+      "step": 3569
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9849633522684395e-05,
+      "loss": 0.7646,
+      "step": 3570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.984830486689413e-05,
+      "loss": 0.7522,
+      "step": 3571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9846975850241928e-05,
+      "loss": 0.6764,
+      "step": 3572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98456464727889e-05,
+      "loss": 0.6902,
+      "step": 3573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.984431673459617e-05,
+      "loss": 0.7527,
+      "step": 3574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.984298663572488e-05,
+      "loss": 0.7294,
+      "step": 3575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9841656176236185e-05,
+      "loss": 0.7153,
+      "step": 3576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9840325356191265e-05,
+      "loss": 0.7118,
+      "step": 3577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98389941756513e-05,
+      "loss": 0.7004,
+      "step": 3578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.983766263467751e-05,
+      "loss": 0.7518,
+      "step": 3579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9836330733331104e-05,
+      "loss": 0.6123,
+      "step": 3580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.983499847167334e-05,
+      "loss": 0.6888,
+      "step": 3581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9833665849765456e-05,
+      "loss": 0.7243,
+      "step": 3582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.983233286766874e-05,
+      "loss": 0.7591,
+      "step": 3583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9830999525444476e-05,
+      "loss": 0.734,
+      "step": 3584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9829665823153972e-05,
+      "loss": 0.7409,
+      "step": 3585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.982833176085855e-05,
+      "loss": 0.7908,
+      "step": 3586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9826997338619547e-05,
+      "loss": 0.7233,
+      "step": 3587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.982566255649832e-05,
+      "loss": 0.6161,
+      "step": 3588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9824327414556247e-05,
+      "loss": 0.6924,
+      "step": 3589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9822991912854717e-05,
+      "loss": 0.7593,
+      "step": 3590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9821656051455127e-05,
+      "loss": 0.7344,
+      "step": 3591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.982031983041891e-05,
+      "loss": 0.6863,
+      "step": 3592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98189832498075e-05,
+      "loss": 0.6986,
+      "step": 3593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.981764630968235e-05,
+      "loss": 0.8243,
+      "step": 3594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9816309010104937e-05,
+      "loss": 0.7989,
+      "step": 3595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9814971351136746e-05,
+      "loss": 0.7435,
+      "step": 3596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9813633332839286e-05,
+      "loss": 0.8102,
+      "step": 3597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.981229495527407e-05,
+      "loss": 0.6545,
+      "step": 3598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9810956218502645e-05,
+      "loss": 0.6861,
+      "step": 3599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9809617122586563e-05,
+      "loss": 0.6901,
+      "step": 3600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9808277667587396e-05,
+      "loss": 0.7463,
+      "step": 3601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.980693785356673e-05,
+      "loss": 0.6588,
+      "step": 3602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9805597680586166e-05,
+      "loss": 0.7102,
+      "step": 3603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.980425714870733e-05,
+      "loss": 0.6744,
+      "step": 3604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9802916257991853e-05,
+      "loss": 0.7767,
+      "step": 3605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9801575008501397e-05,
+      "loss": 0.7012,
+      "step": 3606
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9800233400297628e-05,
+      "loss": 0.6625,
+      "step": 3607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.979889143344223e-05,
+      "loss": 0.6673,
+      "step": 3608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9797549107996906e-05,
+      "loss": 0.7233,
+      "step": 3609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9796206424023384e-05,
+      "loss": 0.6746,
+      "step": 3610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9794863381583387e-05,
+      "loss": 0.665,
+      "step": 3611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9793519980738682e-05,
+      "loss": 0.6302,
+      "step": 3612
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.979217622155103e-05,
+      "loss": 0.6906,
+      "step": 3613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9790832104082214e-05,
+      "loss": 0.6904,
+      "step": 3614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9789487628394037e-05,
+      "loss": 0.7489,
+      "step": 3615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9788142794548322e-05,
+      "loss": 0.8204,
+      "step": 3616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.97867976026069e-05,
+      "loss": 0.7046,
+      "step": 3617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.978545205263163e-05,
+      "loss": 0.712,
+      "step": 3618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.978410614468437e-05,
+      "loss": 0.7506,
+      "step": 3619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9782759878827007e-05,
+      "loss": 0.7296,
+      "step": 3620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9781413255121444e-05,
+      "loss": 0.7157,
+      "step": 3621
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9780066273629597e-05,
+      "loss": 0.7688,
+      "step": 3622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9778718934413398e-05,
+      "loss": 0.785,
+      "step": 3623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.97773712375348e-05,
+      "loss": 0.668,
+      "step": 3624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.977602318305577e-05,
+      "loss": 0.7256,
+      "step": 3625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9774674771038287e-05,
+      "loss": 0.7726,
+      "step": 3626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9773326001544355e-05,
+      "loss": 0.6839,
+      "step": 3627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9771976874635987e-05,
+      "loss": 0.7135,
+      "step": 3628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9770627390375216e-05,
+      "loss": 0.7221,
+      "step": 3629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.976927754882409e-05,
+      "loss": 0.7173,
+      "step": 3630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9767927350044678e-05,
+      "loss": 0.7146,
+      "step": 3631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9766576794099056e-05,
+      "loss": 0.6695,
+      "step": 3632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9765225881049326e-05,
+      "loss": 0.6976,
+      "step": 3633
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.97638746109576e-05,
+      "loss": 0.7103,
+      "step": 3634
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9762522983886007e-05,
+      "loss": 0.7436,
+      "step": 3635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9761170999896703e-05,
+      "loss": 0.7111,
+      "step": 3636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9759818659051843e-05,
+      "loss": 0.7657,
+      "step": 3637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.975846596141361e-05,
+      "loss": 0.7847,
+      "step": 3638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9757112907044202e-05,
+      "loss": 0.7696,
+      "step": 3639
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9755759496005827e-05,
+      "loss": 0.7482,
+      "step": 3640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.975440572836072e-05,
+      "loss": 0.7319,
+      "step": 3641
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9753051604171127e-05,
+      "loss": 0.6837,
+      "step": 3642
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9751697123499306e-05,
+      "loss": 0.7164,
+      "step": 3643
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9750342286407536e-05,
+      "loss": 0.7224,
+      "step": 3644
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.974898709295811e-05,
+      "loss": 0.7346,
+      "step": 3645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9747631543213344e-05,
+      "loss": 0.6441,
+      "step": 3646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9746275637235567e-05,
+      "loss": 0.7036,
+      "step": 3647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9744919375087117e-05,
+      "loss": 0.7289,
+      "step": 3648
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9743562756830357e-05,
+      "loss": 0.6844,
+      "step": 3649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9742205782527664e-05,
+      "loss": 0.6975,
+      "step": 3650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.974084845224143e-05,
+      "loss": 0.8059,
+      "step": 3651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.973949076603407e-05,
+      "loss": 0.6507,
+      "step": 3652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9738132723968002e-05,
+      "loss": 0.807,
+      "step": 3653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9736774326105675e-05,
+      "loss": 0.7165,
+      "step": 3654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.973541557250954e-05,
+      "loss": 0.6482,
+      "step": 3655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.973405646324208e-05,
+      "loss": 0.7425,
+      "step": 3656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.973269699836578e-05,
+      "loss": 0.7543,
+      "step": 3657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.973133717794315e-05,
+      "loss": 0.6838,
+      "step": 3658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9729977002036714e-05,
+      "loss": 0.7725,
+      "step": 3659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9728616470709012e-05,
+      "loss": 0.7322,
+      "step": 3660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.97272555840226e-05,
+      "loss": 0.6605,
+      "step": 3661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.972589434204006e-05,
+      "loss": 0.753,
+      "step": 3662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9724532744823967e-05,
+      "loss": 0.7414,
+      "step": 3663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9723170792436933e-05,
+      "loss": 0.7007,
+      "step": 3664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9721808484941585e-05,
+      "loss": 0.738,
+      "step": 3665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9720445822400553e-05,
+      "loss": 0.7132,
+      "step": 3666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9719082804876493e-05,
+      "loss": 0.7348,
+      "step": 3667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9717719432432086e-05,
+      "loss": 0.6915,
+      "step": 3668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9716355705130006e-05,
+      "loss": 0.7473,
+      "step": 3669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9714991623032962e-05,
+      "loss": 0.7114,
+      "step": 3670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9713627186203677e-05,
+      "loss": 0.6958,
+      "step": 3671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9712262394704882e-05,
+      "loss": 0.6601,
+      "step": 3672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.971089724859933e-05,
+      "loss": 0.7077,
+      "step": 3673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9709531747949798e-05,
+      "loss": 0.7319,
+      "step": 3674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9708165892819062e-05,
+      "loss": 0.6668,
+      "step": 3675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.970679968326993e-05,
+      "loss": 0.649,
+      "step": 3676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9705433119365214e-05,
+      "loss": 0.755,
+      "step": 3677
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9704066201167747e-05,
+      "loss": 0.7462,
+      "step": 3678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.970269892874039e-05,
+      "loss": 0.7286,
+      "step": 3679
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9701331302145995e-05,
+      "loss": 0.7692,
+      "step": 3680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9699963321447455e-05,
+      "loss": 0.6339,
+      "step": 3681
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.969859498670767e-05,
+      "loss": 0.7195,
+      "step": 3682
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9697226297989548e-05,
+      "loss": 0.7561,
+      "step": 3683
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9695857255356027e-05,
+      "loss": 0.7255,
+      "step": 3684
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9694487858870053e-05,
+      "loss": 0.7672,
+      "step": 3685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.969311810859459e-05,
+      "loss": 0.7083,
+      "step": 3686
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9691748004592617e-05,
+      "loss": 0.7498,
+      "step": 3687
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9690377546927135e-05,
+      "loss": 0.7591,
+      "step": 3688
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9689006735661153e-05,
+      "loss": 0.7369,
+      "step": 3689
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9687635570857702e-05,
+      "loss": 0.7432,
+      "step": 3690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9686264052579827e-05,
+      "loss": 0.7115,
+      "step": 3691
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9684892180890594e-05,
+      "loss": 0.7467,
+      "step": 3692
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9683519955853077e-05,
+      "loss": 0.6005,
+      "step": 3693
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9682147377530368e-05,
+      "loss": 0.7214,
+      "step": 3694
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.968077444598558e-05,
+      "loss": 0.7358,
+      "step": 3695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9679401161281845e-05,
+      "loss": 0.7222,
+      "step": 3696
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.96780275234823e-05,
+      "loss": 0.7013,
+      "step": 3697
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9676653532650105e-05,
+      "loss": 0.7404,
+      "step": 3698
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9675279188848435e-05,
+      "loss": 0.6681,
+      "step": 3699
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9673904492140487e-05,
+      "loss": 0.7379,
+      "step": 3700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.967252944258946e-05,
+      "loss": 0.7009,
+      "step": 3701
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9671154040258588e-05,
+      "loss": 0.6795,
+      "step": 3702
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9669778285211103e-05,
+      "loss": 0.6914,
+      "step": 3703
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.966840217751027e-05,
+      "loss": 0.6435,
+      "step": 3704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9667025717219356e-05,
+      "loss": 0.7454,
+      "step": 3705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9665648904401653e-05,
+      "loss": 0.6694,
+      "step": 3706
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.966427173912046e-05,
+      "loss": 0.6853,
+      "step": 3707
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9662894221439106e-05,
+      "loss": 0.7578,
+      "step": 3708
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9661516351420927e-05,
+      "loss": 0.6701,
+      "step": 3709
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9660138129129274e-05,
+      "loss": 0.7212,
+      "step": 3710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.965875955462752e-05,
+      "loss": 0.7131,
+      "step": 3711
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.965738062797905e-05,
+      "loss": 0.6832,
+      "step": 3712
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.965600134924727e-05,
+      "loss": 0.734,
+      "step": 3713
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9654621718495593e-05,
+      "loss": 0.7042,
+      "step": 3714
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9653241735787454e-05,
+      "loss": 0.7022,
+      "step": 3715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.965186140118631e-05,
+      "loss": 0.8068,
+      "step": 3716
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9650480714755623e-05,
+      "loss": 0.705,
+      "step": 3717
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9649099676558878e-05,
+      "loss": 0.71,
+      "step": 3718
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.964771828665958e-05,
+      "loss": 0.7611,
+      "step": 3719
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9646336545121233e-05,
+      "loss": 0.6766,
+      "step": 3720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.964495445200738e-05,
+      "loss": 0.7243,
+      "step": 3721
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.964357200738156e-05,
+      "loss": 0.7263,
+      "step": 3722
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9642189211307347e-05,
+      "loss": 0.7495,
+      "step": 3723
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.964080606384831e-05,
+      "loss": 0.7118,
+      "step": 3724
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9639422565068057e-05,
+      "loss": 0.7219,
+      "step": 3725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9638038715030192e-05,
+      "loss": 0.7019,
+      "step": 3726
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.963665451379835e-05,
+      "loss": 0.7658,
+      "step": 3727
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9635269961436175e-05,
+      "loss": 0.7146,
+      "step": 3728
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.963388505800732e-05,
+      "loss": 0.6526,
+      "step": 3729
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9632499803575474e-05,
+      "loss": 0.7287,
+      "step": 3730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9631114198204325e-05,
+      "loss": 0.6887,
+      "step": 3731
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.962972824195758e-05,
+      "loss": 0.6954,
+      "step": 3732
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.962834193489897e-05,
+      "loss": 0.7613,
+      "step": 3733
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9626955277092237e-05,
+      "loss": 0.7255,
+      "step": 3734
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.962556826860113e-05,
+      "loss": 0.7027,
+      "step": 3735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9624180909489436e-05,
+      "loss": 0.7535,
+      "step": 3736
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9622793199820936e-05,
+      "loss": 0.6969,
+      "step": 3737
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9621405139659438e-05,
+      "loss": 0.6571,
+      "step": 3738
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9620016729068765e-05,
+      "loss": 0.7255,
+      "step": 3739
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9618627968112756e-05,
+      "loss": 0.6362,
+      "step": 3740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.961723885685527e-05,
+      "loss": 0.615,
+      "step": 3741
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9615849395360167e-05,
+      "loss": 0.7231,
+      "step": 3742
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9614459583691347e-05,
+      "loss": 0.7621,
+      "step": 3743
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9613069421912704e-05,
+      "loss": 0.7131,
+      "step": 3744
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9611678910088157e-05,
+      "loss": 0.7079,
+      "step": 3745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.961028804828165e-05,
+      "loss": 0.6899,
+      "step": 3746
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9608896836557122e-05,
+      "loss": 0.7685,
+      "step": 3747
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.960750527497855e-05,
+      "loss": 0.6245,
+      "step": 3748
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.960611336360991e-05,
+      "loss": 0.7021,
+      "step": 3749
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9604721102515208e-05,
+      "loss": 0.8121,
+      "step": 3750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9603328491758455e-05,
+      "loss": 0.7261,
+      "step": 3751
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.960193553140369e-05,
+      "loss": 0.6773,
+      "step": 3752
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9600542221514946e-05,
+      "loss": 0.7509,
+      "step": 3753
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9599148562156304e-05,
+      "loss": 0.7564,
+      "step": 3754
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9597754553391834e-05,
+      "loss": 0.6937,
+      "step": 3755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9596360195285633e-05,
+      "loss": 0.7506,
+      "step": 3756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9594965487901815e-05,
+      "loss": 0.7787,
+      "step": 3757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9593570431304507e-05,
+      "loss": 0.7025,
+      "step": 3758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9592175025557854e-05,
+      "loss": 0.7657,
+      "step": 3759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9590779270726012e-05,
+      "loss": 0.791,
+      "step": 3760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9589383166873164e-05,
+      "loss": 0.7232,
+      "step": 3761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95879867140635e-05,
+      "loss": 0.7322,
+      "step": 3762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9586589912361227e-05,
+      "loss": 0.7658,
+      "step": 3763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958519276183057e-05,
+      "loss": 0.764,
+      "step": 3764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9583795262535768e-05,
+      "loss": 0.7354,
+      "step": 3765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9582397414541078e-05,
+      "loss": 0.6949,
+      "step": 3766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9580999217910774e-05,
+      "loss": 0.6896,
+      "step": 3767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9579600672709143e-05,
+      "loss": 0.6879,
+      "step": 3768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9578201779000492e-05,
+      "loss": 0.7254,
+      "step": 3769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.957680253684914e-05,
+      "loss": 0.7123,
+      "step": 3770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9575402946319423e-05,
+      "loss": 0.7292,
+      "step": 3771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9574003007475693e-05,
+      "loss": 0.7011,
+      "step": 3772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9572602720382323e-05,
+      "loss": 0.6964,
+      "step": 3773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.957120208510369e-05,
+      "loss": 0.6864,
+      "step": 3774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95698011017042e-05,
+      "loss": 0.7308,
+      "step": 3775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.956839977024827e-05,
+      "loss": 0.6724,
+      "step": 3776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.956699809080033e-05,
+      "loss": 0.6765,
+      "step": 3777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.956559606342483e-05,
+      "loss": 0.7545,
+      "step": 3778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9564193688186235e-05,
+      "loss": 0.719,
+      "step": 3779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9562790965149026e-05,
+      "loss": 0.7672,
+      "step": 3780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95613878943777e-05,
+      "loss": 0.6901,
+      "step": 3781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9559984475936766e-05,
+      "loss": 0.7719,
+      "step": 3782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9558580709890756e-05,
+      "loss": 0.7327,
+      "step": 3783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9557176596304213e-05,
+      "loss": 0.6812,
+      "step": 3784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9555772135241702e-05,
+      "loss": 0.7743,
+      "step": 3785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9554367326767794e-05,
+      "loss": 0.7635,
+      "step": 3786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.955296217094708e-05,
+      "loss": 0.7127,
+      "step": 3787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9551556667844178e-05,
+      "loss": 0.7312,
+      "step": 3788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9550150817523703e-05,
+      "loss": 0.7354,
+      "step": 3789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95487446200503e-05,
+      "loss": 0.7407,
+      "step": 3790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9547338075488622e-05,
+      "loss": 0.7096,
+      "step": 3791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954593118390335e-05,
+      "loss": 0.6819,
+      "step": 3792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954452394535916e-05,
+      "loss": 0.6906,
+      "step": 3793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9543116359920762e-05,
+      "loss": 0.6685,
+      "step": 3794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954170842765288e-05,
+      "loss": 0.7357,
+      "step": 3795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9540300148620245e-05,
+      "loss": 0.7276,
+      "step": 3796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953889152288761e-05,
+      "loss": 0.7196,
+      "step": 3797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9537482550519745e-05,
+      "loss": 0.6924,
+      "step": 3798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953607323158143e-05,
+      "loss": 0.679,
+      "step": 3799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953466356613747e-05,
+      "loss": 0.6983,
+      "step": 3800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9533253554252678e-05,
+      "loss": 0.6725,
+      "step": 3801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9531843195991887e-05,
+      "loss": 0.7935,
+      "step": 3802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9530432491419947e-05,
+      "loss": 0.7216,
+      "step": 3803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9529021440601714e-05,
+      "loss": 0.6476,
+      "step": 3804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952761004360207e-05,
+      "loss": 0.679,
+      "step": 3805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952619830048592e-05,
+      "loss": 0.814,
+      "step": 3806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9524786211318164e-05,
+      "loss": 0.769,
+      "step": 3807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9523373776163733e-05,
+      "loss": 0.6963,
+      "step": 3808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9521960995087572e-05,
+      "loss": 0.7002,
+      "step": 3809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9520547868154638e-05,
+      "loss": 0.6926,
+      "step": 3810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951913439542991e-05,
+      "loss": 0.6589,
+      "step": 3811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951772057697837e-05,
+      "loss": 0.6927,
+      "step": 3812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9516306412865032e-05,
+      "loss": 0.7319,
+      "step": 3813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9514891903154918e-05,
+      "loss": 0.7268,
+      "step": 3814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951347704791307e-05,
+      "loss": 0.6777,
+      "step": 3815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951206184720453e-05,
+      "loss": 0.692,
+      "step": 3816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9510646301094383e-05,
+      "loss": 0.6976,
+      "step": 3817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9509230409647707e-05,
+      "loss": 0.7085,
+      "step": 3818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9507814172929604e-05,
+      "loss": 0.7889,
+      "step": 3819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9506397591005196e-05,
+      "loss": 0.731,
+      "step": 3820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9504980663939616e-05,
+      "loss": 0.7,
+      "step": 3821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950356339179801e-05,
+      "loss": 0.7048,
+      "step": 3822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950214577464555e-05,
+      "loss": 0.6489,
+      "step": 3823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950072781254741e-05,
+      "loss": 0.7833,
+      "step": 3824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9499309505568792e-05,
+      "loss": 0.6955,
+      "step": 3825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9497890853774907e-05,
+      "loss": 0.7006,
+      "step": 3826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9496471857230986e-05,
+      "loss": 0.7356,
+      "step": 3827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9495052516002274e-05,
+      "loss": 0.6775,
+      "step": 3828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.949363283015403e-05,
+      "loss": 0.6219,
+      "step": 3829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9492212799751533e-05,
+      "loss": 0.6146,
+      "step": 3830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9490792424860072e-05,
+      "loss": 0.8507,
+      "step": 3831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9489371705544958e-05,
+      "loss": 0.7036,
+      "step": 3832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9487950641871515e-05,
+      "loss": 0.7088,
+      "step": 3833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9486529233905082e-05,
+      "loss": 0.7403,
+      "step": 3834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9485107481711015e-05,
+      "loss": 0.74,
+      "step": 3835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9483685385354685e-05,
+      "loss": 0.7674,
+      "step": 3836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.948226294490148e-05,
+      "loss": 0.7006,
+      "step": 3837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9480840160416803e-05,
+      "loss": 0.7192,
+      "step": 3838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9479417031966073e-05,
+      "loss": 0.6857,
+      "step": 3839
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9477993559614725e-05,
+      "loss": 0.7418,
+      "step": 3840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.947656974342821e-05,
+      "loss": 0.7069,
+      "step": 3841
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9475145583471997e-05,
+      "loss": 0.664,
+      "step": 3842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9473721079811565e-05,
+      "loss": 0.7019,
+      "step": 3843
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.947229623251241e-05,
+      "loss": 0.6755,
+      "step": 3844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9470871041640052e-05,
+      "loss": 0.6982,
+      "step": 3845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9469445507260016e-05,
+      "loss": 0.7784,
+      "step": 3846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9468019629437846e-05,
+      "loss": 0.7013,
+      "step": 3847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.946659340823911e-05,
+      "loss": 0.7205,
+      "step": 3848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9465166843729385e-05,
+      "loss": 0.5712,
+      "step": 3849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.946373993597425e-05,
+      "loss": 0.7782,
+      "step": 3850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.946231268503933e-05,
+      "loss": 0.6702,
+      "step": 3851
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9460885090990245e-05,
+      "loss": 0.6741,
+      "step": 3852
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945945715389263e-05,
+      "loss": 0.7718,
+      "step": 3853
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9458028873812147e-05,
+      "loss": 0.748,
+      "step": 3854
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9456600250814465e-05,
+      "loss": 0.7802,
+      "step": 3855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945517128496527e-05,
+      "loss": 0.7776,
+      "step": 3856
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9453741976330265e-05,
+      "loss": 0.7498,
+      "step": 3857
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9452312324975174e-05,
+      "loss": 0.6802,
+      "step": 3858
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9450882330965728e-05,
+      "loss": 0.6927,
+      "step": 3859
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.944945199436768e-05,
+      "loss": 0.7626,
+      "step": 3860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9448021315246793e-05,
+      "loss": 0.7842,
+      "step": 3861
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9446590293668854e-05,
+      "loss": 0.7984,
+      "step": 3862
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9445158929699653e-05,
+      "loss": 0.7197,
+      "step": 3863
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.944372722340501e-05,
+      "loss": 0.6334,
+      "step": 3864
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9442295174850753e-05,
+      "loss": 0.6644,
+      "step": 3865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9440862784102727e-05,
+      "loss": 0.7193,
+      "step": 3866
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.943943005122679e-05,
+      "loss": 0.7403,
+      "step": 3867
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.943799697628882e-05,
+      "loss": 0.677,
+      "step": 3868
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9436563559354712e-05,
+      "loss": 0.7087,
+      "step": 3869
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9435129800490376e-05,
+      "loss": 0.7276,
+      "step": 3870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9433695699761724e-05,
+      "loss": 0.6726,
+      "step": 3871
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9432261257234704e-05,
+      "loss": 0.6922,
+      "step": 3872
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9430826472975274e-05,
+      "loss": 0.7105,
+      "step": 3873
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9429391347049398e-05,
+      "loss": 0.7882,
+      "step": 3874
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.942795587952307e-05,
+      "loss": 0.705,
+      "step": 3875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9426520070462278e-05,
+      "loss": 0.73,
+      "step": 3876
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9425083919933056e-05,
+      "loss": 0.6771,
+      "step": 3877
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9423647428001432e-05,
+      "loss": 0.7188,
+      "step": 3878
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9422210594733453e-05,
+      "loss": 0.7175,
+      "step": 3879
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9420773420195185e-05,
+      "loss": 0.6558,
+      "step": 3880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9419335904452706e-05,
+      "loss": 0.6767,
+      "step": 3881
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9417898047572122e-05,
+      "loss": 0.6968,
+      "step": 3882
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9416459849619532e-05,
+      "loss": 0.7075,
+      "step": 3883
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9415021310661074e-05,
+      "loss": 0.7278,
+      "step": 3884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9413582430762885e-05,
+      "loss": 0.689,
+      "step": 3885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.941214320999113e-05,
+      "loss": 0.6788,
+      "step": 3886
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.941070364841198e-05,
+      "loss": 0.7547,
+      "step": 3887
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9409263746091627e-05,
+      "loss": 0.7343,
+      "step": 3888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9407823503096276e-05,
+      "loss": 0.6934,
+      "step": 3889
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.940638291949215e-05,
+      "loss": 0.7343,
+      "step": 3890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9404941995345482e-05,
+      "loss": 0.7718,
+      "step": 3891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.940350073072253e-05,
+      "loss": 0.7099,
+      "step": 3892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9402059125689565e-05,
+      "loss": 0.6888,
+      "step": 3893
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9400617180312865e-05,
+      "loss": 0.6952,
+      "step": 3894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9399174894658734e-05,
+      "loss": 0.6733,
+      "step": 3895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9397732268793488e-05,
+      "loss": 0.6441,
+      "step": 3896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9396289302783456e-05,
+      "loss": 0.8085,
+      "step": 3897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9394845996694984e-05,
+      "loss": 0.7474,
+      "step": 3898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9393402350594437e-05,
+      "loss": 0.7272,
+      "step": 3899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9391958364548196e-05,
+      "loss": 0.7762,
+      "step": 3900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9390514038622648e-05,
+      "loss": 0.6694,
+      "step": 3901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.938906937288421e-05,
+      "loss": 0.7201,
+      "step": 3902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93876243673993e-05,
+      "loss": 0.7532,
+      "step": 3903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9386179022234368e-05,
+      "loss": 0.7754,
+      "step": 3904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.938473333745586e-05,
+      "loss": 0.6946,
+      "step": 3905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9383287313130255e-05,
+      "loss": 0.7387,
+      "step": 3906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9381840949324038e-05,
+      "loss": 0.7056,
+      "step": 3907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9380394246103714e-05,
+      "loss": 0.7019,
+      "step": 3908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9378947203535797e-05,
+      "loss": 0.6893,
+      "step": 3909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9377499821686827e-05,
+      "loss": 0.7546,
+      "step": 3910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.937605210062335e-05,
+      "loss": 0.6865,
+      "step": 3911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9374604040411935e-05,
+      "loss": 0.7203,
+      "step": 3912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9373155641119163e-05,
+      "loss": 0.6866,
+      "step": 3913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.937170690281163e-05,
+      "loss": 0.705,
+      "step": 3914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9370257825555948e-05,
+      "loss": 0.7049,
+      "step": 3915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9368808409418744e-05,
+      "loss": 0.6898,
+      "step": 3916
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.936735865446666e-05,
+      "loss": 0.6809,
+      "step": 3917
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.936590856076636e-05,
+      "loss": 0.7227,
+      "step": 3918
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.936445812838452e-05,
+      "loss": 0.7818,
+      "step": 3919
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9363007357387822e-05,
+      "loss": 0.6891,
+      "step": 3920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.936155624784298e-05,
+      "loss": 0.7336,
+      "step": 3921
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.936010479981671e-05,
+      "loss": 0.6941,
+      "step": 3922
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.935865301337575e-05,
+      "loss": 0.6853,
+      "step": 3923
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9357200888586857e-05,
+      "loss": 0.8169,
+      "step": 3924
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9355748425516796e-05,
+      "loss": 0.7349,
+      "step": 3925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9354295624232347e-05,
+      "loss": 0.7466,
+      "step": 3926
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9352842484800314e-05,
+      "loss": 0.7736,
+      "step": 3927
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.935138900728751e-05,
+      "loss": 0.7158,
+      "step": 3928
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.934993519176077e-05,
+      "loss": 0.7136,
+      "step": 3929
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.934848103828693e-05,
+      "loss": 0.8159,
+      "step": 3930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.934702654693286e-05,
+      "loss": 0.6902,
+      "step": 3931
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9345571717765437e-05,
+      "loss": 0.7031,
+      "step": 3932
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9344116550851547e-05,
+      "loss": 0.7726,
+      "step": 3933
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93426610462581e-05,
+      "loss": 0.6375,
+      "step": 3934
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9341205204052023e-05,
+      "loss": 0.8196,
+      "step": 3935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.933974902430025e-05,
+      "loss": 0.6324,
+      "step": 3936
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9338292507069747e-05,
+      "loss": 0.6295,
+      "step": 3937
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9336835652427466e-05,
+      "loss": 0.6506,
+      "step": 3938
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9335378460440406e-05,
+      "loss": 0.6881,
+      "step": 3939
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9333920931175567e-05,
+      "loss": 0.689,
+      "step": 3940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.933246306469996e-05,
+      "loss": 0.7294,
+      "step": 3941
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9331004861080622e-05,
+      "loss": 0.7582,
+      "step": 3942
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9329546320384596e-05,
+      "loss": 0.718,
+      "step": 3943
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.932808744267895e-05,
+      "loss": 0.7338,
+      "step": 3944
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9326628228030762e-05,
+      "loss": 0.6991,
+      "step": 3945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.932516867650712e-05,
+      "loss": 0.6883,
+      "step": 3946
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.932370878817514e-05,
+      "loss": 0.6483,
+      "step": 3947
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9322248563101944e-05,
+      "loss": 0.6411,
+      "step": 3948
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.932078800135468e-05,
+      "loss": 0.7319,
+      "step": 3949
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9319327103000494e-05,
+      "loss": 0.6899,
+      "step": 3950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.931786586810656e-05,
+      "loss": 0.6409,
+      "step": 3951
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9316404296740068e-05,
+      "loss": 0.8064,
+      "step": 3952
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9314942388968217e-05,
+      "loss": 0.665,
+      "step": 3953
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.931348014485823e-05,
+      "loss": 0.6337,
+      "step": 3954
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9312017564477332e-05,
+      "loss": 0.6417,
+      "step": 3955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9310554647892786e-05,
+      "loss": 0.7354,
+      "step": 3956
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9309091395171837e-05,
+      "loss": 0.7123,
+      "step": 3957
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.930762780638178e-05,
+      "loss": 0.6982,
+      "step": 3958
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9306163881589906e-05,
+      "loss": 0.6647,
+      "step": 3959
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9304699620863528e-05,
+      "loss": 0.716,
+      "step": 3960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9303235024269965e-05,
+      "loss": 0.6958,
+      "step": 3961
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9301770091876563e-05,
+      "loss": 0.6793,
+      "step": 3962
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9300304823750678e-05,
+      "loss": 0.729,
+      "step": 3963
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9298839219959686e-05,
+      "loss": 0.7333,
+      "step": 3964
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.929737328057097e-05,
+      "loss": 0.72,
+      "step": 3965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9295907005651938e-05,
+      "loss": 0.6772,
+      "step": 3966
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9294440395270004e-05,
+      "loss": 0.7037,
+      "step": 3967
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9292973449492605e-05,
+      "loss": 0.7473,
+      "step": 3968
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9291506168387188e-05,
+      "loss": 0.8001,
+      "step": 3969
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9290038552021225e-05,
+      "loss": 0.7287,
+      "step": 3970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9288570600462186e-05,
+      "loss": 0.679,
+      "step": 3971
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9287102313777574e-05,
+      "loss": 0.6811,
+      "step": 3972
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9285633692034897e-05,
+      "loss": 0.7711,
+      "step": 3973
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9284164735301687e-05,
+      "loss": 0.673,
+      "step": 3974
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.928269544364548e-05,
+      "loss": 0.6593,
+      "step": 3975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9281225817133836e-05,
+      "loss": 0.8033,
+      "step": 3976
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9279755855834326e-05,
+      "loss": 0.7243,
+      "step": 3977
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9278285559814538e-05,
+      "loss": 0.6293,
+      "step": 3978
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9276814929142078e-05,
+      "loss": 0.7356,
+      "step": 3979
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.927534396388457e-05,
+      "loss": 0.7151,
+      "step": 3980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9273872664109635e-05,
+      "loss": 0.6845,
+      "step": 3981
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9272401029884933e-05,
+      "loss": 0.737,
+      "step": 3982
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.927092906127813e-05,
+      "loss": 0.7459,
+      "step": 3983
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.92694567583569e-05,
+      "loss": 0.7291,
+      "step": 3984
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.926798412118894e-05,
+      "loss": 0.6498,
+      "step": 3985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9266511149841967e-05,
+      "loss": 0.734,
+      "step": 3986
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9265037844383703e-05,
+      "loss": 0.7864,
+      "step": 3987
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9263564204881892e-05,
+      "loss": 0.7139,
+      "step": 3988
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9262090231404288e-05,
+      "loss": 0.6645,
+      "step": 3989
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9260615924018665e-05,
+      "loss": 0.7997,
+      "step": 3990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9259141282792817e-05,
+      "loss": 0.7108,
+      "step": 3991
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9257666307794538e-05,
+      "loss": 0.8065,
+      "step": 3992
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.925619099909165e-05,
+      "loss": 0.7206,
+      "step": 3993
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9254715356751993e-05,
+      "loss": 0.7699,
+      "step": 3994
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9253239380843412e-05,
+      "loss": 0.6872,
+      "step": 3995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9251763071433768e-05,
+      "loss": 0.6855,
+      "step": 3996
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9250286428590946e-05,
+      "loss": 0.6859,
+      "step": 3997
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.924880945238284e-05,
+      "loss": 0.7019,
+      "step": 3998
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.924733214287736e-05,
+      "loss": 0.759,
+      "step": 3999
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9245854500142434e-05,
+      "loss": 0.7119,
+      "step": 4000
+    },
+    {
+      "epoch": 0.26,
+      "eval_text_loss": 0.6687390804290771,
+      "eval_text_runtime": 1.8327,
+      "eval_text_samples_per_second": 40.377,
+      "eval_text_steps_per_second": 1.091,
+      "step": 4000
+    },
+    {
+      "epoch": 0.26,
+      "eval_svit_loss": 0.8611207604408264,
+      "eval_svit_runtime": 6.6057,
+      "eval_svit_samples_per_second": 123.53,
+      "eval_svit_steps_per_second": 1.968,
+      "step": 4000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9244376524246004e-05,
+      "loss": 0.7974,
+      "step": 4001
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.924289821525602e-05,
+      "loss": 0.7214,
+      "step": 4002
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9241419573240464e-05,
+      "loss": 0.7015,
+      "step": 4003
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.923994059826732e-05,
+      "loss": 0.72,
+      "step": 4004
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9238461290404584e-05,
+      "loss": 0.6968,
+      "step": 4005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9236981649720285e-05,
+      "loss": 0.7198,
+      "step": 4006
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9235501676282447e-05,
+      "loss": 0.7401,
+      "step": 4007
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9234021370159124e-05,
+      "loss": 0.7327,
+      "step": 4008
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9232540731418376e-05,
+      "loss": 0.7328,
+      "step": 4009
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9231059760128286e-05,
+      "loss": 0.6759,
+      "step": 4010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9229578456356947e-05,
+      "loss": 0.7139,
+      "step": 4011
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9228096820172467e-05,
+      "loss": 0.6609,
+      "step": 4012
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9226614851642973e-05,
+      "loss": 0.7685,
+      "step": 4013
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9225132550836606e-05,
+      "loss": 0.697,
+      "step": 4014
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9223649917821514e-05,
+      "loss": 0.6708,
+      "step": 4015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.922216695266588e-05,
+      "loss": 0.7436,
+      "step": 4016
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9220683655437887e-05,
+      "loss": 0.7024,
+      "step": 4017
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.921920002620573e-05,
+      "loss": 0.7788,
+      "step": 4018
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.921771606503763e-05,
+      "loss": 0.6931,
+      "step": 4019
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.921623177200182e-05,
+      "loss": 0.7218,
+      "step": 4020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9214747147166543e-05,
+      "loss": 0.7859,
+      "step": 4021
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9213262190600066e-05,
+      "loss": 0.712,
+      "step": 4022
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9211776902370662e-05,
+      "loss": 0.7404,
+      "step": 4023
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9210291282546634e-05,
+      "loss": 0.7206,
+      "step": 4024
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9208805331196274e-05,
+      "loss": 0.6885,
+      "step": 4025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9207319048387923e-05,
+      "loss": 0.7417,
+      "step": 4026
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9205832434189904e-05,
+      "loss": 0.7053,
+      "step": 4027
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.920434548867058e-05,
+      "loss": 0.6664,
+      "step": 4028
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9202858211898323e-05,
+      "loss": 0.7337,
+      "step": 4029
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.920137060394151e-05,
+      "loss": 0.7439,
+      "step": 4030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.919988266486854e-05,
+      "loss": 0.7129,
+      "step": 4031
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9198394394747832e-05,
+      "loss": 0.7659,
+      "step": 4032
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9196905793647818e-05,
+      "loss": 0.6822,
+      "step": 4033
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.919541686163694e-05,
+      "loss": 0.7329,
+      "step": 4034
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.919392759878366e-05,
+      "loss": 0.7837,
+      "step": 4035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.919243800515645e-05,
+      "loss": 0.6948,
+      "step": 4036
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9190948080823803e-05,
+      "loss": 0.7356,
+      "step": 4037
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9189457825854232e-05,
+      "loss": 0.7449,
+      "step": 4038
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9187967240316244e-05,
+      "loss": 0.7066,
+      "step": 4039
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9186476324278386e-05,
+      "loss": 0.6665,
+      "step": 4040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.918498507780921e-05,
+      "loss": 0.6919,
+      "step": 4041
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.918349350097728e-05,
+      "loss": 0.6734,
+      "step": 4042
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9182001593851177e-05,
+      "loss": 0.7028,
+      "step": 4043
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.91805093564995e-05,
+      "loss": 0.6818,
+      "step": 4044
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.917901678899086e-05,
+      "loss": 0.7797,
+      "step": 4045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9177523891393886e-05,
+      "loss": 0.6694,
+      "step": 4046
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9176030663777223e-05,
+      "loss": 0.6685,
+      "step": 4047
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.917453710620952e-05,
+      "loss": 0.7054,
+      "step": 4048
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9173043218759464e-05,
+      "loss": 0.6874,
+      "step": 4049
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9171549001495733e-05,
+      "loss": 0.6499,
+      "step": 4050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9170054454487036e-05,
+      "loss": 0.7385,
+      "step": 4051
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9168559577802088e-05,
+      "loss": 0.7178,
+      "step": 4052
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9167064371509623e-05,
+      "loss": 0.7633,
+      "step": 4053
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9165568835678393e-05,
+      "loss": 0.7965,
+      "step": 4054
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9164072970377162e-05,
+      "loss": 0.6729,
+      "step": 4055
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9162576775674707e-05,
+      "loss": 0.6492,
+      "step": 4056
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.916108025163982e-05,
+      "loss": 0.6578,
+      "step": 4057
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9159583398341315e-05,
+      "loss": 0.7213,
+      "step": 4058
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9158086215848014e-05,
+      "loss": 0.6566,
+      "step": 4059
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.915658870422876e-05,
+      "loss": 0.713,
+      "step": 4060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9155090863552405e-05,
+      "loss": 0.7075,
+      "step": 4061
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9153592693887822e-05,
+      "loss": 0.7024,
+      "step": 4062
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9152094195303893e-05,
+      "loss": 0.716,
+      "step": 4063
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.915059536786952e-05,
+      "loss": 0.6655,
+      "step": 4064
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.914909621165362e-05,
+      "loss": 0.7866,
+      "step": 4065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9147596726725123e-05,
+      "loss": 0.7022,
+      "step": 4066
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.914609691315297e-05,
+      "loss": 0.7304,
+      "step": 4067
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9144596771006128e-05,
+      "loss": 0.7301,
+      "step": 4068
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9143096300353572e-05,
+      "loss": 0.6929,
+      "step": 4069
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9141595501264286e-05,
+      "loss": 0.7103,
+      "step": 4070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9140094373807288e-05,
+      "loss": 0.7911,
+      "step": 4071
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9138592918051593e-05,
+      "loss": 0.7352,
+      "step": 4072
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9137091134066232e-05,
+      "loss": 0.7435,
+      "step": 4073
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9135589021920264e-05,
+      "loss": 0.7148,
+      "step": 4074
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9134086581682755e-05,
+      "loss": 0.756,
+      "step": 4075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9132583813422782e-05,
+      "loss": 0.7305,
+      "step": 4076
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9131080717209447e-05,
+      "loss": 0.7435,
+      "step": 4077
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9129577293111856e-05,
+      "loss": 0.7343,
+      "step": 4078
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9128073541199142e-05,
+      "loss": 0.8273,
+      "step": 4079
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9126569461540446e-05,
+      "loss": 0.6349,
+      "step": 4080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9125065054204916e-05,
+      "loss": 0.6577,
+      "step": 4081
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9123560319261736e-05,
+      "loss": 0.7418,
+      "step": 4082
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.912205525678009e-05,
+      "loss": 0.7167,
+      "step": 4083
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9120549866829176e-05,
+      "loss": 0.6657,
+      "step": 4084
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9119044149478213e-05,
+      "loss": 0.6517,
+      "step": 4085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9117538104796436e-05,
+      "loss": 0.6874,
+      "step": 4086
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.911603173285309e-05,
+      "loss": 0.6896,
+      "step": 4087
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9114525033717437e-05,
+      "loss": 0.7462,
+      "step": 4088
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9113018007458753e-05,
+      "loss": 0.6324,
+      "step": 4089
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9111510654146332e-05,
+      "loss": 0.7174,
+      "step": 4090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9110002973849487e-05,
+      "loss": 0.6876,
+      "step": 4091
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9108494966637532e-05,
+      "loss": 0.6929,
+      "step": 4092
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9106986632579808e-05,
+      "loss": 0.7371,
+      "step": 4093
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.910547797174567e-05,
+      "loss": 0.73,
+      "step": 4094
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.910396898420448e-05,
+      "loss": 0.7382,
+      "step": 4095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9102459670025626e-05,
+      "loss": 0.7294,
+      "step": 4096
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9100950029278506e-05,
+      "loss": 0.682,
+      "step": 4097
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9099440062032524e-05,
+      "loss": 0.695,
+      "step": 4098
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.909792976835712e-05,
+      "loss": 0.6982,
+      "step": 4099
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9096419148321728e-05,
+      "loss": 0.6829,
+      "step": 4100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.909490820199581e-05,
+      "loss": 0.6909,
+      "step": 4101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9093396929448834e-05,
+      "loss": 0.6997,
+      "step": 4102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9091885330750294e-05,
+      "loss": 0.7047,
+      "step": 4103
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.909037340596969e-05,
+      "loss": 0.7226,
+      "step": 4104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9088861155176537e-05,
+      "loss": 0.678,
+      "step": 4105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9087348578440372e-05,
+      "loss": 0.6726,
+      "step": 4106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9085835675830738e-05,
+      "loss": 0.7122,
+      "step": 4107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9084322447417202e-05,
+      "loss": 0.6801,
+      "step": 4108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.908280889326934e-05,
+      "loss": 0.7062,
+      "step": 4109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9081295013456744e-05,
+      "loss": 0.7343,
+      "step": 4110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9079780808049023e-05,
+      "loss": 0.7511,
+      "step": 4111
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9078266277115797e-05,
+      "loss": 0.685,
+      "step": 4112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.907675142072671e-05,
+      "loss": 0.6842,
+      "step": 4113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9075236238951405e-05,
+      "loss": 0.7161,
+      "step": 4114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.907372073185955e-05,
+      "loss": 0.683,
+      "step": 4115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.907220489952084e-05,
+      "loss": 0.6804,
+      "step": 4116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9070688742004962e-05,
+      "loss": 0.7338,
+      "step": 4117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9069172259381628e-05,
+      "loss": 0.7179,
+      "step": 4118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.906765545172057e-05,
+      "loss": 0.7023,
+      "step": 4119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9066138319091525e-05,
+      "loss": 0.687,
+      "step": 4120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9064620861564257e-05,
+      "loss": 0.6986,
+      "step": 4121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.906310307920853e-05,
+      "loss": 0.6533,
+      "step": 4122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9061584972094137e-05,
+      "loss": 0.6983,
+      "step": 4123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9060066540290876e-05,
+      "loss": 0.738,
+      "step": 4124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9058547783868564e-05,
+      "loss": 0.709,
+      "step": 4125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.905702870289704e-05,
+      "loss": 0.7747,
+      "step": 4126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9055509297446145e-05,
+      "loss": 0.6267,
+      "step": 4127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9053989567585737e-05,
+      "loss": 0.664,
+      "step": 4128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.90524695133857e-05,
+      "loss": 0.7119,
+      "step": 4129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9050949134915924e-05,
+      "loss": 0.6784,
+      "step": 4130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9049428432246308e-05,
+      "loss": 0.6877,
+      "step": 4131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9047907405446784e-05,
+      "loss": 0.6941,
+      "step": 4132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.904638605458728e-05,
+      "loss": 0.5992,
+      "step": 4133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.904486437973775e-05,
+      "loss": 0.6947,
+      "step": 4134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.904334238096816e-05,
+      "loss": 0.7272,
+      "step": 4135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9041820058348494e-05,
+      "loss": 0.6941,
+      "step": 4136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.904029741194874e-05,
+      "loss": 0.6539,
+      "step": 4137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.903877444183892e-05,
+      "loss": 0.6385,
+      "step": 4138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9037251148089048e-05,
+      "loss": 0.6823,
+      "step": 4139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9035727530769173e-05,
+      "loss": 0.791,
+      "step": 4140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9034203589949344e-05,
+      "loss": 0.6836,
+      "step": 4141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9032679325699636e-05,
+      "loss": 0.7145,
+      "step": 4142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.903115473809013e-05,
+      "loss": 0.7156,
+      "step": 4143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.902962982719093e-05,
+      "loss": 0.6602,
+      "step": 4144
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9028104593072147e-05,
+      "loss": 0.6925,
+      "step": 4145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9026579035803914e-05,
+      "loss": 0.7184,
+      "step": 4146
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9025053155456373e-05,
+      "loss": 0.762,
+      "step": 4147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9023526952099686e-05,
+      "loss": 0.7594,
+      "step": 4148
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9022000425804023e-05,
+      "loss": 0.7088,
+      "step": 4149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9020473576639578e-05,
+      "loss": 0.7287,
+      "step": 4150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9018946404676554e-05,
+      "loss": 0.6346,
+      "step": 4151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9017418909985165e-05,
+      "loss": 0.777,
+      "step": 4152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.901589109263565e-05,
+      "loss": 0.6629,
+      "step": 4153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9014362952698256e-05,
+      "loss": 0.7173,
+      "step": 4154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9012834490243244e-05,
+      "loss": 0.6467,
+      "step": 4155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9011305705340892e-05,
+      "loss": 0.697,
+      "step": 4156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9009776598061497e-05,
+      "loss": 0.6929,
+      "step": 4157
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9008247168475362e-05,
+      "loss": 0.7524,
+      "step": 4158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.900671741665282e-05,
+      "loss": 0.6389,
+      "step": 4159
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.900518734266419e-05,
+      "loss": 0.7226,
+      "step": 4160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.900365694657984e-05,
+      "loss": 0.6527,
+      "step": 4161
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.900212622847013e-05,
+      "loss": 0.7044,
+      "step": 4162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9000595188405446e-05,
+      "loss": 0.7747,
+      "step": 4163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8999063826456177e-05,
+      "loss": 0.6918,
+      "step": 4164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8997532142692746e-05,
+      "loss": 0.7114,
+      "step": 4165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8996000137185567e-05,
+      "loss": 0.7219,
+      "step": 4166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8994467810005084e-05,
+      "loss": 0.7631,
+      "step": 4167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8992935161221762e-05,
+      "loss": 0.7137,
+      "step": 4168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8991402190906062e-05,
+      "loss": 0.6752,
+      "step": 4169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8989868899128477e-05,
+      "loss": 0.7681,
+      "step": 4170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8988335285959497e-05,
+      "loss": 0.6991,
+      "step": 4171
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8986801351469645e-05,
+      "loss": 0.7353,
+      "step": 4172
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8985267095729444e-05,
+      "loss": 0.7194,
+      "step": 4173
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8983732518809445e-05,
+      "loss": 0.7323,
+      "step": 4174
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.898219762078021e-05,
+      "loss": 0.6621,
+      "step": 4175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8980662401712303e-05,
+      "loss": 0.6788,
+      "step": 4176
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8979126861676314e-05,
+      "loss": 0.7451,
+      "step": 4177
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8977591000742856e-05,
+      "loss": 0.6679,
+      "step": 4178
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8976054818982537e-05,
+      "loss": 0.7944,
+      "step": 4179
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8974518316466e-05,
+      "loss": 0.6919,
+      "step": 4180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8972981493263885e-05,
+      "loss": 0.7109,
+      "step": 4181
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8971444349446854e-05,
+      "loss": 0.6797,
+      "step": 4182
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.896990688508559e-05,
+      "loss": 0.7463,
+      "step": 4183
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.896836910025078e-05,
+      "loss": 0.6919,
+      "step": 4184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8966830995013135e-05,
+      "loss": 0.668,
+      "step": 4185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.896529256944337e-05,
+      "loss": 0.7329,
+      "step": 4186
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8963753823612226e-05,
+      "loss": 0.6998,
+      "step": 4187
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8962214757590454e-05,
+      "loss": 0.724,
+      "step": 4188
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.896067537144882e-05,
+      "loss": 0.7889,
+      "step": 4189
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8959135665258107e-05,
+      "loss": 0.6529,
+      "step": 4190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.89575956390891e-05,
+      "loss": 0.766,
+      "step": 4191
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8956055293012617e-05,
+      "loss": 0.793,
+      "step": 4192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.895451462709948e-05,
+      "loss": 0.692,
+      "step": 4193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.895297364142053e-05,
+      "loss": 0.6584,
+      "step": 4194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.895143233604662e-05,
+      "loss": 0.6819,
+      "step": 4195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.894989071104862e-05,
+      "loss": 0.7504,
+      "step": 4196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8948348766497408e-05,
+      "loss": 0.7045,
+      "step": 4197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.894680650246389e-05,
+      "loss": 0.6985,
+      "step": 4198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.894526391901897e-05,
+      "loss": 0.7566,
+      "step": 4199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8943721016233583e-05,
+      "loss": 0.7117,
+      "step": 4200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8942177794178667e-05,
+      "loss": 0.6847,
+      "step": 4201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.894063425292518e-05,
+      "loss": 0.7705,
+      "step": 4202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8939090392544086e-05,
+      "loss": 0.8616,
+      "step": 4203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8937546213106387e-05,
+      "loss": 0.7288,
+      "step": 4204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.893600171468307e-05,
+      "loss": 0.7289,
+      "step": 4205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8934456897345155e-05,
+      "loss": 0.7163,
+      "step": 4206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8932911761163674e-05,
+      "loss": 0.6502,
+      "step": 4207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.893136630620967e-05,
+      "loss": 0.6782,
+      "step": 4208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8929820532554203e-05,
+      "loss": 0.6915,
+      "step": 4209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8928274440268344e-05,
+      "loss": 0.6935,
+      "step": 4210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.892672802942319e-05,
+      "loss": 0.6962,
+      "step": 4211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8925181300089832e-05,
+      "loss": 0.6898,
+      "step": 4212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8923634252339398e-05,
+      "loss": 0.6883,
+      "step": 4213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.892208688624302e-05,
+      "loss": 0.6995,
+      "step": 4214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8920539201871837e-05,
+      "loss": 0.6998,
+      "step": 4215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8918991199297015e-05,
+      "loss": 0.7368,
+      "step": 4216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8917442878589737e-05,
+      "loss": 0.6286,
+      "step": 4217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8915894239821187e-05,
+      "loss": 0.73,
+      "step": 4218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8914345283062574e-05,
+      "loss": 0.6666,
+      "step": 4219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8912796008385113e-05,
+      "loss": 0.7309,
+      "step": 4220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.891124641586005e-05,
+      "loss": 0.704,
+      "step": 4221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8909696505558623e-05,
+      "loss": 0.7092,
+      "step": 4222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8908146277552102e-05,
+      "loss": 0.7148,
+      "step": 4223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8906595731911763e-05,
+      "loss": 0.698,
+      "step": 4224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8905044868708906e-05,
+      "loss": 0.7319,
+      "step": 4225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.890349368801483e-05,
+      "loss": 0.6964,
+      "step": 4226
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8901942189900868e-05,
+      "loss": 0.7946,
+      "step": 4227
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.890039037443835e-05,
+      "loss": 0.7333,
+      "step": 4228
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.889883824169863e-05,
+      "loss": 0.6759,
+      "step": 4229
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.889728579175307e-05,
+      "loss": 0.6969,
+      "step": 4230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8895733024673056e-05,
+      "loss": 0.72,
+      "step": 4231
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8894179940529984e-05,
+      "loss": 0.7166,
+      "step": 4232
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.889262653939526e-05,
+      "loss": 0.717,
+      "step": 4233
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.889107282134032e-05,
+      "loss": 0.7185,
+      "step": 4234
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8889518786436585e-05,
+      "loss": 0.7164,
+      "step": 4235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8887964434755524e-05,
+      "loss": 0.6574,
+      "step": 4236
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8886409766368594e-05,
+      "loss": 0.7434,
+      "step": 4237
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.888485478134729e-05,
+      "loss": 0.6841,
+      "step": 4238
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8883299479763103e-05,
+      "loss": 0.7026,
+      "step": 4239
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8881743861687546e-05,
+      "loss": 0.7787,
+      "step": 4240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8880187927192144e-05,
+      "loss": 0.6937,
+      "step": 4241
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8878631676348443e-05,
+      "loss": 0.6613,
+      "step": 4242
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8877075109227993e-05,
+      "loss": 0.673,
+      "step": 4243
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8875518225902367e-05,
+      "loss": 0.7389,
+      "step": 4244
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.887396102644315e-05,
+      "loss": 0.7097,
+      "step": 4245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8872403510921943e-05,
+      "loss": 0.6759,
+      "step": 4246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8870845679410357e-05,
+      "loss": 0.7014,
+      "step": 4247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8869287531980025e-05,
+      "loss": 0.7237,
+      "step": 4248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8867729068702583e-05,
+      "loss": 0.698,
+      "step": 4249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8866170289649693e-05,
+      "loss": 0.6742,
+      "step": 4250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8864611194893027e-05,
+      "loss": 0.6972,
+      "step": 4251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8863051784504272e-05,
+      "loss": 0.7832,
+      "step": 4252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8861492058555124e-05,
+      "loss": 0.6848,
+      "step": 4253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8859932017117302e-05,
+      "loss": 0.7387,
+      "step": 4254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.885837166026254e-05,
+      "loss": 0.6584,
+      "step": 4255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8856810988062577e-05,
+      "loss": 0.6916,
+      "step": 4256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8855250000589178e-05,
+      "loss": 0.712,
+      "step": 4257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.885368869791411e-05,
+      "loss": 0.748,
+      "step": 4258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8852127080109167e-05,
+      "loss": 0.7357,
+      "step": 4259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8850565147246144e-05,
+      "loss": 0.6502,
+      "step": 4260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8849002899396867e-05,
+      "loss": 0.6828,
+      "step": 4261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8847440336633158e-05,
+      "loss": 0.7211,
+      "step": 4262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8845877459026872e-05,
+      "loss": 0.6422,
+      "step": 4263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8844314266649865e-05,
+      "loss": 0.643,
+      "step": 4264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.884275075957401e-05,
+      "loss": 0.8154,
+      "step": 4265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8841186937871204e-05,
+      "loss": 0.7327,
+      "step": 4266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.883962280161334e-05,
+      "loss": 0.6554,
+      "step": 4267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8838058350872347e-05,
+      "loss": 0.6914,
+      "step": 4268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8836493585720148e-05,
+      "loss": 0.6411,
+      "step": 4269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.88349285062287e-05,
+      "loss": 0.7485,
+      "step": 4270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8833363112469957e-05,
+      "loss": 0.7623,
+      "step": 4271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.88317974045159e-05,
+      "loss": 0.7573,
+      "step": 4272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.883023138243852e-05,
+      "loss": 0.7468,
+      "step": 4273
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8828665046309818e-05,
+      "loss": 0.7081,
+      "step": 4274
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8827098396201812e-05,
+      "loss": 0.6549,
+      "step": 4275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8825531432186546e-05,
+      "loss": 0.6214,
+      "step": 4276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8823964154336057e-05,
+      "loss": 0.7579,
+      "step": 4277
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8822396562722412e-05,
+      "loss": 0.6848,
+      "step": 4278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.882082865741769e-05,
+      "loss": 0.7659,
+      "step": 4279
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8819260438493983e-05,
+      "loss": 0.6925,
+      "step": 4280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8817691906023395e-05,
+      "loss": 0.7259,
+      "step": 4281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8816123060078045e-05,
+      "loss": 0.7019,
+      "step": 4282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8814553900730075e-05,
+      "loss": 0.738,
+      "step": 4283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8812984428051624e-05,
+      "loss": 0.7569,
+      "step": 4284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8811414642114867e-05,
+      "loss": 0.7721,
+      "step": 4285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.880984454299197e-05,
+      "loss": 0.6658,
+      "step": 4286
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8808274130755134e-05,
+      "loss": 0.7056,
+      "step": 4287
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.880670340547656e-05,
+      "loss": 0.6903,
+      "step": 4288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8805132367228483e-05,
+      "loss": 0.6485,
+      "step": 4289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.880356101608312e-05,
+      "loss": 0.6879,
+      "step": 4290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.880198935211273e-05,
+      "loss": 0.6968,
+      "step": 4291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.880041737538958e-05,
+      "loss": 0.6903,
+      "step": 4292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8798845085985947e-05,
+      "loss": 0.7555,
+      "step": 4293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8797272483974118e-05,
+      "loss": 0.7191,
+      "step": 4294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8795699569426414e-05,
+      "loss": 0.6249,
+      "step": 4295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8794126342415144e-05,
+      "loss": 0.6899,
+      "step": 4296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.879255280301265e-05,
+      "loss": 0.605,
+      "step": 4297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8790978951291285e-05,
+      "loss": 0.7614,
+      "step": 4298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8789404787323407e-05,
+      "loss": 0.7224,
+      "step": 4299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.87878303111814e-05,
+      "loss": 0.7639,
+      "step": 4300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.878625552293766e-05,
+      "loss": 0.6535,
+      "step": 4301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.878468042266459e-05,
+      "loss": 0.725,
+      "step": 4302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8783105010434613e-05,
+      "loss": 0.8151,
+      "step": 4303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.878152928632017e-05,
+      "loss": 0.6737,
+      "step": 4304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8779953250393715e-05,
+      "loss": 0.6947,
+      "step": 4305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8778376902727702e-05,
+      "loss": 0.7402,
+      "step": 4306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.877680024339462e-05,
+      "loss": 0.7009,
+      "step": 4307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.877522327246696e-05,
+      "loss": 0.6873,
+      "step": 4308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8773645990017227e-05,
+      "loss": 0.699,
+      "step": 4309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8772068396117955e-05,
+      "loss": 0.709,
+      "step": 4310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8770490490841666e-05,
+      "loss": 0.794,
+      "step": 4311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.876891227426093e-05,
+      "loss": 0.705,
+      "step": 4312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8767333746448292e-05,
+      "loss": 0.6806,
+      "step": 4313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8765754907476344e-05,
+      "loss": 0.6892,
+      "step": 4314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8764175757417682e-05,
+      "loss": 0.696,
+      "step": 4315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.876259629634491e-05,
+      "loss": 0.6796,
+      "step": 4316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8761016524330653e-05,
+      "loss": 0.7625,
+      "step": 4317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8759436441447545e-05,
+      "loss": 0.7227,
+      "step": 4318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8757856047768242e-05,
+      "loss": 0.6947,
+      "step": 4319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.875627534336541e-05,
+      "loss": 0.7654,
+      "step": 4320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8754694328311728e-05,
+      "loss": 0.7766,
+      "step": 4321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8753113002679884e-05,
+      "loss": 0.7143,
+      "step": 4322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8751531366542595e-05,
+      "loss": 0.6809,
+      "step": 4323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.874994941997258e-05,
+      "loss": 0.7403,
+      "step": 4324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8748367163042578e-05,
+      "loss": 0.754,
+      "step": 4325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.874678459582534e-05,
+      "loss": 0.708,
+      "step": 4326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8745201718393638e-05,
+      "loss": 0.6184,
+      "step": 4327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.874361853082024e-05,
+      "loss": 0.7411,
+      "step": 4328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8742035033177945e-05,
+      "loss": 0.6969,
+      "step": 4329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8740451225539565e-05,
+      "loss": 0.6913,
+      "step": 4330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8738867107977925e-05,
+      "loss": 0.7151,
+      "step": 4331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8737282680565855e-05,
+      "loss": 0.7059,
+      "step": 4332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.873569794337621e-05,
+      "loss": 0.7672,
+      "step": 4333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.873411289648185e-05,
+      "loss": 0.673,
+      "step": 4334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8732527539955665e-05,
+      "loss": 0.7179,
+      "step": 4335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8730941873870542e-05,
+      "loss": 0.7443,
+      "step": 4336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.872935589829939e-05,
+      "loss": 0.6932,
+      "step": 4337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8727769613315133e-05,
+      "loss": 0.6798,
+      "step": 4338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.872618301899071e-05,
+      "loss": 0.7052,
+      "step": 4339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8724596115399067e-05,
+      "loss": 0.6977,
+      "step": 4340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.872300890261317e-05,
+      "loss": 0.6917,
+      "step": 4341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8721421380706e-05,
+      "loss": 0.6962,
+      "step": 4342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8719833549750553e-05,
+      "loss": 0.6686,
+      "step": 4343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8718245409819835e-05,
+      "loss": 0.7099,
+      "step": 4344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8716656960986864e-05,
+      "loss": 0.613,
+      "step": 4345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8715068203324684e-05,
+      "loss": 0.6849,
+      "step": 4346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8713479136906334e-05,
+      "loss": 0.7743,
+      "step": 4347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8711889761804895e-05,
+      "loss": 0.6885,
+      "step": 4348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8710300078093432e-05,
+      "loss": 0.7411,
+      "step": 4349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.870871008584504e-05,
+      "loss": 0.6632,
+      "step": 4350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8707119785132834e-05,
+      "loss": 0.7079,
+      "step": 4351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8705529176029925e-05,
+      "loss": 0.7135,
+      "step": 4352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8703938258609458e-05,
+      "loss": 0.7207,
+      "step": 4353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8702347032944575e-05,
+      "loss": 0.7098,
+      "step": 4354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8700755499108444e-05,
+      "loss": 0.718,
+      "step": 4355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8699163657174243e-05,
+      "loss": 0.7045,
+      "step": 4356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8697571507215166e-05,
+      "loss": 0.7113,
+      "step": 4357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8695979049304413e-05,
+      "loss": 0.7475,
+      "step": 4358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8694386283515216e-05,
+      "loss": 0.6481,
+      "step": 4359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8692793209920795e-05,
+      "loss": 0.6521,
+      "step": 4360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.869119982859441e-05,
+      "loss": 0.6734,
+      "step": 4361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.868960613960932e-05,
+      "loss": 0.6897,
+      "step": 4362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8688012143038804e-05,
+      "loss": 0.7808,
+      "step": 4363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.868641783895615e-05,
+      "loss": 0.6983,
+      "step": 4364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8684823227434666e-05,
+      "loss": 0.6653,
+      "step": 4365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8683228308547673e-05,
+      "loss": 0.7116,
+      "step": 4366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.86816330823685e-05,
+      "loss": 0.7667,
+      "step": 4367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8680037548970498e-05,
+      "loss": 0.7226,
+      "step": 4368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8678441708427026e-05,
+      "loss": 0.7027,
+      "step": 4369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.867684556081147e-05,
+      "loss": 0.7464,
+      "step": 4370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8675249106197204e-05,
+      "loss": 0.674,
+      "step": 4371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8673652344657648e-05,
+      "loss": 0.6705,
+      "step": 4372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.867205527626621e-05,
+      "loss": 0.7162,
+      "step": 4373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.867045790109633e-05,
+      "loss": 0.68,
+      "step": 4374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8668860219221448e-05,
+      "loss": 0.6126,
+      "step": 4375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.866726223071503e-05,
+      "loss": 0.6911,
+      "step": 4376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8665663935650546e-05,
+      "loss": 0.66,
+      "step": 4377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.866406533410149e-05,
+      "loss": 0.7267,
+      "step": 4378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8662466426141364e-05,
+      "loss": 0.691,
+      "step": 4379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.866086721184368e-05,
+      "loss": 0.7185,
+      "step": 4380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8659267691281975e-05,
+      "loss": 0.6199,
+      "step": 4381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8657667864529794e-05,
+      "loss": 0.6604,
+      "step": 4382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8656067731660692e-05,
+      "loss": 0.7402,
+      "step": 4383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8654467292748246e-05,
+      "loss": 0.7386,
+      "step": 4384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8652866547866043e-05,
+      "loss": 0.6966,
+      "step": 4385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8651265497087685e-05,
+      "loss": 0.7311,
+      "step": 4386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8649664140486788e-05,
+      "loss": 0.7717,
+      "step": 4387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8648062478136978e-05,
+      "loss": 0.6548,
+      "step": 4388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8646460510111903e-05,
+      "loss": 0.6973,
+      "step": 4389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.864485823648522e-05,
+      "loss": 0.7434,
+      "step": 4390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8643255657330603e-05,
+      "loss": 0.6907,
+      "step": 4391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.864165277272173e-05,
+      "loss": 0.7281,
+      "step": 4392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.864004958273231e-05,
+      "loss": 0.7082,
+      "step": 4393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8638446087436054e-05,
+      "loss": 0.7629,
+      "step": 4394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8636842286906687e-05,
+      "loss": 0.6491,
+      "step": 4395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8635238181217957e-05,
+      "loss": 0.6782,
+      "step": 4396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8633633770443612e-05,
+      "loss": 0.7482,
+      "step": 4397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.863202905465743e-05,
+      "loss": 0.7233,
+      "step": 4398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.863042403393319e-05,
+      "loss": 0.6884,
+      "step": 4399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.862881870834469e-05,
+      "loss": 0.781,
+      "step": 4400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.862721307796575e-05,
+      "loss": 0.7126,
+      "step": 4401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.862560714287019e-05,
+      "loss": 0.7512,
+      "step": 4402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.862400090313185e-05,
+      "loss": 0.6934,
+      "step": 4403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8622394358824586e-05,
+      "loss": 0.7422,
+      "step": 4404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8620787510022265e-05,
+      "loss": 0.7609,
+      "step": 4405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861918035679877e-05,
+      "loss": 0.7138,
+      "step": 4406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8617572899227994e-05,
+      "loss": 0.7522,
+      "step": 4407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8615965137383855e-05,
+      "loss": 0.8009,
+      "step": 4408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861435707134027e-05,
+      "loss": 0.6834,
+      "step": 4409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861274870117118e-05,
+      "loss": 0.7306,
+      "step": 4410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8611140026950537e-05,
+      "loss": 0.7474,
+      "step": 4411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.860953104875231e-05,
+      "loss": 0.7083,
+      "step": 4412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8607921766650472e-05,
+      "loss": 0.7473,
+      "step": 4413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8606312180719024e-05,
+      "loss": 0.7219,
+      "step": 4414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8604702291031973e-05,
+      "loss": 0.7643,
+      "step": 4415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8603092097663335e-05,
+      "loss": 0.7086,
+      "step": 4416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.860148160068716e-05,
+      "loss": 0.7973,
+      "step": 4417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.859987080017748e-05,
+      "loss": 0.6394,
+      "step": 4418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8598259696208372e-05,
+      "loss": 0.7846,
+      "step": 4419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8596648288853907e-05,
+      "loss": 0.7602,
+      "step": 4420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8595036578188182e-05,
+      "loss": 0.7126,
+      "step": 4421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.85934245642853e-05,
+      "loss": 0.678,
+      "step": 4422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.859181224721938e-05,
+      "loss": 0.7124,
+      "step": 4423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.859019962706456e-05,
+      "loss": 0.7097,
+      "step": 4424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8588586703894983e-05,
+      "loss": 0.6479,
+      "step": 4425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.858697347778481e-05,
+      "loss": 0.6918,
+      "step": 4426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.858535994880822e-05,
+      "loss": 0.7649,
+      "step": 4427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.85837461170394e-05,
+      "loss": 0.7365,
+      "step": 4428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8582131982552555e-05,
+      "loss": 0.7283,
+      "step": 4429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8580517545421902e-05,
+      "loss": 0.6789,
+      "step": 4430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.857890280572167e-05,
+      "loss": 0.6809,
+      "step": 4431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.857728776352611e-05,
+      "loss": 0.656,
+      "step": 4432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8575672418909474e-05,
+      "loss": 0.6972,
+      "step": 4433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8574056771946036e-05,
+      "loss": 0.7276,
+      "step": 4434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.857244082271009e-05,
+      "loss": 0.7458,
+      "step": 4435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8570824571275927e-05,
+      "loss": 0.7576,
+      "step": 4436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8569208017717866e-05,
+      "loss": 0.7426,
+      "step": 4437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8567591162110234e-05,
+      "loss": 0.7197,
+      "step": 4438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.856597400452738e-05,
+      "loss": 0.7052,
+      "step": 4439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8564356545043653e-05,
+      "loss": 0.7579,
+      "step": 4440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.856273878373342e-05,
+      "loss": 0.7246,
+      "step": 4441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8561120720671077e-05,
+      "loss": 0.7527,
+      "step": 4442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.855950235593101e-05,
+      "loss": 0.6628,
+      "step": 4443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.855788368958764e-05,
+      "loss": 0.6701,
+      "step": 4444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8556264721715384e-05,
+      "loss": 0.64,
+      "step": 4445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8554645452388693e-05,
+      "loss": 0.695,
+      "step": 4446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8553025881682004e-05,
+      "loss": 0.7186,
+      "step": 4447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.85514060096698e-05,
+      "loss": 0.71,
+      "step": 4448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8549785836426555e-05,
+      "loss": 0.7121,
+      "step": 4449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8548165362026765e-05,
+      "loss": 0.6469,
+      "step": 4450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8546544586544935e-05,
+      "loss": 0.7442,
+      "step": 4451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8544923510055593e-05,
+      "loss": 0.6747,
+      "step": 4452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8543302132633272e-05,
+      "loss": 0.7171,
+      "step": 4453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8541680454352526e-05,
+      "loss": 0.6634,
+      "step": 4454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8540058475287914e-05,
+      "loss": 0.7342,
+      "step": 4455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8538436195514016e-05,
+      "loss": 0.654,
+      "step": 4456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.853681361510543e-05,
+      "loss": 0.7131,
+      "step": 4457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.853519073413675e-05,
+      "loss": 0.6099,
+      "step": 4458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8533567552682606e-05,
+      "loss": 0.6913,
+      "step": 4459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8531944070817628e-05,
+      "loss": 0.6973,
+      "step": 4460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8530320288616457e-05,
+      "loss": 0.6344,
+      "step": 4461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.852869620615376e-05,
+      "loss": 0.6611,
+      "step": 4462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8527071823504213e-05,
+      "loss": 0.6587,
+      "step": 4463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8525447140742496e-05,
+      "loss": 0.7138,
+      "step": 4464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8523822157943324e-05,
+      "loss": 0.7283,
+      "step": 4465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.85221968751814e-05,
+      "loss": 0.6841,
+      "step": 4466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8520571292531462e-05,
+      "loss": 0.7028,
+      "step": 4467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8518945410068253e-05,
+      "loss": 0.7183,
+      "step": 4468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.851731922786653e-05,
+      "loss": 0.7422,
+      "step": 4469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8515692746001064e-05,
+      "loss": 0.7493,
+      "step": 4470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8514065964546633e-05,
+      "loss": 0.7462,
+      "step": 4471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8512438883578048e-05,
+      "loss": 0.6697,
+      "step": 4472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8510811503170113e-05,
+      "loss": 0.6766,
+      "step": 4473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8509183823397657e-05,
+      "loss": 0.7608,
+      "step": 4474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.850755584433552e-05,
+      "loss": 0.7479,
+      "step": 4475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8505927566058555e-05,
+      "loss": 0.6991,
+      "step": 4476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8504298988641633e-05,
+      "loss": 0.6427,
+      "step": 4477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.850267011215963e-05,
+      "loss": 0.7024,
+      "step": 4478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8501040936687445e-05,
+      "loss": 0.6354,
+      "step": 4479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8499411462299986e-05,
+      "loss": 0.6867,
+      "step": 4480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8497781689072174e-05,
+      "loss": 0.7649,
+      "step": 4481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.849615161707895e-05,
+      "loss": 0.6655,
+      "step": 4482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8494521246395255e-05,
+      "loss": 0.6992,
+      "step": 4483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8492890577096062e-05,
+      "loss": 0.7234,
+      "step": 4484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8491259609256343e-05,
+      "loss": 0.6736,
+      "step": 4485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8489628342951094e-05,
+      "loss": 0.717,
+      "step": 4486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8487996778255314e-05,
+      "loss": 0.7982,
+      "step": 4487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8486364915244024e-05,
+      "loss": 0.6915,
+      "step": 4488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.848473275399226e-05,
+      "loss": 0.6833,
+      "step": 4489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8483100294575063e-05,
+      "loss": 0.7432,
+      "step": 4490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8481467537067497e-05,
+      "loss": 0.7258,
+      "step": 4491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8479834481544636e-05,
+      "loss": 0.6882,
+      "step": 4492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.847820112808156e-05,
+      "loss": 0.7656,
+      "step": 4493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.847656747675338e-05,
+      "loss": 0.7152,
+      "step": 4494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8474933527635202e-05,
+      "loss": 0.6685,
+      "step": 4495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8473299280802158e-05,
+      "loss": 0.7132,
+      "step": 4496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8471664736329393e-05,
+      "loss": 0.6863,
+      "step": 4497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8470029894292057e-05,
+      "loss": 0.7227,
+      "step": 4498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.846839475476532e-05,
+      "loss": 0.6939,
+      "step": 4499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8466759317824372e-05,
+      "loss": 0.6917,
+      "step": 4500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8465123583544403e-05,
+      "loss": 0.6574,
+      "step": 4501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8463487552000626e-05,
+      "loss": 0.683,
+      "step": 4502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.846185122326826e-05,
+      "loss": 0.6284,
+      "step": 4503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8460214597422548e-05,
+      "loss": 0.6781,
+      "step": 4504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8458577674538743e-05,
+      "loss": 0.6677,
+      "step": 4505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8456940454692106e-05,
+      "loss": 0.7337,
+      "step": 4506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8455302937957917e-05,
+      "loss": 0.7042,
+      "step": 4507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.845366512441147e-05,
+      "loss": 0.6979,
+      "step": 4508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8452027014128072e-05,
+      "loss": 0.6489,
+      "step": 4509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8450388607183035e-05,
+      "loss": 0.7344,
+      "step": 4510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84487499036517e-05,
+      "loss": 0.7843,
+      "step": 4511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.844711090360941e-05,
+      "loss": 0.6864,
+      "step": 4512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.844547160713153e-05,
+      "loss": 0.6896,
+      "step": 4513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8443832014293432e-05,
+      "loss": 0.7094,
+      "step": 4514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84421921251705e-05,
+      "loss": 0.6713,
+      "step": 4515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.844055193983814e-05,
+      "loss": 0.6664,
+      "step": 4516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8438911458371765e-05,
+      "loss": 0.7545,
+      "step": 4517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8437270680846806e-05,
+      "loss": 0.7319,
+      "step": 4518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84356296073387e-05,
+      "loss": 0.754,
+      "step": 4519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.843398823792291e-05,
+      "loss": 0.803,
+      "step": 4520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84323465726749e-05,
+      "loss": 0.7454,
+      "step": 4521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8430704611670152e-05,
+      "loss": 0.7274,
+      "step": 4522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.842906235498417e-05,
+      "loss": 0.6011,
+      "step": 4523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.842741980269246e-05,
+      "loss": 0.7082,
+      "step": 4524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8425776954870545e-05,
+      "loss": 0.689,
+      "step": 4525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8424133811593965e-05,
+      "loss": 0.6203,
+      "step": 4526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.842249037293827e-05,
+      "loss": 0.6915,
+      "step": 4527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.842084663897902e-05,
+      "loss": 0.7153,
+      "step": 4528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8419202609791802e-05,
+      "loss": 0.7264,
+      "step": 4529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84175582854522e-05,
+      "loss": 0.7142,
+      "step": 4530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8415913666035823e-05,
+      "loss": 0.7271,
+      "step": 4531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8414268751618294e-05,
+      "loss": 0.7016,
+      "step": 4532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8412623542275236e-05,
+      "loss": 0.714,
+      "step": 4533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8410978038082303e-05,
+      "loss": 0.7418,
+      "step": 4534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8409332239115152e-05,
+      "loss": 0.6568,
+      "step": 4535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8407686145449457e-05,
+      "loss": 0.7211,
+      "step": 4536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84060397571609e-05,
+      "loss": 0.728,
+      "step": 4537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8404393074325187e-05,
+      "loss": 0.6991,
+      "step": 4538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.840274609701803e-05,
+      "loss": 0.6956,
+      "step": 4539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8401098825315158e-05,
+      "loss": 0.7869,
+      "step": 4540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8399451259292308e-05,
+      "loss": 0.6921,
+      "step": 4541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8397803399025238e-05,
+      "loss": 0.6979,
+      "step": 4542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.839615524458971e-05,
+      "loss": 0.6298,
+      "step": 4543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.839450679606151e-05,
+      "loss": 0.7293,
+      "step": 4544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.839285805351644e-05,
+      "loss": 0.6472,
+      "step": 4545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8391209017030298e-05,
+      "loss": 0.8027,
+      "step": 4546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8389559686678906e-05,
+      "loss": 0.7372,
+      "step": 4547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.838791006253811e-05,
+      "loss": 0.7143,
+      "step": 4548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8386260144683745e-05,
+      "loss": 0.6929,
+      "step": 4549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8384609933191687e-05,
+      "loss": 0.6916,
+      "step": 4550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8382959428137805e-05,
+      "loss": 0.6452,
+      "step": 4551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8381308629597985e-05,
+      "loss": 0.712,
+      "step": 4552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8379657537648138e-05,
+      "loss": 0.762,
+      "step": 4553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8378006152364177e-05,
+      "loss": 0.7264,
+      "step": 4554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8376354473822028e-05,
+      "loss": 0.7793,
+      "step": 4555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8374702502097642e-05,
+      "loss": 0.744,
+      "step": 4556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.837305023726697e-05,
+      "loss": 0.7996,
+      "step": 4557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8371397679405988e-05,
+      "loss": 0.7152,
+      "step": 4558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8369744828590677e-05,
+      "loss": 0.6838,
+      "step": 4559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8368091684897034e-05,
+      "loss": 0.6521,
+      "step": 4560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8366438248401067e-05,
+      "loss": 0.695,
+      "step": 4561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8364784519178806e-05,
+      "loss": 0.7211,
+      "step": 4562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8363130497306287e-05,
+      "loss": 0.6496,
+      "step": 4563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.836147618285956e-05,
+      "loss": 0.698,
+      "step": 4564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.835982157591469e-05,
+      "loss": 0.607,
+      "step": 4565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.835816667654776e-05,
+      "loss": 0.7223,
+      "step": 4566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8356511484834852e-05,
+      "loss": 0.7493,
+      "step": 4567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8354856000852077e-05,
+      "loss": 0.6324,
+      "step": 4568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8353200224675553e-05,
+      "loss": 0.719,
+      "step": 4569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8351544156381414e-05,
+      "loss": 0.631,
+      "step": 4570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8349887796045802e-05,
+      "loss": 0.6541,
+      "step": 4571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8348231143744872e-05,
+      "loss": 0.7162,
+      "step": 4572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8346574199554807e-05,
+      "loss": 0.7185,
+      "step": 4573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8344916963551787e-05,
+      "loss": 0.7127,
+      "step": 4574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8343259435812006e-05,
+      "loss": 0.6854,
+      "step": 4575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8341601616411687e-05,
+      "loss": 0.71,
+      "step": 4576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8339943505427044e-05,
+      "loss": 0.7153,
+      "step": 4577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8338285102934326e-05,
+      "loss": 0.7272,
+      "step": 4578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8336626409009782e-05,
+      "loss": 0.7544,
+      "step": 4579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8334967423729678e-05,
+      "loss": 0.6609,
+      "step": 4580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8333308147170292e-05,
+      "loss": 0.7198,
+      "step": 4581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.833164857940792e-05,
+      "loss": 0.69,
+      "step": 4582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8329988720518866e-05,
+      "loss": 0.6795,
+      "step": 4583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.832832857057945e-05,
+      "loss": 0.7761,
+      "step": 4584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.832666812966601e-05,
+      "loss": 0.6868,
+      "step": 4585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.832500739785488e-05,
+      "loss": 0.7166,
+      "step": 4586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8323346375222433e-05,
+      "loss": 0.7072,
+      "step": 4587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8321685061845037e-05,
+      "loss": 0.6805,
+      "step": 4588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8320023457799075e-05,
+      "loss": 0.7343,
+      "step": 4589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8318361563160952e-05,
+      "loss": 0.678,
+      "step": 4590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.831669937800708e-05,
+      "loss": 0.748,
+      "step": 4591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8315036902413888e-05,
+      "loss": 0.7169,
+      "step": 4592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.831337413645781e-05,
+      "loss": 0.6612,
+      "step": 4593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.83117110802153e-05,
+      "loss": 0.7073,
+      "step": 4594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.831004773376283e-05,
+      "loss": 0.7194,
+      "step": 4595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8308384097176873e-05,
+      "loss": 0.6984,
+      "step": 4596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8306720170533933e-05,
+      "loss": 0.6216,
+      "step": 4597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8305055953910506e-05,
+      "loss": 0.6606,
+      "step": 4598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8303391447383115e-05,
+      "loss": 0.7187,
+      "step": 4599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.83017266510283e-05,
+      "loss": 0.7066,
+      "step": 4600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8300061564922598e-05,
+      "loss": 0.7426,
+      "step": 4601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8298396189142572e-05,
+      "loss": 0.7819,
+      "step": 4602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8296730523764797e-05,
+      "loss": 0.628,
+      "step": 4603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8295064568865862e-05,
+      "loss": 0.7148,
+      "step": 4604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8293398324522362e-05,
+      "loss": 0.7379,
+      "step": 4605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8291731790810913e-05,
+      "loss": 0.8157,
+      "step": 4606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.829006496780814e-05,
+      "loss": 0.728,
+      "step": 4607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8288397855590685e-05,
+      "loss": 0.6943,
+      "step": 4608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8286730454235195e-05,
+      "loss": 0.75,
+      "step": 4609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.828506276381835e-05,
+      "loss": 0.7143,
+      "step": 4610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.828339478441681e-05,
+      "loss": 0.6755,
+      "step": 4611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8281726516107288e-05,
+      "loss": 0.6718,
+      "step": 4612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8280057958966478e-05,
+      "loss": 0.7299,
+      "step": 4613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8278389113071103e-05,
+      "loss": 0.7718,
+      "step": 4614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8276719978497895e-05,
+      "loss": 0.6726,
+      "step": 4615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8275050555323606e-05,
+      "loss": 0.6651,
+      "step": 4616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8273380843624984e-05,
+      "loss": 0.7253,
+      "step": 4617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.827171084347881e-05,
+      "loss": 0.6991,
+      "step": 4618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8270040554961868e-05,
+      "loss": 0.7609,
+      "step": 4619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.826836997815096e-05,
+      "loss": 0.7496,
+      "step": 4620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8266699113122893e-05,
+      "loss": 0.691,
+      "step": 4621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8265027959954497e-05,
+      "loss": 0.6949,
+      "step": 4622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8263356518722606e-05,
+      "loss": 0.7344,
+      "step": 4623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.826168478950408e-05,
+      "loss": 0.7521,
+      "step": 4624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8260012772375776e-05,
+      "loss": 0.7171,
+      "step": 4625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.825834046741458e-05,
+      "loss": 0.5883,
+      "step": 4626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.825666787469738e-05,
+      "loss": 0.7165,
+      "step": 4627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8254994994301078e-05,
+      "loss": 0.7256,
+      "step": 4628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.82533218263026e-05,
+      "loss": 0.7396,
+      "step": 4629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8251648370778872e-05,
+      "loss": 0.6735,
+      "step": 4630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.824997462780684e-05,
+      "loss": 0.6518,
+      "step": 4631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8248300597463464e-05,
+      "loss": 0.6842,
+      "step": 4632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8246626279825713e-05,
+      "loss": 0.7306,
+      "step": 4633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8244951674970572e-05,
+      "loss": 0.6345,
+      "step": 4634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8243276782975037e-05,
+      "loss": 0.6735,
+      "step": 4635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8241601603916125e-05,
+      "loss": 0.7369,
+      "step": 4636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8239926137870854e-05,
+      "loss": 0.6716,
+      "step": 4637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8238250384916265e-05,
+      "loss": 0.713,
+      "step": 4638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.82365743451294e-05,
+      "loss": 0.7843,
+      "step": 4639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8234898018587337e-05,
+      "loss": 0.6615,
+      "step": 4640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8233221405367144e-05,
+      "loss": 0.6832,
+      "step": 4641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.823154450554591e-05,
+      "loss": 0.6703,
+      "step": 4642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.822986731920074e-05,
+      "loss": 0.7332,
+      "step": 4643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.822818984640875e-05,
+      "loss": 0.7484,
+      "step": 4644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8226512087247072e-05,
+      "loss": 0.7325,
+      "step": 4645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8224834041792845e-05,
+      "loss": 0.7253,
+      "step": 4646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8223155710123226e-05,
+      "loss": 0.7154,
+      "step": 4647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.822147709231539e-05,
+      "loss": 0.7407,
+      "step": 4648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8219798188446507e-05,
+      "loss": 0.7519,
+      "step": 4649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8218118998593786e-05,
+      "loss": 0.6439,
+      "step": 4650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8216439522834425e-05,
+      "loss": 0.7376,
+      "step": 4651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.821475976124565e-05,
+      "loss": 0.7342,
+      "step": 4652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.821307971390469e-05,
+      "loss": 0.6946,
+      "step": 4653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8211399380888806e-05,
+      "loss": 0.696,
+      "step": 4654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8209718762275245e-05,
+      "loss": 0.736,
+      "step": 4655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.820803785814129e-05,
+      "loss": 0.7361,
+      "step": 4656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8206356668564225e-05,
+      "loss": 0.6834,
+      "step": 4657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.820467519362135e-05,
+      "loss": 0.7157,
+      "step": 4658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8202993433389984e-05,
+      "loss": 0.7456,
+      "step": 4659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8201311387947444e-05,
+      "loss": 0.604,
+      "step": 4660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.819962905737108e-05,
+      "loss": 0.663,
+      "step": 4661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8197946441738233e-05,
+      "loss": 0.6775,
+      "step": 4662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8196263541126283e-05,
+      "loss": 0.7228,
+      "step": 4663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8194580355612592e-05,
+      "loss": 0.6896,
+      "step": 4664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8192896885274573e-05,
+      "loss": 0.6728,
+      "step": 4665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8191213130189616e-05,
+      "loss": 0.7056,
+      "step": 4666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.818952909043514e-05,
+      "loss": 0.6782,
+      "step": 4667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8187844766088587e-05,
+      "loss": 0.6721,
+      "step": 4668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8186160157227393e-05,
+      "loss": 0.6979,
+      "step": 4669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8184475263929015e-05,
+      "loss": 0.7414,
+      "step": 4670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.818279008627093e-05,
+      "loss": 0.6884,
+      "step": 4671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8181104624330617e-05,
+      "loss": 0.7395,
+      "step": 4672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.817941887818557e-05,
+      "loss": 0.7059,
+      "step": 4673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8177732847913304e-05,
+      "loss": 0.7529,
+      "step": 4674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8176046533591342e-05,
+      "loss": 0.7887,
+      "step": 4675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.817435993529722e-05,
+      "loss": 0.7337,
+      "step": 4676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8172673053108484e-05,
+      "loss": 0.732,
+      "step": 4677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.81709858871027e-05,
+      "loss": 0.6967,
+      "step": 4678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.816929843735744e-05,
+      "loss": 0.7195,
+      "step": 4679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8167610703950295e-05,
+      "loss": 0.6536,
+      "step": 4680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8165922686958864e-05,
+      "loss": 0.7391,
+      "step": 4681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8164234386460762e-05,
+      "loss": 0.7919,
+      "step": 4682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8162545802533618e-05,
+      "loss": 0.7324,
+      "step": 4683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.816085693525507e-05,
+      "loss": 0.6368,
+      "step": 4684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8159167784702778e-05,
+      "loss": 0.7202,
+      "step": 4685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8157478350954397e-05,
+      "loss": 0.6756,
+      "step": 4686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8155788634087614e-05,
+      "loss": 0.7122,
+      "step": 4687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8154098634180118e-05,
+      "loss": 0.7342,
+      "step": 4688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.815240835130962e-05,
+      "loss": 0.6666,
+      "step": 4689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.815071778555383e-05,
+      "loss": 0.7077,
+      "step": 4690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.814902693699049e-05,
+      "loss": 0.7744,
+      "step": 4691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8147335805697337e-05,
+      "loss": 0.6794,
+      "step": 4692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.814564439175213e-05,
+      "loss": 0.6693,
+      "step": 4693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8143952695232642e-05,
+      "loss": 0.683,
+      "step": 4694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8142260716216655e-05,
+      "loss": 0.6178,
+      "step": 4695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8140568454781964e-05,
+      "loss": 0.7275,
+      "step": 4696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.813887591100638e-05,
+      "loss": 0.715,
+      "step": 4697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8137183084967725e-05,
+      "loss": 0.7304,
+      "step": 4698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8135489976743835e-05,
+      "loss": 0.6485,
+      "step": 4699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8133796586412563e-05,
+      "loss": 0.6844,
+      "step": 4700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8132102914051757e-05,
+      "loss": 0.7373,
+      "step": 4701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8130408959739307e-05,
+      "loss": 0.7103,
+      "step": 4702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.812871472355309e-05,
+      "loss": 0.7897,
+      "step": 4703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.812702020557101e-05,
+      "loss": 0.7329,
+      "step": 4704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8125325405870977e-05,
+      "loss": 0.7011,
+      "step": 4705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8123630324530925e-05,
+      "loss": 0.6828,
+      "step": 4706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8121934961628787e-05,
+      "loss": 0.6322,
+      "step": 4707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8120239317242516e-05,
+      "loss": 0.6818,
+      "step": 4708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8118543391450077e-05,
+      "loss": 0.7533,
+      "step": 4709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.811684718432945e-05,
+      "loss": 0.6245,
+      "step": 4710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8115150695958626e-05,
+      "loss": 0.684,
+      "step": 4711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8113453926415605e-05,
+      "loss": 0.74,
+      "step": 4712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8111756875778404e-05,
+      "loss": 0.6769,
+      "step": 4713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8110059544125057e-05,
+      "loss": 0.7295,
+      "step": 4714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8108361931533602e-05,
+      "loss": 0.6849,
+      "step": 4715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.81066640380821e-05,
+      "loss": 0.6648,
+      "step": 4716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8104965863848616e-05,
+      "loss": 0.7135,
+      "step": 4717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.810326740891123e-05,
+      "loss": 0.6055,
+      "step": 4718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.810156867334804e-05,
+      "loss": 0.7146,
+      "step": 4719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.809986965723715e-05,
+      "loss": 0.6976,
+      "step": 4720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.809817036065668e-05,
+      "loss": 0.6937,
+      "step": 4721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8096470783684766e-05,
+      "loss": 0.7386,
+      "step": 4722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8094770926399554e-05,
+      "loss": 0.6651,
+      "step": 4723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.80930707888792e-05,
+      "loss": 0.7602,
+      "step": 4724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8091370371201878e-05,
+      "loss": 0.6863,
+      "step": 4725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8089669673445768e-05,
+      "loss": 0.6272,
+      "step": 4726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8087968695689074e-05,
+      "loss": 0.7142,
+      "step": 4727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.808626743801e-05,
+      "loss": 0.7193,
+      "step": 4728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8084565900486768e-05,
+      "loss": 0.7377,
+      "step": 4729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8082864083197623e-05,
+      "loss": 0.7618,
+      "step": 4730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.808116198622081e-05,
+      "loss": 0.7068,
+      "step": 4731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.807945960963459e-05,
+      "loss": 0.778,
+      "step": 4732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.807775695351723e-05,
+      "loss": 0.7273,
+      "step": 4733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.807605401794703e-05,
+      "loss": 0.8132,
+      "step": 4734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.807435080300229e-05,
+      "loss": 0.7362,
+      "step": 4735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.807264730876131e-05,
+      "loss": 0.6221,
+      "step": 4736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8070943535302428e-05,
+      "loss": 0.6734,
+      "step": 4737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8069239482703978e-05,
+      "loss": 0.7157,
+      "step": 4738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.806753515104431e-05,
+      "loss": 0.7339,
+      "step": 4739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8065830540401794e-05,
+      "loss": 0.7156,
+      "step": 4740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8064125650854805e-05,
+      "loss": 0.693,
+      "step": 4741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.806242048248173e-05,
+      "loss": 0.7185,
+      "step": 4742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8060715035360975e-05,
+      "loss": 0.6652,
+      "step": 4743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8059009309570958e-05,
+      "loss": 0.7282,
+      "step": 4744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.80573033051901e-05,
+      "loss": 0.6931,
+      "step": 4745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8055597022296854e-05,
+      "loss": 0.7961,
+      "step": 4746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.805389046096966e-05,
+      "loss": 0.6632,
+      "step": 4747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8052183621286998e-05,
+      "loss": 0.6928,
+      "step": 4748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.805047650332734e-05,
+      "loss": 0.7149,
+      "step": 4749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.804876910716918e-05,
+      "loss": 0.7321,
+      "step": 4750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8047061432891026e-05,
+      "loss": 0.652,
+      "step": 4751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8045353480571396e-05,
+      "loss": 0.7256,
+      "step": 4752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8043645250288818e-05,
+      "loss": 0.7329,
+      "step": 4753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8041936742121838e-05,
+      "loss": 0.6592,
+      "step": 4754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.804022795614901e-05,
+      "loss": 0.7509,
+      "step": 4755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8038518892448902e-05,
+      "loss": 0.768,
+      "step": 4756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8036809551100108e-05,
+      "loss": 0.7227,
+      "step": 4757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8035099932181205e-05,
+      "loss": 0.6922,
+      "step": 4758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8033390035770813e-05,
+      "loss": 0.7148,
+      "step": 4759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.803167986194755e-05,
+      "loss": 0.651,
+      "step": 4760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8029969410790045e-05,
+      "loss": 0.7263,
+      "step": 4761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.802825868237695e-05,
+      "loss": 0.73,
+      "step": 4762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8026547676786917e-05,
+      "loss": 0.7115,
+      "step": 4763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8024836394098625e-05,
+      "loss": 0.7293,
+      "step": 4764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8023124834390753e-05,
+      "loss": 0.7297,
+      "step": 4765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8021412997741995e-05,
+      "loss": 0.6977,
+      "step": 4766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8019700884231066e-05,
+      "loss": 0.7163,
+      "step": 4767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8017988493936692e-05,
+      "loss": 0.7391,
+      "step": 4768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8016275826937596e-05,
+      "loss": 0.7019,
+      "step": 4769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8014562883312535e-05,
+      "loss": 0.7867,
+      "step": 4770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8012849663140267e-05,
+      "loss": 0.7097,
+      "step": 4771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8011136166499565e-05,
+      "loss": 0.6763,
+      "step": 4772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8009422393469214e-05,
+      "loss": 0.7064,
+      "step": 4773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8007708344128015e-05,
+      "loss": 0.8041,
+      "step": 4774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8005994018554775e-05,
+      "loss": 0.6388,
+      "step": 4775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.800427941682832e-05,
+      "loss": 0.6821,
+      "step": 4776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8002564539027493e-05,
+      "loss": 0.6659,
+      "step": 4777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8000849385231136e-05,
+      "loss": 0.7388,
+      "step": 4778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.799913395551811e-05,
+      "loss": 0.7569,
+      "step": 4779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7997418249967298e-05,
+      "loss": 0.701,
+      "step": 4780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7995702268657577e-05,
+      "loss": 0.7232,
+      "step": 4781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7993986011667854e-05,
+      "loss": 0.701,
+      "step": 4782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7992269479077043e-05,
+      "loss": 0.7232,
+      "step": 4783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7990552670964063e-05,
+      "loss": 0.7629,
+      "step": 4784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7988835587407857e-05,
+      "loss": 0.7007,
+      "step": 4785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7987118228487377e-05,
+      "loss": 0.6875,
+      "step": 4786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.798540059428158e-05,
+      "loss": 0.7019,
+      "step": 4787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7983682684869448e-05,
+      "loss": 0.6065,
+      "step": 4788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.798196450032997e-05,
+      "loss": 0.7216,
+      "step": 4789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7980246040742146e-05,
+      "loss": 0.739,
+      "step": 4790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7978527306184987e-05,
+      "loss": 0.7194,
+      "step": 4791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.797680829673752e-05,
+      "loss": 0.6291,
+      "step": 4792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.797508901247879e-05,
+      "loss": 0.7084,
+      "step": 4793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.797336945348785e-05,
+      "loss": 0.7227,
+      "step": 4794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7971649619843752e-05,
+      "loss": 0.7255,
+      "step": 4795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7969929511625588e-05,
+      "loss": 0.6938,
+      "step": 4796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7968209128912438e-05,
+      "loss": 0.659,
+      "step": 4797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7966488471783408e-05,
+      "loss": 0.7114,
+      "step": 4798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7964767540317615e-05,
+      "loss": 0.6896,
+      "step": 4799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7963046334594186e-05,
+      "loss": 0.6821,
+      "step": 4800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7961324854692256e-05,
+      "loss": 0.7242,
+      "step": 4801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7959603100690986e-05,
+      "loss": 0.8103,
+      "step": 4802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7957881072669536e-05,
+      "loss": 0.6119,
+      "step": 4803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7956158770707084e-05,
+      "loss": 0.6448,
+      "step": 4804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7954436194882827e-05,
+      "loss": 0.7093,
+      "step": 4805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7952713345275963e-05,
+      "loss": 0.7721,
+      "step": 4806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7950990221965706e-05,
+      "loss": 0.7521,
+      "step": 4807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794926682503129e-05,
+      "loss": 0.7193,
+      "step": 4808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794754315455195e-05,
+      "loss": 0.7243,
+      "step": 4809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794581921060695e-05,
+      "loss": 0.6393,
+      "step": 4810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7944094993275544e-05,
+      "loss": 0.667,
+      "step": 4811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794237050263702e-05,
+      "loss": 0.6929,
+      "step": 4812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7940645738770666e-05,
+      "loss": 0.7721,
+      "step": 4813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7938920701755782e-05,
+      "loss": 0.7934,
+      "step": 4814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7937195391671692e-05,
+      "loss": 0.6786,
+      "step": 4815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.793546980859772e-05,
+      "loss": 0.7259,
+      "step": 4816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7933743952613207e-05,
+      "loss": 0.716,
+      "step": 4817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.793201782379751e-05,
+      "loss": 0.6573,
+      "step": 4818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7930291422229995e-05,
+      "loss": 0.7414,
+      "step": 4819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7928564747990043e-05,
+      "loss": 0.7111,
+      "step": 4820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.792683780115704e-05,
+      "loss": 0.688,
+      "step": 4821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7925110581810397e-05,
+      "loss": 0.6734,
+      "step": 4822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7923383090029528e-05,
+      "loss": 0.772,
+      "step": 4823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.792165532589386e-05,
+      "loss": 0.7835,
+      "step": 4824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7919927289482842e-05,
+      "loss": 0.7301,
+      "step": 4825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.791819898087592e-05,
+      "loss": 0.7393,
+      "step": 4826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7916470400152566e-05,
+      "loss": 0.6748,
+      "step": 4827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7914741547392256e-05,
+      "loss": 0.7531,
+      "step": 4828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7913012422674488e-05,
+      "loss": 0.6418,
+      "step": 4829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.791128302607876e-05,
+      "loss": 0.7488,
+      "step": 4830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7909553357684592e-05,
+      "loss": 0.6499,
+      "step": 4831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7907823417571513e-05,
+      "loss": 0.6263,
+      "step": 4832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7906093205819068e-05,
+      "loss": 0.7171,
+      "step": 4833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7904362722506805e-05,
+      "loss": 0.7176,
+      "step": 4834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7902631967714294e-05,
+      "loss": 0.6507,
+      "step": 4835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7900900941521116e-05,
+      "loss": 0.6814,
+      "step": 4836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7899169644006863e-05,
+      "loss": 0.6884,
+      "step": 4837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7897438075251135e-05,
+      "loss": 0.7146,
+      "step": 4838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7895706235333554e-05,
+      "loss": 0.763,
+      "step": 4839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7893974124333748e-05,
+      "loss": 0.7809,
+      "step": 4840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7892241742331357e-05,
+      "loss": 0.7237,
+      "step": 4841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7890509089406034e-05,
+      "loss": 0.6836,
+      "step": 4842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.788877616563745e-05,
+      "loss": 0.7182,
+      "step": 4843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7887042971105285e-05,
+      "loss": 0.7453,
+      "step": 4844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7885309505889223e-05,
+      "loss": 0.7029,
+      "step": 4845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7883575770068973e-05,
+      "loss": 0.6997,
+      "step": 4846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7881841763724253e-05,
+      "loss": 0.7007,
+      "step": 4847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.788010748693479e-05,
+      "loss": 0.7182,
+      "step": 4848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7878372939780326e-05,
+      "loss": 0.6932,
+      "step": 4849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7876638122340616e-05,
+      "loss": 0.6232,
+      "step": 4850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.787490303469542e-05,
+      "loss": 0.6807,
+      "step": 4851
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.787316767692452e-05,
+      "loss": 0.7328,
+      "step": 4852
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7871432049107715e-05,
+      "loss": 0.7404,
+      "step": 4853
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7869696151324796e-05,
+      "loss": 0.6287,
+      "step": 4854
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7867959983655587e-05,
+      "loss": 0.6816,
+      "step": 4855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.786622354617991e-05,
+      "loss": 0.674,
+      "step": 4856
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7864486838977616e-05,
+      "loss": 0.7102,
+      "step": 4857
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7862749862128546e-05,
+      "loss": 0.7281,
+      "step": 4858
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7861012615712576e-05,
+      "loss": 0.6692,
+      "step": 4859
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7859275099809574e-05,
+      "loss": 0.7541,
+      "step": 4860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7857537314499438e-05,
+      "loss": 0.7155,
+      "step": 4861
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7855799259862068e-05,
+      "loss": 0.6133,
+      "step": 4862
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7854060935977376e-05,
+      "loss": 0.6744,
+      "step": 4863
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7852322342925296e-05,
+      "loss": 0.7332,
+      "step": 4864
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7850583480785765e-05,
+      "loss": 0.6935,
+      "step": 4865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7848844349638732e-05,
+      "loss": 0.738,
+      "step": 4866
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7847104949564166e-05,
+      "loss": 0.7516,
+      "step": 4867
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.784536528064204e-05,
+      "loss": 0.683,
+      "step": 4868
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7843625342952346e-05,
+      "loss": 0.746,
+      "step": 4869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7841885136575086e-05,
+      "loss": 0.6941,
+      "step": 4870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.784014466159027e-05,
+      "loss": 0.6844,
+      "step": 4871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7838403918077934e-05,
+      "loss": 0.7223,
+      "step": 4872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7836662906118104e-05,
+      "loss": 0.688,
+      "step": 4873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.783492162579084e-05,
+      "loss": 0.645,
+      "step": 4874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.78331800771762e-05,
+      "loss": 0.7526,
+      "step": 4875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.783143826035426e-05,
+      "loss": 0.6977,
+      "step": 4876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7829696175405113e-05,
+      "loss": 0.7323,
+      "step": 4877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7827953822408857e-05,
+      "loss": 0.6279,
+      "step": 4878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7826211201445603e-05,
+      "loss": 0.6741,
+      "step": 4879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7824468312595477e-05,
+      "loss": 0.7519,
+      "step": 4880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7822725155938616e-05,
+      "loss": 0.7099,
+      "step": 4881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7820981731555176e-05,
+      "loss": 0.7184,
+      "step": 4882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7819238039525306e-05,
+      "loss": 0.7521,
+      "step": 4883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.781749407992919e-05,
+      "loss": 0.7258,
+      "step": 4884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7815749852847016e-05,
+      "loss": 0.6476,
+      "step": 4885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7814005358358972e-05,
+      "loss": 0.7143,
+      "step": 4886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7812260596545277e-05,
+      "loss": 0.6897,
+      "step": 4887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7810515567486158e-05,
+      "loss": 0.6818,
+      "step": 4888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7808770271261843e-05,
+      "loss": 0.7806,
+      "step": 4889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7807024707952583e-05,
+      "loss": 0.7527,
+      "step": 4890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.780527887763864e-05,
+      "loss": 0.6514,
+      "step": 4891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.780353278040028e-05,
+      "loss": 0.6132,
+      "step": 4892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.78017864163178e-05,
+      "loss": 0.7418,
+      "step": 4893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.780003978547149e-05,
+      "loss": 0.6817,
+      "step": 4894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7798292887941657e-05,
+      "loss": 0.6596,
+      "step": 4895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7796545723808625e-05,
+      "loss": 0.6574,
+      "step": 4896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.779479829315273e-05,
+      "loss": 0.6964,
+      "step": 4897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.779305059605431e-05,
+      "loss": 0.672,
+      "step": 4898
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7791302632593735e-05,
+      "loss": 0.6991,
+      "step": 4899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.778955440285137e-05,
+      "loss": 0.7008,
+      "step": 4900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.77878059069076e-05,
+      "loss": 0.7426,
+      "step": 4901
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7786057144842815e-05,
+      "loss": 0.6726,
+      "step": 4902
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7784308116737427e-05,
+      "loss": 0.7632,
+      "step": 4903
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7782558822671853e-05,
+      "loss": 0.7318,
+      "step": 4904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.778080926272653e-05,
+      "loss": 0.7085,
+      "step": 4905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7779059436981894e-05,
+      "loss": 0.7405,
+      "step": 4906
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7777309345518405e-05,
+      "loss": 0.6501,
+      "step": 4907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7775558988416537e-05,
+      "loss": 0.7515,
+      "step": 4908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.777380836575676e-05,
+      "loss": 0.6787,
+      "step": 4909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7772057477619576e-05,
+      "loss": 0.6966,
+      "step": 4910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7770306324085486e-05,
+      "loss": 0.7137,
+      "step": 4911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.776855490523501e-05,
+      "loss": 0.6811,
+      "step": 4912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7766803221148677e-05,
+      "loss": 0.7337,
+      "step": 4913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7765051271907022e-05,
+      "loss": 0.7247,
+      "step": 4914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7763299057590607e-05,
+      "loss": 0.7034,
+      "step": 4915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7761546578279996e-05,
+      "loss": 0.6931,
+      "step": 4916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7759793834055765e-05,
+      "loss": 0.7833,
+      "step": 4917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.775804082499851e-05,
+      "loss": 0.7653,
+      "step": 4918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7756287551188827e-05,
+      "loss": 0.6919,
+      "step": 4919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7754534012707338e-05,
+      "loss": 0.7068,
+      "step": 4920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7752780209634663e-05,
+      "loss": 0.6799,
+      "step": 4921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.775102614205144e-05,
+      "loss": 0.6585,
+      "step": 4922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.774927181003833e-05,
+      "loss": 0.688,
+      "step": 4923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.774751721367599e-05,
+      "loss": 0.6675,
+      "step": 4924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7745762353045096e-05,
+      "loss": 0.6956,
+      "step": 4925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7744007228226333e-05,
+      "loss": 0.6802,
+      "step": 4926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7742251839300408e-05,
+      "loss": 0.7226,
+      "step": 4927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7740496186348027e-05,
+      "loss": 0.714,
+      "step": 4928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7738740269449918e-05,
+      "loss": 0.6309,
+      "step": 4929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7736984088686815e-05,
+      "loss": 0.7112,
+      "step": 4930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7735227644139466e-05,
+      "loss": 0.7282,
+      "step": 4931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7733470935888635e-05,
+      "loss": 0.6496,
+      "step": 4932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7731713964015093e-05,
+      "loss": 0.7706,
+      "step": 4933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7729956728599622e-05,
+      "loss": 0.7769,
+      "step": 4934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7728199229723024e-05,
+      "loss": 0.6585,
+      "step": 4935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7726441467466105e-05,
+      "loss": 0.7037,
+      "step": 4936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7724683441909686e-05,
+      "loss": 0.6796,
+      "step": 4937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7722925153134606e-05,
+      "loss": 0.6843,
+      "step": 4938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7721166601221698e-05,
+      "loss": 0.6827,
+      "step": 4939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7719407786251832e-05,
+      "loss": 0.7286,
+      "step": 4940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.771764870830587e-05,
+      "loss": 0.6731,
+      "step": 4941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.77158893674647e-05,
+      "loss": 0.7102,
+      "step": 4942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.771412976380921e-05,
+      "loss": 0.68,
+      "step": 4943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7712369897420307e-05,
+      "loss": 0.7012,
+      "step": 4944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.771060976837891e-05,
+      "loss": 0.7095,
+      "step": 4945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7708849376765953e-05,
+      "loss": 0.7262,
+      "step": 4946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.770708872266237e-05,
+      "loss": 0.6758,
+      "step": 4947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7705327806149122e-05,
+      "loss": 0.7158,
+      "step": 4948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.770356662730717e-05,
+      "loss": 0.6947,
+      "step": 4949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7701805186217497e-05,
+      "loss": 0.745,
+      "step": 4950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.770004348296109e-05,
+      "loss": 0.6512,
+      "step": 4951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7698281517618952e-05,
+      "loss": 0.6995,
+      "step": 4952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7696519290272104e-05,
+      "loss": 0.6959,
+      "step": 4953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.769475680100156e-05,
+      "loss": 0.6891,
+      "step": 4954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7692994049888365e-05,
+      "loss": 0.6654,
+      "step": 4955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7691231037013574e-05,
+      "loss": 0.7253,
+      "step": 4956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7689467762458243e-05,
+      "loss": 0.6541,
+      "step": 4957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.768770422630345e-05,
+      "loss": 0.7265,
+      "step": 4958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.768594042863028e-05,
+      "loss": 0.6503,
+      "step": 4959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.768417636951983e-05,
+      "loss": 0.6993,
+      "step": 4960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7682412049053218e-05,
+      "loss": 0.6739,
+      "step": 4961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.768064746731156e-05,
+      "loss": 0.7583,
+      "step": 4962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7678882624375994e-05,
+      "loss": 0.6895,
+      "step": 4963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7677117520327664e-05,
+      "loss": 0.7217,
+      "step": 4964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.767535215524773e-05,
+      "loss": 0.661,
+      "step": 4965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7673586529217364e-05,
+      "loss": 0.6421,
+      "step": 4966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.767182064231775e-05,
+      "loss": 0.6823,
+      "step": 4967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7670054494630076e-05,
+      "loss": 0.6669,
+      "step": 4968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7668288086235556e-05,
+      "loss": 0.7089,
+      "step": 4969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7666521417215408e-05,
+      "loss": 0.7881,
+      "step": 4970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.766475448765086e-05,
+      "loss": 0.7244,
+      "step": 4971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.766298729762315e-05,
+      "loss": 0.6877,
+      "step": 4972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7661219847213543e-05,
+      "loss": 0.7228,
+      "step": 4973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.76594521365033e-05,
+      "loss": 0.689,
+      "step": 4974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7657684165573706e-05,
+      "loss": 0.667,
+      "step": 4975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.765591593450604e-05,
+      "loss": 0.6841,
+      "step": 4976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.765414744338161e-05,
+      "loss": 0.6851,
+      "step": 4977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7652378692281736e-05,
+      "loss": 0.7251,
+      "step": 4978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7650609681287737e-05,
+      "loss": 0.7551,
+      "step": 4979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7648840410480954e-05,
+      "loss": 0.7487,
+      "step": 4980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7647070879942736e-05,
+      "loss": 0.6187,
+      "step": 4981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7645301089754452e-05,
+      "loss": 0.6964,
+      "step": 4982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7643531039997468e-05,
+      "loss": 0.6815,
+      "step": 4983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7641760730753174e-05,
+      "loss": 0.754,
+      "step": 4984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7639990162102966e-05,
+      "loss": 0.7524,
+      "step": 4985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.763821933412826e-05,
+      "loss": 0.6736,
+      "step": 4986
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7636448246910467e-05,
+      "loss": 0.613,
+      "step": 4987
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.763467690053103e-05,
+      "loss": 0.7118,
+      "step": 4988
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7632905295071393e-05,
+      "loss": 0.7109,
+      "step": 4989
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7631133430613012e-05,
+      "loss": 0.6858,
+      "step": 4990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7629361307237357e-05,
+      "loss": 0.6599,
+      "step": 4991
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.762758892502591e-05,
+      "loss": 0.8004,
+      "step": 4992
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7625816284060167e-05,
+      "loss": 0.6587,
+      "step": 4993
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7624043384421628e-05,
+      "loss": 0.7848,
+      "step": 4994
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7622270226191814e-05,
+      "loss": 0.6712,
+      "step": 4995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7620496809452254e-05,
+      "loss": 0.6953,
+      "step": 4996
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7618723134284484e-05,
+      "loss": 0.7278,
+      "step": 4997
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7616949200770066e-05,
+      "loss": 0.7828,
+      "step": 4998
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7615175008990555e-05,
+      "loss": 0.6536,
+      "step": 4999
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.761340055902754e-05,
+      "loss": 0.736,
+      "step": 5000
+    },
+    {
+      "epoch": 0.33,
+      "eval_text_loss": 0.6746059656143188,
+      "eval_text_runtime": 1.8909,
+      "eval_text_samples_per_second": 39.135,
+      "eval_text_steps_per_second": 1.058,
+      "step": 5000
+    },
+    {
+      "epoch": 0.33,
+      "eval_svit_loss": 0.8486446142196655,
+      "eval_svit_runtime": 6.7896,
+      "eval_svit_samples_per_second": 120.185,
+      "eval_svit_steps_per_second": 1.915,
+      "step": 5000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.761162585096259e-05,
+      "loss": 0.7134,
+      "step": 5001
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7609850884877324e-05,
+      "loss": 0.7189,
+      "step": 5002
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.760807566085335e-05,
+      "loss": 0.7107,
+      "step": 5003
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.760630017897229e-05,
+      "loss": 0.69,
+      "step": 5004
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7604524439315772e-05,
+      "loss": 0.7579,
+      "step": 5005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.760274844196546e-05,
+      "loss": 0.6688,
+      "step": 5006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7600972187002997e-05,
+      "loss": 0.7726,
+      "step": 5007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7599195674510068e-05,
+      "loss": 0.6762,
+      "step": 5008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7597418904568346e-05,
+      "loss": 0.6888,
+      "step": 5009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7595641877259535e-05,
+      "loss": 0.6737,
+      "step": 5010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7593864592665334e-05,
+      "loss": 0.6529,
+      "step": 5011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.759208705086747e-05,
+      "loss": 0.684,
+      "step": 5012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7590309251947668e-05,
+      "loss": 0.7272,
+      "step": 5013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7588531195987673e-05,
+      "loss": 0.7891,
+      "step": 5014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7586752883069237e-05,
+      "loss": 0.7175,
+      "step": 5015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7584974313274127e-05,
+      "loss": 0.6723,
+      "step": 5016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.758319548668412e-05,
+      "loss": 0.7281,
+      "step": 5017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.758141640338101e-05,
+      "loss": 0.6724,
+      "step": 5018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7579637063446594e-05,
+      "loss": 0.7855,
+      "step": 5019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7577857466962686e-05,
+      "loss": 0.6657,
+      "step": 5020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7576077614011114e-05,
+      "loss": 0.6453,
+      "step": 5021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.757429750467371e-05,
+      "loss": 0.7348,
+      "step": 5022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.757251713903233e-05,
+      "loss": 0.7131,
+      "step": 5023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.757073651716883e-05,
+      "loss": 0.6893,
+      "step": 5024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.756895563916508e-05,
+      "loss": 0.6552,
+      "step": 5025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7567174505102964e-05,
+      "loss": 0.6849,
+      "step": 5026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7565393115064385e-05,
+      "loss": 0.6705,
+      "step": 5027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7563611469131243e-05,
+      "loss": 0.641,
+      "step": 5028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7561829567385465e-05,
+      "loss": 0.6899,
+      "step": 5029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7560047409908974e-05,
+      "loss": 0.7214,
+      "step": 5030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7558264996783716e-05,
+      "loss": 0.7076,
+      "step": 5031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7556482328091647e-05,
+      "loss": 0.6339,
+      "step": 5032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7554699403914732e-05,
+      "loss": 0.6633,
+      "step": 5033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7552916224334948e-05,
+      "loss": 0.7253,
+      "step": 5034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.755113278943429e-05,
+      "loss": 0.7146,
+      "step": 5035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7549349099294754e-05,
+      "loss": 0.6812,
+      "step": 5036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7547565153998354e-05,
+      "loss": 0.6694,
+      "step": 5037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.754578095362712e-05,
+      "loss": 0.7129,
+      "step": 5038
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7543996498263085e-05,
+      "loss": 0.7019,
+      "step": 5039
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7542211787988296e-05,
+      "loss": 0.7247,
+      "step": 5040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7540426822884814e-05,
+      "loss": 0.6605,
+      "step": 5041
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7538641603034716e-05,
+      "loss": 0.6762,
+      "step": 5042
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.753685612852008e-05,
+      "loss": 0.6709,
+      "step": 5043
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7535070399423e-05,
+      "loss": 0.7107,
+      "step": 5044
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7533284415825594e-05,
+      "loss": 0.7189,
+      "step": 5045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7531498177809967e-05,
+      "loss": 0.6937,
+      "step": 5046
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7529711685458258e-05,
+      "loss": 0.743,
+      "step": 5047
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7527924938852607e-05,
+      "loss": 0.7129,
+      "step": 5048
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7526137938075165e-05,
+      "loss": 0.7244,
+      "step": 5049
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7524350683208103e-05,
+      "loss": 0.673,
+      "step": 5050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7522563174333593e-05,
+      "loss": 0.7296,
+      "step": 5051
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7520775411533832e-05,
+      "loss": 0.7332,
+      "step": 5052
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7518987394891013e-05,
+      "loss": 0.7155,
+      "step": 5053
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7517199124487348e-05,
+      "loss": 0.6781,
+      "step": 5054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7515410600405063e-05,
+      "loss": 0.6854,
+      "step": 5055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7513621822726397e-05,
+      "loss": 0.7287,
+      "step": 5056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7511832791533595e-05,
+      "loss": 0.7034,
+      "step": 5057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.751004350690891e-05,
+      "loss": 0.7361,
+      "step": 5058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7508253968934627e-05,
+      "loss": 0.7242,
+      "step": 5059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7506464177693012e-05,
+      "loss": 0.6901,
+      "step": 5060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.750467413326637e-05,
+      "loss": 0.7049,
+      "step": 5061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7502883835737e-05,
+      "loss": 0.7306,
+      "step": 5062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7501093285187223e-05,
+      "loss": 0.7329,
+      "step": 5063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7499302481699366e-05,
+      "loss": 0.7071,
+      "step": 5064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.749751142535577e-05,
+      "loss": 0.7435,
+      "step": 5065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7495720116238786e-05,
+      "loss": 0.7392,
+      "step": 5066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.749392855443078e-05,
+      "loss": 0.6723,
+      "step": 5067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7492136740014126e-05,
+      "loss": 0.7418,
+      "step": 5068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7490344673071212e-05,
+      "loss": 0.6816,
+      "step": 5069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7488552353684434e-05,
+      "loss": 0.7836,
+      "step": 5070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7486759781936203e-05,
+      "loss": 0.7354,
+      "step": 5071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7484966957908947e-05,
+      "loss": 0.7422,
+      "step": 5072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7483173881685087e-05,
+      "loss": 0.722,
+      "step": 5073
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7481380553347075e-05,
+      "loss": 0.6586,
+      "step": 5074
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7479586972977368e-05,
+      "loss": 0.7135,
+      "step": 5075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7477793140658435e-05,
+      "loss": 0.7644,
+      "step": 5076
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7475999056472755e-05,
+      "loss": 0.7067,
+      "step": 5077
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7474204720502817e-05,
+      "loss": 0.787,
+      "step": 5078
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7472410132831124e-05,
+      "loss": 0.6059,
+      "step": 5079
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.747061529354019e-05,
+      "loss": 0.7521,
+      "step": 5080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7468820202712544e-05,
+      "loss": 0.6864,
+      "step": 5081
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.746702486043072e-05,
+      "loss": 0.7377,
+      "step": 5082
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.746522926677727e-05,
+      "loss": 0.6501,
+      "step": 5083
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7463433421834757e-05,
+      "loss": 0.745,
+      "step": 5084
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7461637325685747e-05,
+      "loss": 0.6855,
+      "step": 5085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.745984097841283e-05,
+      "loss": 0.7144,
+      "step": 5086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.74580443800986e-05,
+      "loss": 0.759,
+      "step": 5087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7456247530825657e-05,
+      "loss": 0.6773,
+      "step": 5088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7454450430676626e-05,
+      "loss": 0.6725,
+      "step": 5089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7452653079734137e-05,
+      "loss": 0.7042,
+      "step": 5090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.745085547808083e-05,
+      "loss": 0.725,
+      "step": 5091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7449057625799354e-05,
+      "loss": 0.7146,
+      "step": 5092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.744725952297238e-05,
+      "loss": 0.736,
+      "step": 5093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.744546116968258e-05,
+      "loss": 0.6773,
+      "step": 5094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7443662566012647e-05,
+      "loss": 0.694,
+      "step": 5095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7441863712045273e-05,
+      "loss": 0.6898,
+      "step": 5096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7440064607863173e-05,
+      "loss": 0.7489,
+      "step": 5097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7438265253549067e-05,
+      "loss": 0.7114,
+      "step": 5098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.743646564918569e-05,
+      "loss": 0.6981,
+      "step": 5099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7434665794855787e-05,
+      "loss": 0.6633,
+      "step": 5100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7432865690642115e-05,
+      "loss": 0.6546,
+      "step": 5101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7431065336627438e-05,
+      "loss": 0.7166,
+      "step": 5102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.742926473289454e-05,
+      "loss": 0.623,
+      "step": 5103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7427463879526213e-05,
+      "loss": 0.6711,
+      "step": 5104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7425662776605255e-05,
+      "loss": 0.7037,
+      "step": 5105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7423861424214484e-05,
+      "loss": 0.7368,
+      "step": 5106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7422059822436725e-05,
+      "loss": 0.687,
+      "step": 5107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.742025797135481e-05,
+      "loss": 0.6447,
+      "step": 5108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7418455871051593e-05,
+      "loss": 0.5712,
+      "step": 5109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.741665352160993e-05,
+      "loss": 0.8263,
+      "step": 5110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7414850923112696e-05,
+      "loss": 0.7511,
+      "step": 5111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7413048075642772e-05,
+      "loss": 0.7615,
+      "step": 5112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7411244979283055e-05,
+      "loss": 0.6476,
+      "step": 5113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7409441634116446e-05,
+      "loss": 0.6778,
+      "step": 5114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7407638040225862e-05,
+      "loss": 0.6836,
+      "step": 5115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7405834197694232e-05,
+      "loss": 0.6926,
+      "step": 5116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.74040301066045e-05,
+      "loss": 0.6887,
+      "step": 5117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7402225767039616e-05,
+      "loss": 0.6981,
+      "step": 5118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7400421179082543e-05,
+      "loss": 0.704,
+      "step": 5119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.739861634281625e-05,
+      "loss": 0.7555,
+      "step": 5120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.739681125832373e-05,
+      "loss": 0.7323,
+      "step": 5121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7395005925687975e-05,
+      "loss": 0.723,
+      "step": 5122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7393200344991994e-05,
+      "loss": 0.669,
+      "step": 5123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.739139451631881e-05,
+      "loss": 0.7288,
+      "step": 5124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7389588439751453e-05,
+      "loss": 0.7187,
+      "step": 5125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7387782115372962e-05,
+      "loss": 0.7379,
+      "step": 5126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7385975543266394e-05,
+      "loss": 0.6898,
+      "step": 5127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7384168723514816e-05,
+      "loss": 0.7485,
+      "step": 5128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7382361656201308e-05,
+      "loss": 0.7211,
+      "step": 5129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7380554341408947e-05,
+      "loss": 0.7248,
+      "step": 5130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.737874677922084e-05,
+      "loss": 0.7524,
+      "step": 5131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7376938969720096e-05,
+      "loss": 0.7446,
+      "step": 5132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7375130912989845e-05,
+      "loss": 0.8096,
+      "step": 5133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.737332260911321e-05,
+      "loss": 0.6701,
+      "step": 5134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7371514058173338e-05,
+      "loss": 0.7313,
+      "step": 5135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.736970526025339e-05,
+      "loss": 0.6466,
+      "step": 5136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7367896215436534e-05,
+      "loss": 0.7472,
+      "step": 5137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7366086923805945e-05,
+      "loss": 0.6959,
+      "step": 5138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7364277385444815e-05,
+      "loss": 0.7342,
+      "step": 5139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.736246760043635e-05,
+      "loss": 0.6792,
+      "step": 5140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7360657568863753e-05,
+      "loss": 0.7357,
+      "step": 5141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.735884729081026e-05,
+      "loss": 0.6668,
+      "step": 5142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7357036766359102e-05,
+      "loss": 0.7276,
+      "step": 5143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7355225995593522e-05,
+      "loss": 0.7082,
+      "step": 5144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7353414978596782e-05,
+      "loss": 0.6209,
+      "step": 5145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7351603715452156e-05,
+      "loss": 0.7255,
+      "step": 5146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.734979220624292e-05,
+      "loss": 0.6691,
+      "step": 5147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.734798045105237e-05,
+      "loss": 0.6525,
+      "step": 5148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7346168449963803e-05,
+      "loss": 0.6921,
+      "step": 5149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7344356203060542e-05,
+      "loss": 0.7356,
+      "step": 5150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.734254371042591e-05,
+      "loss": 0.7087,
+      "step": 5151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.734073097214324e-05,
+      "loss": 0.7326,
+      "step": 5152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7338917988295892e-05,
+      "loss": 0.6925,
+      "step": 5153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.733710475896722e-05,
+      "loss": 0.7047,
+      "step": 5154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7335291284240596e-05,
+      "loss": 0.7403,
+      "step": 5155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.73334775641994e-05,
+      "loss": 0.6537,
+      "step": 5156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.733166359892703e-05,
+      "loss": 0.7227,
+      "step": 5157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.732984938850689e-05,
+      "loss": 0.6522,
+      "step": 5158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7328034933022398e-05,
+      "loss": 0.6874,
+      "step": 5159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7326220232556977e-05,
+      "loss": 0.7088,
+      "step": 5160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7324405287194073e-05,
+      "loss": 0.7491,
+      "step": 5161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.732259009701713e-05,
+      "loss": 0.681,
+      "step": 5162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7320774662109616e-05,
+      "loss": 0.706,
+      "step": 5163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7318958982555e-05,
+      "loss": 0.7146,
+      "step": 5164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.731714305843677e-05,
+      "loss": 0.6968,
+      "step": 5165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7315326889838416e-05,
+      "loss": 0.7221,
+      "step": 5166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7313510476843452e-05,
+      "loss": 0.681,
+      "step": 5167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7311693819535384e-05,
+      "loss": 0.6577,
+      "step": 5168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.730987691799775e-05,
+      "loss": 0.7401,
+      "step": 5169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7308059772314096e-05,
+      "loss": 0.7725,
+      "step": 5170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7306242382567962e-05,
+      "loss": 0.6874,
+      "step": 5171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7304424748842915e-05,
+      "loss": 0.724,
+      "step": 5172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.730260687122253e-05,
+      "loss": 0.6967,
+      "step": 5173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7300788749790393e-05,
+      "loss": 0.7448,
+      "step": 5174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.72989703846301e-05,
+      "loss": 0.6506,
+      "step": 5175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7297151775825253e-05,
+      "loss": 0.7235,
+      "step": 5176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.729533292345948e-05,
+      "loss": 0.7045,
+      "step": 5177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7293513827616406e-05,
+      "loss": 0.6531,
+      "step": 5178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7291694488379674e-05,
+      "loss": 0.7788,
+      "step": 5179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7289874905832935e-05,
+      "loss": 0.7588,
+      "step": 5180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7288055080059855e-05,
+      "loss": 0.6769,
+      "step": 5181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7286235011144107e-05,
+      "loss": 0.7013,
+      "step": 5182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.728441469916938e-05,
+      "loss": 0.7641,
+      "step": 5183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7282594144219363e-05,
+      "loss": 0.7231,
+      "step": 5184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7280773346377774e-05,
+      "loss": 0.6979,
+      "step": 5185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7278952305728327e-05,
+      "loss": 0.7241,
+      "step": 5186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7277131022354757e-05,
+      "loss": 0.7337,
+      "step": 5187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.72753094963408e-05,
+      "loss": 0.6819,
+      "step": 5188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7273487727770218e-05,
+      "loss": 0.7728,
+      "step": 5189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.727166571672676e-05,
+      "loss": 0.7,
+      "step": 5190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.726984346329422e-05,
+      "loss": 0.7016,
+      "step": 5191
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.726802096755637e-05,
+      "loss": 0.691,
+      "step": 5192
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7266198229597014e-05,
+      "loss": 0.7148,
+      "step": 5193
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7264375249499957e-05,
+      "loss": 0.6493,
+      "step": 5194
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.726255202734903e-05,
+      "loss": 0.725,
+      "step": 5195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7260728563228045e-05,
+      "loss": 0.6479,
+      "step": 5196
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7258904857220862e-05,
+      "loss": 0.6757,
+      "step": 5197
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7257080909411323e-05,
+      "loss": 0.7085,
+      "step": 5198
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.72552567198833e-05,
+      "loss": 0.6535,
+      "step": 5199
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7253432288720664e-05,
+      "loss": 0.7149,
+      "step": 5200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7251607616007305e-05,
+      "loss": 0.6552,
+      "step": 5201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7249782701827115e-05,
+      "loss": 0.7112,
+      "step": 5202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7247957546264006e-05,
+      "loss": 0.7103,
+      "step": 5203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.72461321494019e-05,
+      "loss": 0.7,
+      "step": 5204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7244306511324723e-05,
+      "loss": 0.7236,
+      "step": 5205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7242480632116424e-05,
+      "loss": 0.6409,
+      "step": 5206
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.724065451186095e-05,
+      "loss": 0.7123,
+      "step": 5207
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.723882815064227e-05,
+      "loss": 0.751,
+      "step": 5208
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7237001548544355e-05,
+      "loss": 0.7154,
+      "step": 5209
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.723517470565119e-05,
+      "loss": 0.7179,
+      "step": 5210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.723334762204678e-05,
+      "loss": 0.6919,
+      "step": 5211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.723152029781513e-05,
+      "loss": 0.7159,
+      "step": 5212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.722969273304026e-05,
+      "loss": 0.7098,
+      "step": 5213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.72278649278062e-05,
+      "loss": 0.7557,
+      "step": 5214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7226036882196987e-05,
+      "loss": 0.7018,
+      "step": 5215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7224208596296686e-05,
+      "loss": 0.645,
+      "step": 5216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7222380070189346e-05,
+      "loss": 0.7197,
+      "step": 5217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7220551303959056e-05,
+      "loss": 0.6929,
+      "step": 5218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.721872229768989e-05,
+      "loss": 0.648,
+      "step": 5219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.721689305146595e-05,
+      "loss": 0.7224,
+      "step": 5220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7215063565371347e-05,
+      "loss": 0.6648,
+      "step": 5221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7213233839490197e-05,
+      "loss": 0.6717,
+      "step": 5222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7211403873906634e-05,
+      "loss": 0.7274,
+      "step": 5223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7209573668704788e-05,
+      "loss": 0.6833,
+      "step": 5224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7207743223968822e-05,
+      "loss": 0.699,
+      "step": 5225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7205912539782893e-05,
+      "loss": 0.6958,
+      "step": 5226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.720408161623118e-05,
+      "loss": 0.6752,
+      "step": 5227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7202250453397864e-05,
+      "loss": 0.6709,
+      "step": 5228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7200419051367145e-05,
+      "loss": 0.6957,
+      "step": 5229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.719858741022323e-05,
+      "loss": 0.7889,
+      "step": 5230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7196755530050333e-05,
+      "loss": 0.7154,
+      "step": 5231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7194923410932684e-05,
+      "loss": 0.6593,
+      "step": 5232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7193091052954526e-05,
+      "loss": 0.72,
+      "step": 5233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7191258456200108e-05,
+      "loss": 0.6992,
+      "step": 5234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7189425620753696e-05,
+      "loss": 0.6774,
+      "step": 5235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7187592546699556e-05,
+      "loss": 0.689,
+      "step": 5236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.718575923412198e-05,
+      "loss": 0.7149,
+      "step": 5237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7183925683105253e-05,
+      "loss": 0.6476,
+      "step": 5238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7182091893733687e-05,
+      "loss": 0.6362,
+      "step": 5239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7180257866091603e-05,
+      "loss": 0.7023,
+      "step": 5240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7178423600263323e-05,
+      "loss": 0.6059,
+      "step": 5241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7176589096333187e-05,
+      "loss": 0.698,
+      "step": 5242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7174754354385545e-05,
+      "loss": 0.7002,
+      "step": 5243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7172919374504753e-05,
+      "loss": 0.769,
+      "step": 5244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7171084156775192e-05,
+      "loss": 0.7234,
+      "step": 5245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.716924870128124e-05,
+      "loss": 0.6707,
+      "step": 5246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.716741300810729e-05,
+      "loss": 0.7325,
+      "step": 5247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7165577077337744e-05,
+      "loss": 0.7078,
+      "step": 5248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7163740909057025e-05,
+      "loss": 0.6742,
+      "step": 5249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7161904503349553e-05,
+      "loss": 0.6567,
+      "step": 5250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7160067860299765e-05,
+      "loss": 0.7459,
+      "step": 5251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7158230979992115e-05,
+      "loss": 0.7548,
+      "step": 5252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7156393862511053e-05,
+      "loss": 0.7078,
+      "step": 5253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7154556507941057e-05,
+      "loss": 0.6887,
+      "step": 5254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7152718916366604e-05,
+      "loss": 0.7093,
+      "step": 5255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7150881087872184e-05,
+      "loss": 0.6944,
+      "step": 5256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7149043022542304e-05,
+      "loss": 0.6745,
+      "step": 5257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7147204720461476e-05,
+      "loss": 0.6877,
+      "step": 5258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7145366181714223e-05,
+      "loss": 0.6863,
+      "step": 5259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7143527406385083e-05,
+      "loss": 0.6714,
+      "step": 5260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.71416883945586e-05,
+      "loss": 0.7045,
+      "step": 5261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7139849146319326e-05,
+      "loss": 0.7602,
+      "step": 5262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7138009661751838e-05,
+      "loss": 0.6946,
+      "step": 5263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.713616994094071e-05,
+      "loss": 0.6962,
+      "step": 5264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7134329983970537e-05,
+      "loss": 0.7165,
+      "step": 5265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.713248979092591e-05,
+      "loss": 0.6829,
+      "step": 5266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.713064936189145e-05,
+      "loss": 0.6726,
+      "step": 5267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7128808696951773e-05,
+      "loss": 0.706,
+      "step": 5268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7126967796191513e-05,
+      "loss": 0.6644,
+      "step": 5269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.712512665969531e-05,
+      "loss": 0.6982,
+      "step": 5270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7123285287547833e-05,
+      "loss": 0.69,
+      "step": 5271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7121443679833735e-05,
+      "loss": 0.7387,
+      "step": 5272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7119601836637692e-05,
+      "loss": 0.645,
+      "step": 5273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7117759758044396e-05,
+      "loss": 0.7348,
+      "step": 5274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7115917444138544e-05,
+      "loss": 0.664,
+      "step": 5275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7114074895004846e-05,
+      "loss": 0.7152,
+      "step": 5276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7112232110728016e-05,
+      "loss": 0.769,
+      "step": 5277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.711038909139279e-05,
+      "loss": 0.712,
+      "step": 5278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.710854583708391e-05,
+      "loss": 0.7681,
+      "step": 5279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7106702347886124e-05,
+      "loss": 0.6661,
+      "step": 5280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7104858623884196e-05,
+      "loss": 0.7268,
+      "step": 5281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7103014665162902e-05,
+      "loss": 0.6865,
+      "step": 5282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7101170471807024e-05,
+      "loss": 0.6889,
+      "step": 5283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.709932604390136e-05,
+      "loss": 0.6573,
+      "step": 5284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7097481381530716e-05,
+      "loss": 0.7296,
+      "step": 5285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.70956364847799e-05,
+      "loss": 0.8039,
+      "step": 5286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7093791353733753e-05,
+      "loss": 0.7464,
+      "step": 5287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7091945988477107e-05,
+      "loss": 0.6698,
+      "step": 5288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.709010038909481e-05,
+      "loss": 0.6701,
+      "step": 5289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7088254555671724e-05,
+      "loss": 0.6516,
+      "step": 5290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.708640848829272e-05,
+      "loss": 0.7114,
+      "step": 5291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7084562187042678e-05,
+      "loss": 0.6948,
+      "step": 5292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7082715652006494e-05,
+      "loss": 0.6757,
+      "step": 5293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.708086888326906e-05,
+      "loss": 0.6765,
+      "step": 5294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7079021880915303e-05,
+      "loss": 0.7673,
+      "step": 5295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7077174645030137e-05,
+      "loss": 0.6729,
+      "step": 5296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7075327175698506e-05,
+      "loss": 0.6721,
+      "step": 5297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7073479473005347e-05,
+      "loss": 0.7444,
+      "step": 5298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.707163153703563e-05,
+      "loss": 0.7125,
+      "step": 5299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7069783367874304e-05,
+      "loss": 0.7201,
+      "step": 5300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7067934965606365e-05,
+      "loss": 0.7812,
+      "step": 5301
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.706608633031679e-05,
+      "loss": 0.6762,
+      "step": 5302
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7064237462090577e-05,
+      "loss": 0.7508,
+      "step": 5303
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.706238836101275e-05,
+      "loss": 0.7962,
+      "step": 5304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7060539027168318e-05,
+      "loss": 0.712,
+      "step": 5305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7058689460642317e-05,
+      "loss": 0.7293,
+      "step": 5306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7056839661519785e-05,
+      "loss": 0.7077,
+      "step": 5307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.705498962988578e-05,
+      "loss": 0.7121,
+      "step": 5308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7053139365825366e-05,
+      "loss": 0.7163,
+      "step": 5309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7051288869423614e-05,
+      "loss": 0.7101,
+      "step": 5310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.704943814076561e-05,
+      "loss": 0.701,
+      "step": 5311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7047587179936453e-05,
+      "loss": 0.7245,
+      "step": 5312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7045735987021243e-05,
+      "loss": 0.6817,
+      "step": 5313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7043884562105105e-05,
+      "loss": 0.6464,
+      "step": 5314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.704203290527316e-05,
+      "loss": 0.704,
+      "step": 5315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7040181016610548e-05,
+      "loss": 0.7476,
+      "step": 5316
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7038328896202423e-05,
+      "loss": 0.6478,
+      "step": 5317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7036476544133938e-05,
+      "loss": 0.6805,
+      "step": 5318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7034623960490264e-05,
+      "loss": 0.7327,
+      "step": 5319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7032771145356588e-05,
+      "loss": 0.7693,
+      "step": 5320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.70309180988181e-05,
+      "loss": 0.7308,
+      "step": 5321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7029064820959997e-05,
+      "loss": 0.6448,
+      "step": 5322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7027211311867498e-05,
+      "loss": 0.7228,
+      "step": 5323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.702535757162582e-05,
+      "loss": 0.7012,
+      "step": 5324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7023503600320207e-05,
+      "loss": 0.6872,
+      "step": 5325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7021649398035893e-05,
+      "loss": 0.6709,
+      "step": 5326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7019794964858144e-05,
+      "loss": 0.6518,
+      "step": 5327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7017940300872216e-05,
+      "loss": 0.6829,
+      "step": 5328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.701608540616339e-05,
+      "loss": 0.6803,
+      "step": 5329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7014230280816954e-05,
+      "loss": 0.6915,
+      "step": 5330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.701237492491821e-05,
+      "loss": 0.7364,
+      "step": 5331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.701051933855246e-05,
+      "loss": 0.7806,
+      "step": 5332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7008663521805026e-05,
+      "loss": 0.7182,
+      "step": 5333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7006807474761235e-05,
+      "loss": 0.6567,
+      "step": 5334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7004951197506433e-05,
+      "loss": 0.6538,
+      "step": 5335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7003094690125962e-05,
+      "loss": 0.725,
+      "step": 5336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7001237952705193e-05,
+      "loss": 0.6979,
+      "step": 5337
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.699938098532949e-05,
+      "loss": 0.6617,
+      "step": 5338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6997523788084242e-05,
+      "loss": 0.7676,
+      "step": 5339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6995666361054842e-05,
+      "loss": 0.6154,
+      "step": 5340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6993808704326688e-05,
+      "loss": 0.6775,
+      "step": 5341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.69919508179852e-05,
+      "loss": 0.6732,
+      "step": 5342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6990092702115795e-05,
+      "loss": 0.7743,
+      "step": 5343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.698823435680392e-05,
+      "loss": 0.6779,
+      "step": 5344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6986375782135013e-05,
+      "loss": 0.6543,
+      "step": 5345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.698451697819453e-05,
+      "loss": 0.6865,
+      "step": 5346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6982657945067942e-05,
+      "loss": 0.6799,
+      "step": 5347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6980798682840726e-05,
+      "loss": 0.7045,
+      "step": 5348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.697893919159837e-05,
+      "loss": 0.6181,
+      "step": 5349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.697707947142637e-05,
+      "loss": 0.7698,
+      "step": 5350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.697521952241024e-05,
+      "loss": 0.7192,
+      "step": 5351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6973359344635494e-05,
+      "loss": 0.6808,
+      "step": 5352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6971498938187668e-05,
+      "loss": 0.668,
+      "step": 5353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6969638303152297e-05,
+      "loss": 0.644,
+      "step": 5354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6967777439614935e-05,
+      "loss": 0.661,
+      "step": 5355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6965916347661146e-05,
+      "loss": 0.6024,
+      "step": 5356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.69640550273765e-05,
+      "loss": 0.6802,
+      "step": 5357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.696219347884658e-05,
+      "loss": 0.6255,
+      "step": 5358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.696033170215698e-05,
+      "loss": 0.6942,
+      "step": 5359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6958469697393297e-05,
+      "loss": 0.6866,
+      "step": 5360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6956607464641155e-05,
+      "loss": 0.762,
+      "step": 5361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6954745003986176e-05,
+      "loss": 0.6217,
+      "step": 5362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6952882315513992e-05,
+      "loss": 0.7234,
+      "step": 5363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.695101939931025e-05,
+      "loss": 0.6881,
+      "step": 5364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.694915625546061e-05,
+      "loss": 0.6034,
+      "step": 5365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6947292884050734e-05,
+      "loss": 0.7578,
+      "step": 5366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.69454292851663e-05,
+      "loss": 0.7114,
+      "step": 5367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6943565458893e-05,
+      "loss": 0.7819,
+      "step": 5368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6941701405316524e-05,
+      "loss": 0.7477,
+      "step": 5369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6939837124522586e-05,
+      "loss": 0.6666,
+      "step": 5370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6937972616596905e-05,
+      "loss": 0.713,
+      "step": 5371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6936107881625205e-05,
+      "loss": 0.6947,
+      "step": 5372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.693424291969323e-05,
+      "loss": 0.7296,
+      "step": 5373
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.693237773088673e-05,
+      "loss": 0.7069,
+      "step": 5374
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6930512315291466e-05,
+      "loss": 0.7483,
+      "step": 5375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.692864667299321e-05,
+      "loss": 0.6324,
+      "step": 5376
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6926780804077742e-05,
+      "loss": 0.7262,
+      "step": 5377
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.692491470863085e-05,
+      "loss": 0.6409,
+      "step": 5378
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6923048386738345e-05,
+      "loss": 0.6571,
+      "step": 5379
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6921181838486034e-05,
+      "loss": 0.7,
+      "step": 5380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6919315063959736e-05,
+      "loss": 0.6781,
+      "step": 5381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6917448063245294e-05,
+      "loss": 0.7414,
+      "step": 5382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.691558083642855e-05,
+      "loss": 0.6677,
+      "step": 5383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6913713383595353e-05,
+      "loss": 0.6901,
+      "step": 5384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6911845704831568e-05,
+      "loss": 0.7236,
+      "step": 5385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6909977800223074e-05,
+      "loss": 0.7289,
+      "step": 5386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.690810966985576e-05,
+      "loss": 0.6811,
+      "step": 5387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6906241313815512e-05,
+      "loss": 0.7459,
+      "step": 5388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6904372732188243e-05,
+      "loss": 0.6335,
+      "step": 5389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6902503925059865e-05,
+      "loss": 0.7708,
+      "step": 5390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6900634892516314e-05,
+      "loss": 0.7928,
+      "step": 5391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6898765634643514e-05,
+      "loss": 0.6779,
+      "step": 5392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6896896151527428e-05,
+      "loss": 0.6775,
+      "step": 5393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6895026443254003e-05,
+      "loss": 0.6725,
+      "step": 5394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689315650990921e-05,
+      "loss": 0.6852,
+      "step": 5395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689128635157903e-05,
+      "loss": 0.7196,
+      "step": 5396
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6889415968349447e-05,
+      "loss": 0.7245,
+      "step": 5397
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6887545360306467e-05,
+      "loss": 0.6885,
+      "step": 5398
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.68856745275361e-05,
+      "loss": 0.6589,
+      "step": 5399
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6883803470124357e-05,
+      "loss": 0.6391,
+      "step": 5400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.688193218815728e-05,
+      "loss": 0.6421,
+      "step": 5401
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.68800606817209e-05,
+      "loss": 0.6453,
+      "step": 5402
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6878188950901275e-05,
+      "loss": 0.6823,
+      "step": 5403
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6876316995784462e-05,
+      "loss": 0.6897,
+      "step": 5404
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6874444816456534e-05,
+      "loss": 0.6521,
+      "step": 5405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.687257241300358e-05,
+      "loss": 0.6228,
+      "step": 5406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.687069978551168e-05,
+      "loss": 0.6657,
+      "step": 5407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6868826934066946e-05,
+      "loss": 0.6772,
+      "step": 5408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6866953858755485e-05,
+      "loss": 0.5761,
+      "step": 5409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6865080559663424e-05,
+      "loss": 0.7302,
+      "step": 5410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6863207036876895e-05,
+      "loss": 0.6653,
+      "step": 5411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6861333290482042e-05,
+      "loss": 0.6775,
+      "step": 5412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.685945932056502e-05,
+      "loss": 0.7398,
+      "step": 5413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6857585127211994e-05,
+      "loss": 0.7159,
+      "step": 5414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6855710710509135e-05,
+      "loss": 0.6742,
+      "step": 5415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6853836070542635e-05,
+      "loss": 0.6929,
+      "step": 5416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.685196120739868e-05,
+      "loss": 0.6682,
+      "step": 5417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6850086121163484e-05,
+      "loss": 0.6462,
+      "step": 5418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6848210811923255e-05,
+      "loss": 0.6832,
+      "step": 5419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6846335279764226e-05,
+      "loss": 0.5841,
+      "step": 5420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6844459524772627e-05,
+      "loss": 0.7322,
+      "step": 5421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6842583547034708e-05,
+      "loss": 0.6826,
+      "step": 5422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6840707346636726e-05,
+      "loss": 0.6645,
+      "step": 5423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.683883092366495e-05,
+      "loss": 0.6634,
+      "step": 5424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.683695427820565e-05,
+      "loss": 0.6589,
+      "step": 5425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.683507741034512e-05,
+      "loss": 0.6086,
+      "step": 5426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6833200320169655e-05,
+      "loss": 0.6666,
+      "step": 5427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6831323007765563e-05,
+      "loss": 0.7399,
+      "step": 5428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6829445473219162e-05,
+      "loss": 0.5934,
+      "step": 5429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.682756771661678e-05,
+      "loss": 0.6862,
+      "step": 5430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6825689738044758e-05,
+      "loss": 0.7219,
+      "step": 5431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.682381153758944e-05,
+      "loss": 0.6756,
+      "step": 5432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6821933115337193e-05,
+      "loss": 0.7329,
+      "step": 5433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6820054471374373e-05,
+      "loss": 0.706,
+      "step": 5434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6818175605787375e-05,
+      "loss": 0.7209,
+      "step": 5435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6816296518662576e-05,
+      "loss": 0.6532,
+      "step": 5436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6814417210086383e-05,
+      "loss": 0.7105,
+      "step": 5437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6812537680145203e-05,
+      "loss": 0.6852,
+      "step": 5438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6810657928925452e-05,
+      "loss": 0.6938,
+      "step": 5439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.680877795651357e-05,
+      "loss": 0.6514,
+      "step": 5440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6806897762995988e-05,
+      "loss": 0.6949,
+      "step": 5441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6805017348459162e-05,
+      "loss": 0.6725,
+      "step": 5442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.680313671298955e-05,
+      "loss": 0.6669,
+      "step": 5443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6801255856673624e-05,
+      "loss": 0.7682,
+      "step": 5444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6799374779597865e-05,
+      "loss": 0.6965,
+      "step": 5445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6797493481848764e-05,
+      "loss": 0.6683,
+      "step": 5446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.679561196351282e-05,
+      "loss": 0.7166,
+      "step": 5447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6793730224676553e-05,
+      "loss": 0.7064,
+      "step": 5448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6791848265426473e-05,
+      "loss": 0.6499,
+      "step": 5449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6789966085849124e-05,
+      "loss": 0.668,
+      "step": 5450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6788083686031035e-05,
+      "loss": 0.6569,
+      "step": 5451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6786201066058766e-05,
+      "loss": 0.6842,
+      "step": 5452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6784318226018877e-05,
+      "loss": 0.6801,
+      "step": 5453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.678243516599794e-05,
+      "loss": 0.7689,
+      "step": 5454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6780551886082537e-05,
+      "loss": 0.7598,
+      "step": 5455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6778668386359264e-05,
+      "loss": 0.7121,
+      "step": 5456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.677678466691472e-05,
+      "loss": 0.7101,
+      "step": 5457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6774900727835517e-05,
+      "loss": 0.6754,
+      "step": 5458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6773016569208282e-05,
+      "loss": 0.7167,
+      "step": 5459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6771132191119642e-05,
+      "loss": 0.706,
+      "step": 5460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6769247593656245e-05,
+      "loss": 0.6451,
+      "step": 5461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6767362776904742e-05,
+      "loss": 0.6858,
+      "step": 5462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6765477740951796e-05,
+      "loss": 0.6846,
+      "step": 5463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.676359248588408e-05,
+      "loss": 0.6424,
+      "step": 5464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6761707011788278e-05,
+      "loss": 0.7008,
+      "step": 5465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6759821318751084e-05,
+      "loss": 0.6859,
+      "step": 5466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.67579354068592e-05,
+      "loss": 0.7641,
+      "step": 5467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.675604927619934e-05,
+      "loss": 0.6952,
+      "step": 5468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.675416292685823e-05,
+      "loss": 0.7286,
+      "step": 5469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6752276358922595e-05,
+      "loss": 0.767,
+      "step": 5470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6750389572479186e-05,
+      "loss": 0.6684,
+      "step": 5471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.674850256761476e-05,
+      "loss": 0.7268,
+      "step": 5472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6746615344416072e-05,
+      "loss": 0.6821,
+      "step": 5473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6744727902969905e-05,
+      "loss": 0.6615,
+      "step": 5474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6742840243363033e-05,
+      "loss": 0.6567,
+      "step": 5475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6740952365682256e-05,
+      "loss": 0.6967,
+      "step": 5476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.673906427001438e-05,
+      "loss": 0.7198,
+      "step": 5477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.673717595644621e-05,
+      "loss": 0.7854,
+      "step": 5478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.673528742506458e-05,
+      "loss": 0.7081,
+      "step": 5479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6733398675956313e-05,
+      "loss": 0.6575,
+      "step": 5480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.673150970920826e-05,
+      "loss": 0.6601,
+      "step": 5481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6729620524907278e-05,
+      "loss": 0.6121,
+      "step": 5482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6727731123140224e-05,
+      "loss": 0.7143,
+      "step": 5483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.672584150399398e-05,
+      "loss": 0.6426,
+      "step": 5484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.672395166755542e-05,
+      "loss": 0.6994,
+      "step": 5485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6722061613911446e-05,
+      "loss": 0.7042,
+      "step": 5486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6720171343148954e-05,
+      "loss": 0.7584,
+      "step": 5487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6718280855354864e-05,
+      "loss": 0.6309,
+      "step": 5488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6716390150616103e-05,
+      "loss": 0.6758,
+      "step": 5489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6714499229019597e-05,
+      "loss": 0.6817,
+      "step": 5490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6712608090652294e-05,
+      "loss": 0.7118,
+      "step": 5491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6710716735601145e-05,
+      "loss": 0.7167,
+      "step": 5492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6708825163953117e-05,
+      "loss": 0.7458,
+      "step": 5493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6706933375795184e-05,
+      "loss": 0.7468,
+      "step": 5494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6705041371214326e-05,
+      "loss": 0.715,
+      "step": 5495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.670314915029754e-05,
+      "loss": 0.7532,
+      "step": 5496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6701256713131826e-05,
+      "loss": 0.7757,
+      "step": 5497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6699364059804202e-05,
+      "loss": 0.7521,
+      "step": 5498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6697471190401688e-05,
+      "loss": 0.7043,
+      "step": 5499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.669557810501132e-05,
+      "loss": 0.7583,
+      "step": 5500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.669368480372014e-05,
+      "loss": 0.7346,
+      "step": 5501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6691791286615197e-05,
+      "loss": 0.6815,
+      "step": 5502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6689897553783564e-05,
+      "loss": 0.6929,
+      "step": 5503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6688003605312304e-05,
+      "loss": 0.6939,
+      "step": 5504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6686109441288506e-05,
+      "loss": 0.6686,
+      "step": 5505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6684215061799263e-05,
+      "loss": 0.7365,
+      "step": 5506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6682320466931672e-05,
+      "loss": 0.7343,
+      "step": 5507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6680425656772856e-05,
+      "loss": 0.7524,
+      "step": 5508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6678530631409928e-05,
+      "loss": 0.7112,
+      "step": 5509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6676635390930023e-05,
+      "loss": 0.7293,
+      "step": 5510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6674739935420285e-05,
+      "loss": 0.7327,
+      "step": 5511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.667284426496787e-05,
+      "loss": 0.5857,
+      "step": 5512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6670948379659928e-05,
+      "loss": 0.7247,
+      "step": 5513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.666905227958364e-05,
+      "loss": 0.6933,
+      "step": 5514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.666715596482619e-05,
+      "loss": 0.7034,
+      "step": 5515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6665259435474765e-05,
+      "loss": 0.6612,
+      "step": 5516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.666336269161657e-05,
+      "loss": 0.6863,
+      "step": 5517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6661465733338815e-05,
+      "loss": 0.6952,
+      "step": 5518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6659568560728718e-05,
+      "loss": 0.693,
+      "step": 5519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6657671173873518e-05,
+      "loss": 0.661,
+      "step": 5520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6655773572860448e-05,
+      "loss": 0.661,
+      "step": 5521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6653875757776764e-05,
+      "loss": 0.658,
+      "step": 5522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6651977728709725e-05,
+      "loss": 0.6761,
+      "step": 5523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.66500794857466e-05,
+      "loss": 0.7302,
+      "step": 5524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6648181028974675e-05,
+      "loss": 0.6919,
+      "step": 5525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6646282358481234e-05,
+      "loss": 0.6889,
+      "step": 5526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.664438347435358e-05,
+      "loss": 0.6677,
+      "step": 5527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6642484376679023e-05,
+      "loss": 0.7136,
+      "step": 5528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6640585065544883e-05,
+      "loss": 0.7362,
+      "step": 5529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.663868554103849e-05,
+      "loss": 0.7338,
+      "step": 5530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6636785803247177e-05,
+      "loss": 0.7018,
+      "step": 5531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.66348858522583e-05,
+      "loss": 0.6766,
+      "step": 5532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6632985688159223e-05,
+      "loss": 0.7078,
+      "step": 5533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.66310853110373e-05,
+      "loss": 0.686,
+      "step": 5534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6629184720979926e-05,
+      "loss": 0.7041,
+      "step": 5535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6627283918074474e-05,
+      "loss": 0.6757,
+      "step": 5536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6625382902408356e-05,
+      "loss": 0.6963,
+      "step": 5537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.662348167406897e-05,
+      "loss": 0.7676,
+      "step": 5538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6621580233143734e-05,
+      "loss": 0.7125,
+      "step": 5539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6619678579720082e-05,
+      "loss": 0.7212,
+      "step": 5540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.661777671388545e-05,
+      "loss": 0.7257,
+      "step": 5541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.661587463572728e-05,
+      "loss": 0.7639,
+      "step": 5542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6613972345333033e-05,
+      "loss": 0.662,
+      "step": 5543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6612069842790174e-05,
+      "loss": 0.6523,
+      "step": 5544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6610167128186177e-05,
+      "loss": 0.7063,
+      "step": 5545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6608264201608532e-05,
+      "loss": 0.7195,
+      "step": 5546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6606361063144735e-05,
+      "loss": 0.6812,
+      "step": 5547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6604457712882287e-05,
+      "loss": 0.6919,
+      "step": 5548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6602554150908708e-05,
+      "loss": 0.6766,
+      "step": 5549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6600650377311523e-05,
+      "loss": 0.7011,
+      "step": 5550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6598746392178264e-05,
+      "loss": 0.6765,
+      "step": 5551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6596842195596477e-05,
+      "loss": 0.6379,
+      "step": 5552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6594937787653712e-05,
+      "loss": 0.5797,
+      "step": 5553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6593033168437542e-05,
+      "loss": 0.6904,
+      "step": 5554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6591128338035533e-05,
+      "loss": 0.7211,
+      "step": 5555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6589223296535267e-05,
+      "loss": 0.5879,
+      "step": 5556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6587318044024346e-05,
+      "loss": 0.6902,
+      "step": 5557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6585412580590363e-05,
+      "loss": 0.6249,
+      "step": 5558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.658350690632094e-05,
+      "loss": 0.7195,
+      "step": 5559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6581601021303692e-05,
+      "loss": 0.6873,
+      "step": 5560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6579694925626255e-05,
+      "loss": 0.6611,
+      "step": 5561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.657778861937626e-05,
+      "loss": 0.7116,
+      "step": 5562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.657588210264137e-05,
+      "loss": 0.6881,
+      "step": 5563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6573975375509247e-05,
+      "loss": 0.6646,
+      "step": 5564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6572068438067554e-05,
+      "loss": 0.7481,
+      "step": 5565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6570161290403975e-05,
+      "loss": 0.759,
+      "step": 5566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.65682539326062e-05,
+      "loss": 0.6883,
+      "step": 5567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6566346364761925e-05,
+      "loss": 0.6934,
+      "step": 5568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6564438586958864e-05,
+      "loss": 0.7055,
+      "step": 5569
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6562530599284736e-05,
+      "loss": 0.7604,
+      "step": 5570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6560622401827262e-05,
+      "loss": 0.6867,
+      "step": 5571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.655871399467419e-05,
+      "loss": 0.6881,
+      "step": 5572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.655680537791326e-05,
+      "loss": 0.6802,
+      "step": 5573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6554896551632235e-05,
+      "loss": 0.6584,
+      "step": 5574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6552987515918874e-05,
+      "loss": 0.5998,
+      "step": 5575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6551078270860965e-05,
+      "loss": 0.7308,
+      "step": 5576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.654916881654629e-05,
+      "loss": 0.6972,
+      "step": 5577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.654725915306264e-05,
+      "loss": 0.7661,
+      "step": 5578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6545349280497827e-05,
+      "loss": 0.575,
+      "step": 5579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.654343919893966e-05,
+      "loss": 0.7507,
+      "step": 5580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.654152890847597e-05,
+      "loss": 0.7454,
+      "step": 5581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6539618409194586e-05,
+      "loss": 0.6654,
+      "step": 5582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6537707701183355e-05,
+      "loss": 0.6113,
+      "step": 5583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6535796784530133e-05,
+      "loss": 0.7493,
+      "step": 5584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.653388565932278e-05,
+      "loss": 0.7155,
+      "step": 5585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.653197432564917e-05,
+      "loss": 0.6312,
+      "step": 5586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6530062783597185e-05,
+      "loss": 0.7777,
+      "step": 5587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.652815103325471e-05,
+      "loss": 0.7207,
+      "step": 5588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6526239074709662e-05,
+      "loss": 0.7733,
+      "step": 5589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6524326908049936e-05,
+      "loss": 0.7412,
+      "step": 5590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.652241453336346e-05,
+      "loss": 0.646,
+      "step": 5591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6520501950738167e-05,
+      "loss": 0.6778,
+      "step": 5592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6518589160261996e-05,
+      "loss": 0.735,
+      "step": 5593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.651667616202289e-05,
+      "loss": 0.6739,
+      "step": 5594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6514762956108815e-05,
+      "loss": 0.6482,
+      "step": 5595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6512849542607733e-05,
+      "loss": 0.706,
+      "step": 5596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.651093592160763e-05,
+      "loss": 0.6418,
+      "step": 5597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6509022093196485e-05,
+      "loss": 0.6955,
+      "step": 5598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6507108057462297e-05,
+      "loss": 0.7284,
+      "step": 5599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.650519381449308e-05,
+      "loss": 0.6707,
+      "step": 5600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6503279364376844e-05,
+      "loss": 0.7925,
+      "step": 5601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6501364707201613e-05,
+      "loss": 0.6629,
+      "step": 5602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6499449843055427e-05,
+      "loss": 0.7065,
+      "step": 5603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6497534772026325e-05,
+      "loss": 0.6403,
+      "step": 5604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.649561949420237e-05,
+      "loss": 0.7202,
+      "step": 5605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6493704009671614e-05,
+      "loss": 0.7258,
+      "step": 5606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.649178831852214e-05,
+      "loss": 0.7035,
+      "step": 5607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6489872420842028e-05,
+      "loss": 0.6602,
+      "step": 5608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6487956316719366e-05,
+      "loss": 0.7428,
+      "step": 5609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.648604000624226e-05,
+      "loss": 0.7329,
+      "step": 5610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6484123489498822e-05,
+      "loss": 0.6932,
+      "step": 5611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6482206766577173e-05,
+      "loss": 0.6225,
+      "step": 5612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.648028983756544e-05,
+      "loss": 0.6259,
+      "step": 5613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6478372702551757e-05,
+      "loss": 0.6509,
+      "step": 5614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6476455361624285e-05,
+      "loss": 0.6619,
+      "step": 5615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6474537814871178e-05,
+      "loss": 0.7396,
+      "step": 5616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6472620062380602e-05,
+      "loss": 0.6972,
+      "step": 5617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.647070210424074e-05,
+      "loss": 0.6652,
+      "step": 5618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.646878394053977e-05,
+      "loss": 0.6066,
+      "step": 5619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6466865571365896e-05,
+      "loss": 0.7393,
+      "step": 5620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.646494699680732e-05,
+      "loss": 0.6749,
+      "step": 5621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.646302821695226e-05,
+      "loss": 0.6883,
+      "step": 5622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6461109231888936e-05,
+      "loss": 0.718,
+      "step": 5623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6459190041705587e-05,
+      "loss": 0.6993,
+      "step": 5624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6457270646490457e-05,
+      "loss": 0.667,
+      "step": 5625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6455351046331792e-05,
+      "loss": 0.6816,
+      "step": 5626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6453431241317863e-05,
+      "loss": 0.7355,
+      "step": 5627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.645151123153694e-05,
+      "loss": 0.6955,
+      "step": 5628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.64495910170773e-05,
+      "loss": 0.7123,
+      "step": 5629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6447670598027236e-05,
+      "loss": 0.6869,
+      "step": 5630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.644574997447505e-05,
+      "loss": 0.619,
+      "step": 5631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6443829146509045e-05,
+      "loss": 0.6894,
+      "step": 5632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6441908114217552e-05,
+      "loss": 0.675,
+      "step": 5633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.643998687768889e-05,
+      "loss": 0.6551,
+      "step": 5634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6438065437011398e-05,
+      "loss": 0.6979,
+      "step": 5635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.643614379227342e-05,
+      "loss": 0.6525,
+      "step": 5636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6434221943563322e-05,
+      "loss": 0.7195,
+      "step": 5637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.643229989096946e-05,
+      "loss": 0.7545,
+      "step": 5638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6430377634580216e-05,
+      "loss": 0.6854,
+      "step": 5639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6428455174483973e-05,
+      "loss": 0.6611,
+      "step": 5640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.642653251076912e-05,
+      "loss": 0.6797,
+      "step": 5641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.642460964352407e-05,
+      "loss": 0.6871,
+      "step": 5642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6422686572837225e-05,
+      "loss": 0.7599,
+      "step": 5643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6420763298797014e-05,
+      "loss": 0.7369,
+      "step": 5644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6418839821491864e-05,
+      "loss": 0.6461,
+      "step": 5645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6416916141010222e-05,
+      "loss": 0.6834,
+      "step": 5646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.641499225744053e-05,
+      "loss": 0.7205,
+      "step": 5647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.641306817087125e-05,
+      "loss": 0.7013,
+      "step": 5648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6411143881390857e-05,
+      "loss": 0.7149,
+      "step": 5649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6409219389087824e-05,
+      "loss": 0.6608,
+      "step": 5650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6407294694050637e-05,
+      "loss": 0.7302,
+      "step": 5651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6405369796367796e-05,
+      "loss": 0.7345,
+      "step": 5652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6403444696127804e-05,
+      "loss": 0.6559,
+      "step": 5653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6401519393419178e-05,
+      "loss": 0.6801,
+      "step": 5654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6399593888330444e-05,
+      "loss": 0.6882,
+      "step": 5655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.639766818095013e-05,
+      "loss": 0.6747,
+      "step": 5656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.639574227136679e-05,
+      "loss": 0.6446,
+      "step": 5657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6393816159668962e-05,
+      "loss": 0.6746,
+      "step": 5658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6391889845945223e-05,
+      "loss": 0.6984,
+      "step": 5659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6389963330284132e-05,
+      "loss": 0.7229,
+      "step": 5660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.638803661277428e-05,
+      "loss": 0.6761,
+      "step": 5661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6386109693504248e-05,
+      "loss": 0.6795,
+      "step": 5662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6384182572562643e-05,
+      "loss": 0.6878,
+      "step": 5663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6382255250038062e-05,
+      "loss": 0.6915,
+      "step": 5664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.638032772601913e-05,
+      "loss": 0.6823,
+      "step": 5665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6378400000594474e-05,
+      "loss": 0.6654,
+      "step": 5666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6376472073852733e-05,
+      "loss": 0.6644,
+      "step": 5667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.637454394588254e-05,
+      "loss": 0.7086,
+      "step": 5668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6372615616772564e-05,
+      "loss": 0.6661,
+      "step": 5669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.637068708661146e-05,
+      "loss": 0.6657,
+      "step": 5670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6368758355487907e-05,
+      "loss": 0.6962,
+      "step": 5671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.636682942349058e-05,
+      "loss": 0.7152,
+      "step": 5672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6364900290708175e-05,
+      "loss": 0.7006,
+      "step": 5673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6362970957229393e-05,
+      "loss": 0.7047,
+      "step": 5674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6361041423142942e-05,
+      "loss": 0.7306,
+      "step": 5675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6359111688537546e-05,
+      "loss": 0.6632,
+      "step": 5676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6357181753501927e-05,
+      "loss": 0.7753,
+      "step": 5677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6355251618124826e-05,
+      "loss": 0.7116,
+      "step": 5678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6353321282494987e-05,
+      "loss": 0.7133,
+      "step": 5679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6351390746701173e-05,
+      "loss": 0.7588,
+      "step": 5680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6349460010832142e-05,
+      "loss": 0.7481,
+      "step": 5681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.634752907497667e-05,
+      "loss": 0.7346,
+      "step": 5682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.634559793922354e-05,
+      "loss": 0.6847,
+      "step": 5683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.634366660366155e-05,
+      "loss": 0.6608,
+      "step": 5684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6341735068379498e-05,
+      "loss": 0.6326,
+      "step": 5685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.63398033334662e-05,
+      "loss": 0.7511,
+      "step": 5686
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6337871399010468e-05,
+      "loss": 0.6382,
+      "step": 5687
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6335939265101133e-05,
+      "loss": 0.7342,
+      "step": 5688
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.633400693182704e-05,
+      "loss": 0.6437,
+      "step": 5689
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.633207439927703e-05,
+      "loss": 0.6619,
+      "step": 5690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.633014166753997e-05,
+      "loss": 0.7061,
+      "step": 5691
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6328208736704716e-05,
+      "loss": 0.7205,
+      "step": 5692
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6326275606860148e-05,
+      "loss": 0.7115,
+      "step": 5693
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.632434227809515e-05,
+      "loss": 0.6982,
+      "step": 5694
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6322408750498615e-05,
+      "loss": 0.7066,
+      "step": 5695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6320475024159445e-05,
+      "loss": 0.6842,
+      "step": 5696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6318541099166557e-05,
+      "loss": 0.739,
+      "step": 5697
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6316606975608864e-05,
+      "loss": 0.6558,
+      "step": 5698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6314672653575303e-05,
+      "loss": 0.6379,
+      "step": 5699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.631273813315481e-05,
+      "loss": 0.6839,
+      "step": 5700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6310803414436336e-05,
+      "loss": 0.7897,
+      "step": 5701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6308868497508835e-05,
+      "loss": 0.6781,
+      "step": 5702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6306933382461276e-05,
+      "loss": 0.6187,
+      "step": 5703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6304998069382636e-05,
+      "loss": 0.7406,
+      "step": 5704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.63030625583619e-05,
+      "loss": 0.6188,
+      "step": 5705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6301126849488055e-05,
+      "loss": 0.7307,
+      "step": 5706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.629919094285011e-05,
+      "loss": 0.6476,
+      "step": 5707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6297254838537083e-05,
+      "loss": 0.6554,
+      "step": 5708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6295318536637983e-05,
+      "loss": 0.6339,
+      "step": 5709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6293382037241853e-05,
+      "loss": 0.6527,
+      "step": 5710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.629144534043772e-05,
+      "loss": 0.7369,
+      "step": 5711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6289508446314642e-05,
+      "loss": 0.668,
+      "step": 5712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6287571354961673e-05,
+      "loss": 0.7166,
+      "step": 5713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6285634066467877e-05,
+      "loss": 0.6261,
+      "step": 5714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6283696580922335e-05,
+      "loss": 0.6754,
+      "step": 5715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6281758898414127e-05,
+      "loss": 0.6793,
+      "step": 5716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.627982101903235e-05,
+      "loss": 0.6896,
+      "step": 5717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.627788294286611e-05,
+      "loss": 0.7252,
+      "step": 5718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6275944670004513e-05,
+      "loss": 0.7099,
+      "step": 5719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.627400620053668e-05,
+      "loss": 0.6887,
+      "step": 5720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6272067534551745e-05,
+      "loss": 0.7073,
+      "step": 5721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6270128672138847e-05,
+      "loss": 0.7627,
+      "step": 5722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.626818961338713e-05,
+      "loss": 0.7153,
+      "step": 5723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6266250358385752e-05,
+      "loss": 0.6688,
+      "step": 5724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6264310907223885e-05,
+      "loss": 0.6038,
+      "step": 5725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6262371259990698e-05,
+      "loss": 0.7374,
+      "step": 5726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6260431416775378e-05,
+      "loss": 0.706,
+      "step": 5727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6258491377667118e-05,
+      "loss": 0.6541,
+      "step": 5728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6256551142755117e-05,
+      "loss": 0.6652,
+      "step": 5729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.625461071212859e-05,
+      "loss": 0.6577,
+      "step": 5730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.625267008587676e-05,
+      "loss": 0.7453,
+      "step": 5731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6250729264088847e-05,
+      "loss": 0.7312,
+      "step": 5732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6248788246854095e-05,
+      "loss": 0.7593,
+      "step": 5733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.624684703426175e-05,
+      "loss": 0.7688,
+      "step": 5734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6244905626401073e-05,
+      "loss": 0.6643,
+      "step": 5735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6242964023361323e-05,
+      "loss": 0.7206,
+      "step": 5736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6241022225231776e-05,
+      "loss": 0.6506,
+      "step": 5737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6239080232101715e-05,
+      "loss": 0.6998,
+      "step": 5738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6237138044060433e-05,
+      "loss": 0.7696,
+      "step": 5739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.623519566119723e-05,
+      "loss": 0.7006,
+      "step": 5740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6233253083601414e-05,
+      "loss": 0.7003,
+      "step": 5741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6231310311362314e-05,
+      "loss": 0.7173,
+      "step": 5742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6229367344569243e-05,
+      "loss": 0.6455,
+      "step": 5743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6227424183311545e-05,
+      "loss": 0.6626,
+      "step": 5744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.622548082767857e-05,
+      "loss": 0.726,
+      "step": 5745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6223537277759667e-05,
+      "loss": 0.6797,
+      "step": 5746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.62215935336442e-05,
+      "loss": 0.6741,
+      "step": 5747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6219649595421543e-05,
+      "loss": 0.6603,
+      "step": 5748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6217705463181078e-05,
+      "loss": 0.7144,
+      "step": 5749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6215761137012195e-05,
+      "loss": 0.6494,
+      "step": 5750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6213816617004295e-05,
+      "loss": 0.7118,
+      "step": 5751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.621187190324678e-05,
+      "loss": 0.7115,
+      "step": 5752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6209926995829077e-05,
+      "loss": 0.6807,
+      "step": 5753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6207981894840604e-05,
+      "loss": 0.6766,
+      "step": 5754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6206036600370796e-05,
+      "loss": 0.7074,
+      "step": 5755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.62040911125091e-05,
+      "loss": 0.7026,
+      "step": 5756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.620214543134497e-05,
+      "loss": 0.6982,
+      "step": 5757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6200199556967868e-05,
+      "loss": 0.6601,
+      "step": 5758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.619825348946726e-05,
+      "loss": 0.6258,
+      "step": 5759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6196307228932628e-05,
+      "loss": 0.7102,
+      "step": 5760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6194360775453458e-05,
+      "loss": 0.6665,
+      "step": 5761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.619241412911925e-05,
+      "loss": 0.6821,
+      "step": 5762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.619046729001951e-05,
+      "loss": 0.7101,
+      "step": 5763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6188520258243755e-05,
+      "loss": 0.6376,
+      "step": 5764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.61865730338815e-05,
+      "loss": 0.6986,
+      "step": 5765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6184625617022285e-05,
+      "loss": 0.7813,
+      "step": 5766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6182678007755654e-05,
+      "loss": 0.5435,
+      "step": 5767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.618073020617115e-05,
+      "loss": 0.6962,
+      "step": 5768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6178782212358335e-05,
+      "loss": 0.6618,
+      "step": 5769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6176834026406778e-05,
+      "loss": 0.7073,
+      "step": 5770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6174885648406058e-05,
+      "loss": 0.6948,
+      "step": 5771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6172937078445754e-05,
+      "loss": 0.7164,
+      "step": 5772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6170988316615466e-05,
+      "loss": 0.6615,
+      "step": 5773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6169039363004796e-05,
+      "loss": 0.6774,
+      "step": 5774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6167090217703353e-05,
+      "loss": 0.7162,
+      "step": 5775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6165140880800764e-05,
+      "loss": 0.6799,
+      "step": 5776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.616319135238665e-05,
+      "loss": 0.7568,
+      "step": 5777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.616124163255066e-05,
+      "loss": 0.7346,
+      "step": 5778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6159291721382435e-05,
+      "loss": 0.7588,
+      "step": 5779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.615734161897163e-05,
+      "loss": 0.7061,
+      "step": 5780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6155391325407917e-05,
+      "loss": 0.7497,
+      "step": 5781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6153440840780963e-05,
+      "loss": 0.7079,
+      "step": 5782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.615149016518045e-05,
+      "loss": 0.7354,
+      "step": 5783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6149539298696074e-05,
+      "loss": 0.7123,
+      "step": 5784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.614758824141753e-05,
+      "loss": 0.6574,
+      "step": 5785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6145636993434534e-05,
+      "loss": 0.691,
+      "step": 5786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6143685554836797e-05,
+      "loss": 0.6354,
+      "step": 5787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6141733925714046e-05,
+      "loss": 0.6475,
+      "step": 5788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6139782106156022e-05,
+      "loss": 0.6769,
+      "step": 5789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6137830096252462e-05,
+      "loss": 0.7292,
+      "step": 5790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6135877896093122e-05,
+      "loss": 0.754,
+      "step": 5791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6133925505767764e-05,
+      "loss": 0.6373,
+      "step": 5792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6131972925366155e-05,
+      "loss": 0.6641,
+      "step": 5793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6130020154978075e-05,
+      "loss": 0.6154,
+      "step": 5794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6128067194693317e-05,
+      "loss": 0.7659,
+      "step": 5795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6126114044601668e-05,
+      "loss": 0.6749,
+      "step": 5796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.612416070479294e-05,
+      "loss": 0.7161,
+      "step": 5797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.612220717535694e-05,
+      "loss": 0.7238,
+      "step": 5798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6120253456383497e-05,
+      "loss": 0.714,
+      "step": 5799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.611829954796244e-05,
+      "loss": 0.7455,
+      "step": 5800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6116345450183612e-05,
+      "loss": 0.6277,
+      "step": 5801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.611439116313686e-05,
+      "loss": 0.755,
+      "step": 5802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6112436686912038e-05,
+      "loss": 0.6418,
+      "step": 5803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6110482021599013e-05,
+      "loss": 0.7344,
+      "step": 5804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6108527167287663e-05,
+      "loss": 0.6848,
+      "step": 5805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6106572124067866e-05,
+      "loss": 0.7423,
+      "step": 5806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.610461689202952e-05,
+      "loss": 0.6654,
+      "step": 5807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6102661471262522e-05,
+      "loss": 0.6926,
+      "step": 5808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6100705861856783e-05,
+      "loss": 0.6955,
+      "step": 5809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.609875006390222e-05,
+      "loss": 0.7365,
+      "step": 5810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6096794077488764e-05,
+      "loss": 0.6674,
+      "step": 5811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6094837902706348e-05,
+      "loss": 0.6796,
+      "step": 5812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.609288153964491e-05,
+      "loss": 0.7167,
+      "step": 5813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6090924988394412e-05,
+      "loss": 0.7043,
+      "step": 5814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6088968249044807e-05,
+      "loss": 0.6695,
+      "step": 5815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.608701132168607e-05,
+      "loss": 0.6422,
+      "step": 5816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6085054206408184e-05,
+      "loss": 0.7147,
+      "step": 5817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6083096903301133e-05,
+      "loss": 0.714,
+      "step": 5818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6081139412454905e-05,
+      "loss": 0.7265,
+      "step": 5819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.607918173395952e-05,
+      "loss": 0.6654,
+      "step": 5820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6077223867904973e-05,
+      "loss": 0.7027,
+      "step": 5821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.60752658143813e-05,
+      "loss": 0.7143,
+      "step": 5822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6073307573478527e-05,
+      "loss": 0.7058,
+      "step": 5823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6071349145286697e-05,
+      "loss": 0.6952,
+      "step": 5824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6069390529895848e-05,
+      "loss": 0.6794,
+      "step": 5825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6067431727396045e-05,
+      "loss": 0.787,
+      "step": 5826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.606547273787735e-05,
+      "loss": 0.6439,
+      "step": 5827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6063513561429834e-05,
+      "loss": 0.6519,
+      "step": 5828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6061554198143588e-05,
+      "loss": 0.6917,
+      "step": 5829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.605959464810869e-05,
+      "loss": 0.6626,
+      "step": 5830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6057634911415254e-05,
+      "loss": 0.7357,
+      "step": 5831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6055674988153378e-05,
+      "loss": 0.7246,
+      "step": 5832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6053714878413173e-05,
+      "loss": 0.7594,
+      "step": 5833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.605175458228478e-05,
+      "loss": 0.6883,
+      "step": 5834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.604979409985832e-05,
+      "loss": 0.681,
+      "step": 5835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6047833431223937e-05,
+      "loss": 0.7363,
+      "step": 5836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.604587257647179e-05,
+      "loss": 0.6705,
+      "step": 5837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.604391153569203e-05,
+      "loss": 0.6614,
+      "step": 5838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6041950308974827e-05,
+      "loss": 0.6027,
+      "step": 5839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.603998889641036e-05,
+      "loss": 0.6265,
+      "step": 5840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.603802729808881e-05,
+      "loss": 0.7233,
+      "step": 5841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6036065514100374e-05,
+      "loss": 0.6495,
+      "step": 5842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6034103544535247e-05,
+      "loss": 0.6833,
+      "step": 5843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.603214138948365e-05,
+      "loss": 0.7003,
+      "step": 5844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6030179049035795e-05,
+      "loss": 0.7161,
+      "step": 5845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6028216523281912e-05,
+      "loss": 0.7903,
+      "step": 5846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6026253812312237e-05,
+      "loss": 0.7139,
+      "step": 5847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6024290916217014e-05,
+      "loss": 0.6211,
+      "step": 5848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6022327835086497e-05,
+      "loss": 0.7588,
+      "step": 5849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6020364569010945e-05,
+      "loss": 0.6594,
+      "step": 5850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6018401118080633e-05,
+      "loss": 0.6828,
+      "step": 5851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6016437482385835e-05,
+      "loss": 0.6656,
+      "step": 5852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6014473662016845e-05,
+      "loss": 0.6871,
+      "step": 5853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.601250965706395e-05,
+      "loss": 0.6265,
+      "step": 5854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.601054546761746e-05,
+      "loss": 0.6742,
+      "step": 5855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6008581093767685e-05,
+      "loss": 0.6785,
+      "step": 5856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6006616535604947e-05,
+      "loss": 0.7243,
+      "step": 5857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6004651793219574e-05,
+      "loss": 0.6517,
+      "step": 5858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6002686866701913e-05,
+      "loss": 0.6457,
+      "step": 5859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6000721756142295e-05,
+      "loss": 0.6753,
+      "step": 5860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5998756461631086e-05,
+      "loss": 0.7149,
+      "step": 5861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.599679098325865e-05,
+      "loss": 0.7228,
+      "step": 5862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.599482532111535e-05,
+      "loss": 0.7183,
+      "step": 5863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5992859475291574e-05,
+      "loss": 0.7138,
+      "step": 5864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.599089344587771e-05,
+      "loss": 0.6381,
+      "step": 5865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5988927232964153e-05,
+      "loss": 0.6003,
+      "step": 5866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5986960836641312e-05,
+      "loss": 0.6795,
+      "step": 5867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5984994256999595e-05,
+      "loss": 0.633,
+      "step": 5868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.598302749412943e-05,
+      "loss": 0.6407,
+      "step": 5869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5981060548121247e-05,
+      "loss": 0.6821,
+      "step": 5870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5979093419065483e-05,
+      "loss": 0.684,
+      "step": 5871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.597712610705259e-05,
+      "loss": 0.719,
+      "step": 5872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.597515861217302e-05,
+      "loss": 0.6984,
+      "step": 5873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5973190934517238e-05,
+      "loss": 0.6648,
+      "step": 5874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.597122307417572e-05,
+      "loss": 0.7423,
+      "step": 5875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5969255031238948e-05,
+      "loss": 0.6686,
+      "step": 5876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5967286805797405e-05,
+      "loss": 0.7155,
+      "step": 5877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.59653183979416e-05,
+      "loss": 0.7016,
+      "step": 5878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.596334980776203e-05,
+      "loss": 0.6517,
+      "step": 5879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5961381035349217e-05,
+      "loss": 0.6784,
+      "step": 5880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5959412080793676e-05,
+      "loss": 0.684,
+      "step": 5881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5957442944185945e-05,
+      "loss": 0.7046,
+      "step": 5882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5955473625616562e-05,
+      "loss": 0.6861,
+      "step": 5883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5953504125176077e-05,
+      "loss": 0.6697,
+      "step": 5884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5951534442955047e-05,
+      "loss": 0.6626,
+      "step": 5885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5949564579044037e-05,
+      "loss": 0.7029,
+      "step": 5886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.594759453353362e-05,
+      "loss": 0.6761,
+      "step": 5887
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5945624306514377e-05,
+      "loss": 0.7421,
+      "step": 5888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5943653898076897e-05,
+      "loss": 0.622,
+      "step": 5889
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5941683308311784e-05,
+      "loss": 0.7095,
+      "step": 5890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5939712537309636e-05,
+      "loss": 0.6847,
+      "step": 5891
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5937741585161077e-05,
+      "loss": 0.6892,
+      "step": 5892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5935770451956732e-05,
+      "loss": 0.6931,
+      "step": 5893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.593379913778722e-05,
+      "loss": 0.6596,
+      "step": 5894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5931827642743196e-05,
+      "loss": 0.6846,
+      "step": 5895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.59298559669153e-05,
+      "loss": 0.7172,
+      "step": 5896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5927884110394194e-05,
+      "loss": 0.7443,
+      "step": 5897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5925912073270534e-05,
+      "loss": 0.6792,
+      "step": 5898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5923939855635006e-05,
+      "loss": 0.6731,
+      "step": 5899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.592196745757828e-05,
+      "loss": 0.7158,
+      "step": 5900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5919994879191055e-05,
+      "loss": 0.6778,
+      "step": 5901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5918022120564023e-05,
+      "loss": 0.6705,
+      "step": 5902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5916049181787896e-05,
+      "loss": 0.7225,
+      "step": 5903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5914076062953384e-05,
+      "loss": 0.6555,
+      "step": 5904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5912102764151216e-05,
+      "loss": 0.6415,
+      "step": 5905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5910129285472117e-05,
+      "loss": 0.6947,
+      "step": 5906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5908155627006834e-05,
+      "loss": 0.6901,
+      "step": 5907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.590618178884611e-05,
+      "loss": 0.6796,
+      "step": 5908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5904207771080697e-05,
+      "loss": 0.7569,
+      "step": 5909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.590223357380137e-05,
+      "loss": 0.634,
+      "step": 5910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5900259197098897e-05,
+      "loss": 0.6599,
+      "step": 5911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5898284641064058e-05,
+      "loss": 0.7239,
+      "step": 5912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.589630990578764e-05,
+      "loss": 0.7228,
+      "step": 5913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5894334991360447e-05,
+      "loss": 0.7264,
+      "step": 5914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.589235989787328e-05,
+      "loss": 0.7516,
+      "step": 5915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5890384625416954e-05,
+      "loss": 0.7259,
+      "step": 5916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5888409174082292e-05,
+      "loss": 0.7342,
+      "step": 5917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.588643354396012e-05,
+      "loss": 0.6853,
+      "step": 5918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5884457735141287e-05,
+      "loss": 0.6966,
+      "step": 5919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5882481747716634e-05,
+      "loss": 0.7416,
+      "step": 5920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5880505581777013e-05,
+      "loss": 0.6564,
+      "step": 5921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.587852923741329e-05,
+      "loss": 0.673,
+      "step": 5922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5876552714716337e-05,
+      "loss": 0.6875,
+      "step": 5923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.587457601377703e-05,
+      "loss": 0.7095,
+      "step": 5924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5872599134686263e-05,
+      "loss": 0.7177,
+      "step": 5925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.587062207753493e-05,
+      "loss": 0.7359,
+      "step": 5926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5868644842413937e-05,
+      "loss": 0.6769,
+      "step": 5927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.586666742941419e-05,
+      "loss": 0.6911,
+      "step": 5928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5864689838626617e-05,
+      "loss": 0.7111,
+      "step": 5929
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.586271207014214e-05,
+      "loss": 0.6852,
+      "step": 5930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5860734124051704e-05,
+      "loss": 0.7344,
+      "step": 5931
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5858756000446245e-05,
+      "loss": 0.746,
+      "step": 5932
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5856777699416725e-05,
+      "loss": 0.6706,
+      "step": 5933
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5854799221054095e-05,
+      "loss": 0.6417,
+      "step": 5934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5852820565449343e-05,
+      "loss": 0.7013,
+      "step": 5935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5850841732693426e-05,
+      "loss": 0.6932,
+      "step": 5936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.584886272287734e-05,
+      "loss": 0.6915,
+      "step": 5937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.584688353609208e-05,
+      "loss": 0.6639,
+      "step": 5938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.584490417242865e-05,
+      "loss": 0.6526,
+      "step": 5939
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.584292463197805e-05,
+      "loss": 0.7787,
+      "step": 5940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.584094491483131e-05,
+      "loss": 0.6076,
+      "step": 5941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5838965021079448e-05,
+      "loss": 0.6526,
+      "step": 5942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5836984950813505e-05,
+      "loss": 0.6991,
+      "step": 5943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.583500470412452e-05,
+      "loss": 0.69,
+      "step": 5944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5833024281103545e-05,
+      "loss": 0.7827,
+      "step": 5945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.583104368184164e-05,
+      "loss": 0.6578,
+      "step": 5946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5829062906429872e-05,
+      "loss": 0.6536,
+      "step": 5947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5827081954959315e-05,
+      "loss": 0.7364,
+      "step": 5948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5825100827521057e-05,
+      "loss": 0.6624,
+      "step": 5949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.582311952420618e-05,
+      "loss": 0.6243,
+      "step": 5950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5821138045105793e-05,
+      "loss": 0.6724,
+      "step": 5951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5819156390311e-05,
+      "loss": 0.6842,
+      "step": 5952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5817174559912912e-05,
+      "loss": 0.6821,
+      "step": 5953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.581519255400266e-05,
+      "loss": 0.6892,
+      "step": 5954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5813210372671373e-05,
+      "loss": 0.7163,
+      "step": 5955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5811228016010192e-05,
+      "loss": 0.6755,
+      "step": 5956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5809245484110264e-05,
+      "loss": 0.7101,
+      "step": 5957
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5807262777062743e-05,
+      "loss": 0.7268,
+      "step": 5958
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5805279894958795e-05,
+      "loss": 0.7057,
+      "step": 5959
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5803296837889593e-05,
+      "loss": 0.7412,
+      "step": 5960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.580131360594632e-05,
+      "loss": 0.717,
+      "step": 5961
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5799330199220155e-05,
+      "loss": 0.6553,
+      "step": 5962
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5797346617802302e-05,
+      "loss": 0.6627,
+      "step": 5963
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5795362861783962e-05,
+      "loss": 0.7418,
+      "step": 5964
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.579337893125635e-05,
+      "loss": 0.7323,
+      "step": 5965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5791394826310686e-05,
+      "loss": 0.678,
+      "step": 5966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5789410547038196e-05,
+      "loss": 0.6671,
+      "step": 5967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5787426093530115e-05,
+      "loss": 0.6668,
+      "step": 5968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5785441465877694e-05,
+      "loss": 0.7314,
+      "step": 5969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5783456664172176e-05,
+      "loss": 0.6484,
+      "step": 5970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5781471688504834e-05,
+      "loss": 0.6681,
+      "step": 5971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5779486538966924e-05,
+      "loss": 0.6708,
+      "step": 5972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.577750121564973e-05,
+      "loss": 0.641,
+      "step": 5973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.577551571864453e-05,
+      "loss": 0.6817,
+      "step": 5974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5773530048042624e-05,
+      "loss": 0.6735,
+      "step": 5975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5771544203935305e-05,
+      "loss": 0.6484,
+      "step": 5976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5769558186413887e-05,
+      "loss": 0.6842,
+      "step": 5977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5767571995569684e-05,
+      "loss": 0.6691,
+      "step": 5978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5765585631494022e-05,
+      "loss": 0.6734,
+      "step": 5979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.576359909427823e-05,
+      "loss": 0.6621,
+      "step": 5980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5761612384013647e-05,
+      "loss": 0.7142,
+      "step": 5981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.575962550079163e-05,
+      "loss": 0.705,
+      "step": 5982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5757638444703526e-05,
+      "loss": 0.75,
+      "step": 5983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.57556512158407e-05,
+      "loss": 0.7176,
+      "step": 5984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5753663814294528e-05,
+      "loss": 0.6175,
+      "step": 5985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.575167624015639e-05,
+      "loss": 0.7125,
+      "step": 5986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5749688493517668e-05,
+      "loss": 0.6743,
+      "step": 5987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5747700574469764e-05,
+      "loss": 0.6728,
+      "step": 5988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.574571248310408e-05,
+      "loss": 0.7147,
+      "step": 5989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5743724219512026e-05,
+      "loss": 0.7055,
+      "step": 5990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5741735783785022e-05,
+      "loss": 0.6553,
+      "step": 5991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.57397471760145e-05,
+      "loss": 0.6296,
+      "step": 5992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5737758396291888e-05,
+      "loss": 0.634,
+      "step": 5993
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5735769444708633e-05,
+      "loss": 0.7206,
+      "step": 5994
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5733780321356186e-05,
+      "loss": 0.732,
+      "step": 5995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5731791026326007e-05,
+      "loss": 0.7625,
+      "step": 5996
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5729801559709562e-05,
+      "loss": 0.6851,
+      "step": 5997
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5727811921598324e-05,
+      "loss": 0.7146,
+      "step": 5998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5725822112083782e-05,
+      "loss": 0.5894,
+      "step": 5999
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.572383213125742e-05,
+      "loss": 0.7013,
+      "step": 6000
+    },
+    {
+      "epoch": 0.4,
+      "eval_text_loss": 0.6642442345619202,
+      "eval_text_runtime": 1.7998,
+      "eval_text_samples_per_second": 41.115,
+      "eval_text_steps_per_second": 1.111,
+      "step": 6000
+    },
+    {
+      "epoch": 0.4,
+      "eval_svit_loss": 0.8306663632392883,
+      "eval_svit_runtime": 6.6143,
+      "eval_svit_samples_per_second": 123.369,
+      "eval_svit_steps_per_second": 1.965,
+      "step": 6000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.572184197921074e-05,
+      "loss": 0.6997,
+      "step": 6001
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5719851656035245e-05,
+      "loss": 0.6979,
+      "step": 6002
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.571786116182245e-05,
+      "loss": 0.6808,
+      "step": 6003
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5715870496663885e-05,
+      "loss": 0.6464,
+      "step": 6004
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.571387966065107e-05,
+      "loss": 0.6472,
+      "step": 6005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5711888653875545e-05,
+      "loss": 0.6933,
+      "step": 6006
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5709897476428862e-05,
+      "loss": 0.7138,
+      "step": 6007
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5707906128402564e-05,
+      "loss": 0.7132,
+      "step": 6008
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5705914609888225e-05,
+      "loss": 0.6779,
+      "step": 6009
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5703922920977403e-05,
+      "loss": 0.6616,
+      "step": 6010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.570193106176168e-05,
+      "loss": 0.6808,
+      "step": 6011
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.569993903233264e-05,
+      "loss": 0.6102,
+      "step": 6012
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5697946832781875e-05,
+      "loss": 0.6253,
+      "step": 6013
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.569595446320099e-05,
+      "loss": 0.7147,
+      "step": 6014
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5693961923681588e-05,
+      "loss": 0.7469,
+      "step": 6015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.569196921431529e-05,
+      "loss": 0.681,
+      "step": 6016
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5689976335193714e-05,
+      "loss": 0.6357,
+      "step": 6017
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5687983286408494e-05,
+      "loss": 0.603,
+      "step": 6018
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5685990068051272e-05,
+      "loss": 0.6934,
+      "step": 6019
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5683996680213698e-05,
+      "loss": 0.6216,
+      "step": 6020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5682003122987417e-05,
+      "loss": 0.6202,
+      "step": 6021
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.56800093964641e-05,
+      "loss": 0.7031,
+      "step": 6022
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5678015500735418e-05,
+      "loss": 0.6984,
+      "step": 6023
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.567602143589305e-05,
+      "loss": 0.7441,
+      "step": 6024
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.567402720202867e-05,
+      "loss": 0.7064,
+      "step": 6025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5672032799233986e-05,
+      "loss": 0.6358,
+      "step": 6026
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.56700382276007e-05,
+      "loss": 0.686,
+      "step": 6027
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5668043487220514e-05,
+      "loss": 0.6403,
+      "step": 6028
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.566604857818515e-05,
+      "loss": 0.6875,
+      "step": 6029
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.566405350058633e-05,
+      "loss": 0.7129,
+      "step": 6030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.566205825451579e-05,
+      "loss": 0.7642,
+      "step": 6031
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5660062840065267e-05,
+      "loss": 0.7455,
+      "step": 6032
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5658067257326514e-05,
+      "loss": 0.7043,
+      "step": 6033
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5656071506391283e-05,
+      "loss": 0.7591,
+      "step": 6034
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5654075587351343e-05,
+      "loss": 0.6757,
+      "step": 6035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5652079500298457e-05,
+      "loss": 0.6087,
+      "step": 6036
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5650083245324412e-05,
+      "loss": 0.6794,
+      "step": 6037
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.564808682252099e-05,
+      "loss": 0.6342,
+      "step": 6038
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.564609023197999e-05,
+      "loss": 0.6705,
+      "step": 6039
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5644093473793214e-05,
+      "loss": 0.7056,
+      "step": 6040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.564209654805247e-05,
+      "loss": 0.6401,
+      "step": 6041
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5640099454849578e-05,
+      "loss": 0.735,
+      "step": 6042
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.563810219427636e-05,
+      "loss": 0.6246,
+      "step": 6043
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.563610476642465e-05,
+      "loss": 0.6299,
+      "step": 6044
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.563410717138629e-05,
+      "loss": 0.7107,
+      "step": 6045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.563210940925313e-05,
+      "loss": 0.6655,
+      "step": 6046
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.563011148011703e-05,
+      "loss": 0.7108,
+      "step": 6047
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.562811338406984e-05,
+      "loss": 0.7104,
+      "step": 6048
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5626115121203446e-05,
+      "loss": 0.7284,
+      "step": 6049
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5624116691609725e-05,
+      "loss": 0.754,
+      "step": 6050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.562211809538056e-05,
+      "loss": 0.6891,
+      "step": 6051
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.562011933260784e-05,
+      "loss": 0.6356,
+      "step": 6052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.561812040338348e-05,
+      "loss": 0.6643,
+      "step": 6053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5616121307799385e-05,
+      "loss": 0.6833,
+      "step": 6054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.561412204594747e-05,
+      "loss": 0.7271,
+      "step": 6055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5612122617919662e-05,
+      "loss": 0.6734,
+      "step": 6056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.561012302380789e-05,
+      "loss": 0.6708,
+      "step": 6057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5608123263704105e-05,
+      "loss": 0.68,
+      "step": 6058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5606123337700244e-05,
+      "loss": 0.651,
+      "step": 6059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.560412324588827e-05,
+      "loss": 0.6862,
+      "step": 6060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5602122988360145e-05,
+      "loss": 0.6878,
+      "step": 6061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5600122565207837e-05,
+      "loss": 0.715,
+      "step": 6062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5598121976523326e-05,
+      "loss": 0.6716,
+      "step": 6063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.55961212223986e-05,
+      "loss": 0.6521,
+      "step": 6064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.559412030292565e-05,
+      "loss": 0.712,
+      "step": 6065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.559211921819648e-05,
+      "loss": 0.614,
+      "step": 6066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.55901179683031e-05,
+      "loss": 0.6513,
+      "step": 6067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.558811655333752e-05,
+      "loss": 0.6803,
+      "step": 6068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5586114973391775e-05,
+      "loss": 0.6468,
+      "step": 6069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5584113228557886e-05,
+      "loss": 0.5869,
+      "step": 6070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5582111318927906e-05,
+      "loss": 0.6818,
+      "step": 6071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5580109244593863e-05,
+      "loss": 0.6261,
+      "step": 6072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.557810700564783e-05,
+      "loss": 0.6924,
+      "step": 6073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5576104602181857e-05,
+      "loss": 0.7,
+      "step": 6074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.557410203428802e-05,
+      "loss": 0.7326,
+      "step": 6075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5572099302058393e-05,
+      "loss": 0.7498,
+      "step": 6076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.557009640558506e-05,
+      "loss": 0.7316,
+      "step": 6077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.556809334496012e-05,
+      "loss": 0.7225,
+      "step": 6078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5566090120275662e-05,
+      "loss": 0.6439,
+      "step": 6079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5564086731623808e-05,
+      "loss": 0.6822,
+      "step": 6080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5562083179096658e-05,
+      "loss": 0.6893,
+      "step": 6081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5560079462786347e-05,
+      "loss": 0.6583,
+      "step": 6082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5558075582784994e-05,
+      "loss": 0.6887,
+      "step": 6083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5556071539184747e-05,
+      "loss": 0.702,
+      "step": 6084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5554067332077743e-05,
+      "loss": 0.6236,
+      "step": 6085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5552062961556138e-05,
+      "loss": 0.6578,
+      "step": 6086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5550058427712093e-05,
+      "loss": 0.7008,
+      "step": 6087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5548053730637775e-05,
+      "loss": 0.6429,
+      "step": 6088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5546048870425356e-05,
+      "loss": 0.6042,
+      "step": 6089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5544043847167026e-05,
+      "loss": 0.7192,
+      "step": 6090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.554203866095497e-05,
+      "loss": 0.6589,
+      "step": 6091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5540033311881384e-05,
+      "loss": 0.7422,
+      "step": 6092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5538027800038475e-05,
+      "loss": 0.7491,
+      "step": 6093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5536022125518457e-05,
+      "loss": 0.6522,
+      "step": 6094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.553401628841355e-05,
+      "loss": 0.6995,
+      "step": 6095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.553201028881598e-05,
+      "loss": 0.671,
+      "step": 6096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5530004126817983e-05,
+      "loss": 0.7268,
+      "step": 6097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5527997802511804e-05,
+      "loss": 0.7276,
+      "step": 6098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.552599131598969e-05,
+      "loss": 0.6819,
+      "step": 6099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5523984667343895e-05,
+      "loss": 0.7059,
+      "step": 6100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5521977856666692e-05,
+      "loss": 0.7004,
+      "step": 6101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5519970884050345e-05,
+      "loss": 0.6891,
+      "step": 6102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.551796374958714e-05,
+      "loss": 0.6998,
+      "step": 6103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5515956453369362e-05,
+      "loss": 0.6419,
+      "step": 6104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5513948995489307e-05,
+      "loss": 0.6646,
+      "step": 6105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5511941376039273e-05,
+      "loss": 0.6918,
+      "step": 6106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.550993359511158e-05,
+      "loss": 0.6507,
+      "step": 6107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5507925652798527e-05,
+      "loss": 0.6794,
+      "step": 6108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5505917549192456e-05,
+      "loss": 0.7556,
+      "step": 6109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5503909284385688e-05,
+      "loss": 0.6904,
+      "step": 6110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5501900858470567e-05,
+      "loss": 0.7014,
+      "step": 6111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5499892271539442e-05,
+      "loss": 0.6883,
+      "step": 6112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.549788352368466e-05,
+      "loss": 0.7069,
+      "step": 6113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.549587461499859e-05,
+      "loss": 0.6628,
+      "step": 6114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.549386554557359e-05,
+      "loss": 0.7011,
+      "step": 6115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.549185631550205e-05,
+      "loss": 0.6595,
+      "step": 6116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.548984692487634e-05,
+      "loss": 0.7744,
+      "step": 6117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.548783737378886e-05,
+      "loss": 0.6626,
+      "step": 6118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5485827662332006e-05,
+      "loss": 0.6586,
+      "step": 6119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5483817790598183e-05,
+      "loss": 0.6831,
+      "step": 6120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.548180775867981e-05,
+      "loss": 0.6859,
+      "step": 6121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5479797566669296e-05,
+      "loss": 0.7496,
+      "step": 6122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.547778721465908e-05,
+      "loss": 0.687,
+      "step": 6123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.547577670274159e-05,
+      "loss": 0.6639,
+      "step": 6124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5473766031009267e-05,
+      "loss": 0.6101,
+      "step": 6125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.547175519955457e-05,
+      "loss": 0.7187,
+      "step": 6126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5469744208469952e-05,
+      "loss": 0.702,
+      "step": 6127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5467733057847872e-05,
+      "loss": 0.6769,
+      "step": 6128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.546572174778081e-05,
+      "loss": 0.691,
+      "step": 6129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5463710278361237e-05,
+      "loss": 0.6879,
+      "step": 6130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.546169864968165e-05,
+      "loss": 0.6235,
+      "step": 6131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5459686861834535e-05,
+      "loss": 0.6308,
+      "step": 6132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5457674914912394e-05,
+      "loss": 0.7721,
+      "step": 6133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5455662809007735e-05,
+      "loss": 0.6772,
+      "step": 6134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.545365054421308e-05,
+      "loss": 0.6937,
+      "step": 6135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.545163812062094e-05,
+      "loss": 0.7069,
+      "step": 6136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5449625538323862e-05,
+      "loss": 0.7257,
+      "step": 6137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.544761279741437e-05,
+      "loss": 0.6842,
+      "step": 6138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.544559989798502e-05,
+      "loss": 0.7066,
+      "step": 6139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5443586840128352e-05,
+      "loss": 0.7046,
+      "step": 6140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5441573623936935e-05,
+      "loss": 0.6677,
+      "step": 6141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.543956024950333e-05,
+      "loss": 0.7477,
+      "step": 6142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.543754671692012e-05,
+      "loss": 0.6976,
+      "step": 6143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5435533026279873e-05,
+      "loss": 0.632,
+      "step": 6144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.543351917767519e-05,
+      "loss": 0.6531,
+      "step": 6145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5431505171198657e-05,
+      "loss": 0.7727,
+      "step": 6146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5429491006942886e-05,
+      "loss": 0.7024,
+      "step": 6147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5427476685000486e-05,
+      "loss": 0.6464,
+      "step": 6148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5425462205464066e-05,
+      "loss": 0.6682,
+      "step": 6149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.542344756842626e-05,
+      "loss": 0.6819,
+      "step": 6150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5421432773979698e-05,
+      "loss": 0.6816,
+      "step": 6151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.541941782221702e-05,
+      "loss": 0.6926,
+      "step": 6152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.541740271323087e-05,
+      "loss": 0.6718,
+      "step": 6153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5415387447113906e-05,
+      "loss": 0.6012,
+      "step": 6154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5413372023958785e-05,
+      "loss": 0.6387,
+      "step": 6155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.541135644385818e-05,
+      "loss": 0.6746,
+      "step": 6156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5409340706904757e-05,
+      "loss": 0.6798,
+      "step": 6157
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5407324813191213e-05,
+      "loss": 0.7266,
+      "step": 6158
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5405308762810227e-05,
+      "loss": 0.6094,
+      "step": 6159
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.54032925558545e-05,
+      "loss": 0.5917,
+      "step": 6160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5401276192416736e-05,
+      "loss": 0.6621,
+      "step": 6161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5399259672589644e-05,
+      "loss": 0.7216,
+      "step": 6162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5397242996465948e-05,
+      "loss": 0.6702,
+      "step": 6163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5395226164138372e-05,
+      "loss": 0.6585,
+      "step": 6164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.539320917569965e-05,
+      "loss": 0.7061,
+      "step": 6165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5391192031242516e-05,
+      "loss": 0.7023,
+      "step": 6166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5389174730859726e-05,
+      "loss": 0.641,
+      "step": 6167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.538715727464403e-05,
+      "loss": 0.6385,
+      "step": 6168
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.538513966268819e-05,
+      "loss": 0.6729,
+      "step": 6169
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5383121895084973e-05,
+      "loss": 0.6952,
+      "step": 6170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5381103971927165e-05,
+      "loss": 0.7194,
+      "step": 6171
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.537908589330754e-05,
+      "loss": 0.6353,
+      "step": 6172
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5377067659318888e-05,
+      "loss": 0.7815,
+      "step": 6173
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.537504927005401e-05,
+      "loss": 0.7009,
+      "step": 6174
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5373030725605715e-05,
+      "loss": 0.6941,
+      "step": 6175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.53710120260668e-05,
+      "loss": 0.6443,
+      "step": 6176
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5368993171530103e-05,
+      "loss": 0.6448,
+      "step": 6177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5366974162088438e-05,
+      "loss": 0.7193,
+      "step": 6178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.536495499783464e-05,
+      "loss": 0.6407,
+      "step": 6179
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5362935678861553e-05,
+      "loss": 0.6663,
+      "step": 6180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.536091620526202e-05,
+      "loss": 0.7333,
+      "step": 6181
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.53588965771289e-05,
+      "loss": 0.7209,
+      "step": 6182
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.535687679455505e-05,
+      "loss": 0.6653,
+      "step": 6183
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5354856857633345e-05,
+      "loss": 0.6595,
+      "step": 6184
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5352836766456653e-05,
+      "loss": 0.7198,
+      "step": 6185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5350816521117864e-05,
+      "loss": 0.7346,
+      "step": 6186
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5348796121709863e-05,
+      "loss": 0.5804,
+      "step": 6187
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.534677556832555e-05,
+      "loss": 0.651,
+      "step": 6188
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.534475486105783e-05,
+      "loss": 0.6343,
+      "step": 6189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5342733999999615e-05,
+      "loss": 0.6773,
+      "step": 6190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5340712985243815e-05,
+      "loss": 0.6591,
+      "step": 6191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5338691816883367e-05,
+      "loss": 0.6334,
+      "step": 6192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5336670495011194e-05,
+      "loss": 0.734,
+      "step": 6193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5334649019720246e-05,
+      "loss": 0.7482,
+      "step": 6194
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5332627391103457e-05,
+      "loss": 0.6991,
+      "step": 6195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.533060560925379e-05,
+      "loss": 0.7116,
+      "step": 6196
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5328583674264204e-05,
+      "loss": 0.7066,
+      "step": 6197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5326561586227664e-05,
+      "loss": 0.6789,
+      "step": 6198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5324539345237147e-05,
+      "loss": 0.6075,
+      "step": 6199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5322516951385633e-05,
+      "loss": 0.6692,
+      "step": 6200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5320494404766116e-05,
+      "loss": 0.7352,
+      "step": 6201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5318471705471587e-05,
+      "loss": 0.8003,
+      "step": 6202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5316448853595046e-05,
+      "loss": 0.6938,
+      "step": 6203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.531442584922951e-05,
+      "loss": 0.7393,
+      "step": 6204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5312402692467993e-05,
+      "loss": 0.6727,
+      "step": 6205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5310379383403515e-05,
+      "loss": 0.6436,
+      "step": 6206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5308355922129117e-05,
+      "loss": 0.6821,
+      "step": 6207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5306332308737828e-05,
+      "loss": 0.683,
+      "step": 6208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5304308543322694e-05,
+      "loss": 0.7176,
+      "step": 6209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.530228462597677e-05,
+      "loss": 0.6612,
+      "step": 6210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.530026055679311e-05,
+      "loss": 0.6222,
+      "step": 6211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5298236335864786e-05,
+      "loss": 0.6144,
+      "step": 6212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.529621196328487e-05,
+      "loss": 0.6874,
+      "step": 6213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5294187439146442e-05,
+      "loss": 0.6966,
+      "step": 6214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5292162763542582e-05,
+      "loss": 0.7448,
+      "step": 6215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.529013793656639e-05,
+      "loss": 0.7103,
+      "step": 6216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5288112958310965e-05,
+      "loss": 0.6891,
+      "step": 6217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5286087828869417e-05,
+      "loss": 0.6734,
+      "step": 6218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5284062548334857e-05,
+      "loss": 0.6676,
+      "step": 6219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5282037116800407e-05,
+      "loss": 0.6726,
+      "step": 6220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5280011534359198e-05,
+      "loss": 0.6835,
+      "step": 6221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5277985801104365e-05,
+      "loss": 0.6569,
+      "step": 6222
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.527595991712905e-05,
+      "loss": 0.5991,
+      "step": 6223
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5273933882526402e-05,
+      "loss": 0.6612,
+      "step": 6224
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5271907697389576e-05,
+      "loss": 0.7248,
+      "step": 6225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.526988136181174e-05,
+      "loss": 0.6761,
+      "step": 6226
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5267854875886058e-05,
+      "loss": 0.6686,
+      "step": 6227
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5265828239705708e-05,
+      "loss": 0.6788,
+      "step": 6228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.526380145336388e-05,
+      "loss": 0.6381,
+      "step": 6229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5261774516953754e-05,
+      "loss": 0.7414,
+      "step": 6230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.525974743056854e-05,
+      "loss": 0.6262,
+      "step": 6231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5257720194301433e-05,
+      "loss": 0.8005,
+      "step": 6232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5255692808245652e-05,
+      "loss": 0.7815,
+      "step": 6233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.525366527249441e-05,
+      "loss": 0.7085,
+      "step": 6234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5251637587140933e-05,
+      "loss": 0.6587,
+      "step": 6235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5249609752278453e-05,
+      "loss": 0.652,
+      "step": 6236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5247581768000212e-05,
+      "loss": 0.6852,
+      "step": 6237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5245553634399453e-05,
+      "loss": 0.6873,
+      "step": 6238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5243525351569433e-05,
+      "loss": 0.6581,
+      "step": 6239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5241496919603405e-05,
+      "loss": 0.6431,
+      "step": 6240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.523946833859464e-05,
+      "loss": 0.6485,
+      "step": 6241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5237439608636411e-05,
+      "loss": 0.7621,
+      "step": 6242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5235410729821997e-05,
+      "loss": 0.6959,
+      "step": 6243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5233381702244685e-05,
+      "loss": 0.6088,
+      "step": 6244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5231352525997768e-05,
+      "loss": 0.6968,
+      "step": 6245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5229323201174553e-05,
+      "loss": 0.6459,
+      "step": 6246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5227293727868339e-05,
+      "loss": 0.6741,
+      "step": 6247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5225264106172446e-05,
+      "loss": 0.6957,
+      "step": 6248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5223234336180192e-05,
+      "loss": 0.6027,
+      "step": 6249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5221204417984906e-05,
+      "loss": 0.7187,
+      "step": 6250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5219174351679927e-05,
+      "loss": 0.6709,
+      "step": 6251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.521714413735859e-05,
+      "loss": 0.6583,
+      "step": 6252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5215113775114244e-05,
+      "loss": 0.6711,
+      "step": 6253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5213083265040247e-05,
+      "loss": 0.6611,
+      "step": 6254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.521105260722996e-05,
+      "loss": 0.6884,
+      "step": 6255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5209021801776758e-05,
+      "loss": 0.6822,
+      "step": 6256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5206990848774006e-05,
+      "loss": 0.6717,
+      "step": 6257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5204959748315093e-05,
+      "loss": 0.7148,
+      "step": 6258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5202928500493406e-05,
+      "loss": 0.6314,
+      "step": 6259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5200897105402341e-05,
+      "loss": 0.6814,
+      "step": 6260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5198865563135302e-05,
+      "loss": 0.6909,
+      "step": 6261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5196833873785697e-05,
+      "loss": 0.6636,
+      "step": 6262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5194802037446945e-05,
+      "loss": 0.6641,
+      "step": 6263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5192770054212464e-05,
+      "loss": 0.6654,
+      "step": 6264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5190737924175688e-05,
+      "loss": 0.7266,
+      "step": 6265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5188705647430053e-05,
+      "loss": 0.7159,
+      "step": 6266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5186673224069004e-05,
+      "loss": 0.7517,
+      "step": 6267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5184640654185986e-05,
+      "loss": 0.6956,
+      "step": 6268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5182607937874461e-05,
+      "loss": 0.6889,
+      "step": 6269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5180575075227886e-05,
+      "loss": 0.7022,
+      "step": 6270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5178542066339736e-05,
+      "loss": 0.7154,
+      "step": 6271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5176508911303486e-05,
+      "loss": 0.6623,
+      "step": 6272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5174475610212623e-05,
+      "loss": 0.6426,
+      "step": 6273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5172442163160633e-05,
+      "loss": 0.6651,
+      "step": 6274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.517040857024102e-05,
+      "loss": 0.6831,
+      "step": 6275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5168374831547278e-05,
+      "loss": 0.6888,
+      "step": 6276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5166340947172928e-05,
+      "loss": 0.7069,
+      "step": 6277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5164306917211475e-05,
+      "loss": 0.6905,
+      "step": 6278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5162272741756454e-05,
+      "loss": 0.6532,
+      "step": 6279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.516023842090139e-05,
+      "loss": 0.6739,
+      "step": 6280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.515820395473982e-05,
+      "loss": 0.6877,
+      "step": 6281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5156169343365292e-05,
+      "loss": 0.5745,
+      "step": 6282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5154134586871352e-05,
+      "loss": 0.6096,
+      "step": 6283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5152099685351562e-05,
+      "loss": 0.7726,
+      "step": 6284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.515006463889948e-05,
+      "loss": 0.6532,
+      "step": 6285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.514802944760868e-05,
+      "loss": 0.7038,
+      "step": 6286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.514599411157274e-05,
+      "loss": 0.66,
+      "step": 6287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5143958630885243e-05,
+      "loss": 0.6495,
+      "step": 6288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.514192300563978e-05,
+      "loss": 0.752,
+      "step": 6289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5139887235929948e-05,
+      "loss": 0.711,
+      "step": 6290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.513785132184935e-05,
+      "loss": 0.6446,
+      "step": 6291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.51358152634916e-05,
+      "loss": 0.6766,
+      "step": 6292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5133779060950309e-05,
+      "loss": 0.579,
+      "step": 6293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5131742714319106e-05,
+      "loss": 0.6589,
+      "step": 6294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5129706223691618e-05,
+      "loss": 0.667,
+      "step": 6295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5127669589161484e-05,
+      "loss": 0.7423,
+      "step": 6296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.512563281082235e-05,
+      "loss": 0.6523,
+      "step": 6297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5123595888767861e-05,
+      "loss": 0.6808,
+      "step": 6298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5121558823091677e-05,
+      "loss": 0.6962,
+      "step": 6299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5119521613887463e-05,
+      "loss": 0.6497,
+      "step": 6300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5117484261248885e-05,
+      "loss": 0.7058,
+      "step": 6301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5115446765269621e-05,
+      "loss": 0.7099,
+      "step": 6302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5113409126043357e-05,
+      "loss": 0.6496,
+      "step": 6303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5111371343663777e-05,
+      "loss": 0.7633,
+      "step": 6304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5109333418224583e-05,
+      "loss": 0.6445,
+      "step": 6305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5107295349819477e-05,
+      "loss": 0.6834,
+      "step": 6306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5105257138542169e-05,
+      "loss": 0.6888,
+      "step": 6307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5103218784486371e-05,
+      "loss": 0.7278,
+      "step": 6308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.510118028774581e-05,
+      "loss": 0.6866,
+      "step": 6309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5099141648414216e-05,
+      "loss": 0.6546,
+      "step": 6310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.509710286658532e-05,
+      "loss": 0.7173,
+      "step": 6311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5095063942352866e-05,
+      "loss": 0.6186,
+      "step": 6312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5093024875810604e-05,
+      "loss": 0.8181,
+      "step": 6313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5090985667052294e-05,
+      "loss": 0.6666,
+      "step": 6314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5088946316171688e-05,
+      "loss": 0.6797,
+      "step": 6315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5086906823262564e-05,
+      "loss": 0.6292,
+      "step": 6316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.508486718841869e-05,
+      "loss": 0.7048,
+      "step": 6317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5082827411733855e-05,
+      "loss": 0.6907,
+      "step": 6318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5080787493301838e-05,
+      "loss": 0.6453,
+      "step": 6319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5078747433216441e-05,
+      "loss": 0.6435,
+      "step": 6320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.507670723157146e-05,
+      "loss": 0.6947,
+      "step": 6321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5074666888460709e-05,
+      "loss": 0.6172,
+      "step": 6322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5072626403977995e-05,
+      "loss": 0.6286,
+      "step": 6323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5070585778217146e-05,
+      "loss": 0.6978,
+      "step": 6324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5068545011271982e-05,
+      "loss": 0.6168,
+      "step": 6325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5066504103236341e-05,
+      "loss": 0.6515,
+      "step": 6326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.506446305420406e-05,
+      "loss": 0.664,
+      "step": 6327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5062421864268993e-05,
+      "loss": 0.6777,
+      "step": 6328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5060380533524984e-05,
+      "loss": 0.6639,
+      "step": 6329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5058339062065895e-05,
+      "loss": 0.6718,
+      "step": 6330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5056297449985595e-05,
+      "loss": 0.6691,
+      "step": 6331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5054255697377954e-05,
+      "loss": 0.7379,
+      "step": 6332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5052213804336854e-05,
+      "loss": 0.6823,
+      "step": 6333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5050171770956176e-05,
+      "loss": 0.7012,
+      "step": 6334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5048129597329816e-05,
+      "loss": 0.7101,
+      "step": 6335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.504608728355167e-05,
+      "loss": 0.6817,
+      "step": 6336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.504404482971564e-05,
+      "loss": 0.574,
+      "step": 6337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5042002235915645e-05,
+      "loss": 0.6823,
+      "step": 6338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5039959502245595e-05,
+      "loss": 0.692,
+      "step": 6339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.503791662879942e-05,
+      "loss": 0.7484,
+      "step": 6340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5035873615671047e-05,
+      "loss": 0.6095,
+      "step": 6341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5033830462954413e-05,
+      "loss": 0.642,
+      "step": 6342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5031787170743462e-05,
+      "loss": 0.6493,
+      "step": 6343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5029743739132143e-05,
+      "loss": 0.7087,
+      "step": 6344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5027700168214416e-05,
+      "loss": 0.6892,
+      "step": 6345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5025656458084236e-05,
+      "loss": 0.6554,
+      "step": 6346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5023612608835577e-05,
+      "loss": 0.5976,
+      "step": 6347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5021568620562418e-05,
+      "loss": 0.6347,
+      "step": 6348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5019524493358734e-05,
+      "loss": 0.6783,
+      "step": 6349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5017480227318517e-05,
+      "loss": 0.6672,
+      "step": 6350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5015435822535758e-05,
+      "loss": 0.6206,
+      "step": 6351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5013391279104462e-05,
+      "loss": 0.7032,
+      "step": 6352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5011346597118635e-05,
+      "loss": 0.7127,
+      "step": 6353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5009301776672287e-05,
+      "loss": 0.6749,
+      "step": 6354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5007256817859443e-05,
+      "loss": 0.7329,
+      "step": 6355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5005211720774128e-05,
+      "loss": 0.7016,
+      "step": 6356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5003166485510374e-05,
+      "loss": 0.7558,
+      "step": 6357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5001121112162223e-05,
+      "loss": 0.6933,
+      "step": 6358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4999075600823714e-05,
+      "loss": 0.6137,
+      "step": 6359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4997029951588905e-05,
+      "loss": 0.6304,
+      "step": 6360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4994984164551853e-05,
+      "loss": 0.6142,
+      "step": 6361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.499293823980662e-05,
+      "loss": 0.6831,
+      "step": 6362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4990892177447276e-05,
+      "loss": 0.7743,
+      "step": 6363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4988845977567903e-05,
+      "loss": 0.7086,
+      "step": 6364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4986799640262582e-05,
+      "loss": 0.6054,
+      "step": 6365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4984753165625401e-05,
+      "loss": 0.6445,
+      "step": 6366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4982706553750459e-05,
+      "loss": 0.5928,
+      "step": 6367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4980659804731854e-05,
+      "loss": 0.7175,
+      "step": 6368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4978612918663702e-05,
+      "loss": 0.6619,
+      "step": 6369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.497656589564011e-05,
+      "loss": 0.6916,
+      "step": 6370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4974518735755203e-05,
+      "loss": 0.635,
+      "step": 6371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4972471439103108e-05,
+      "loss": 0.6833,
+      "step": 6372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4970424005777963e-05,
+      "loss": 0.653,
+      "step": 6373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4968376435873903e-05,
+      "loss": 0.6672,
+      "step": 6374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4966328729485077e-05,
+      "loss": 0.7181,
+      "step": 6375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4964280886705635e-05,
+      "loss": 0.6955,
+      "step": 6376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4962232907629742e-05,
+      "loss": 0.687,
+      "step": 6377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4960184792351554e-05,
+      "loss": 0.6588,
+      "step": 6378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.495813654096525e-05,
+      "loss": 0.7081,
+      "step": 6379
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4956088153565004e-05,
+      "loss": 0.6273,
+      "step": 6380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4954039630245003e-05,
+      "loss": 0.6723,
+      "step": 6381
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4951990971099435e-05,
+      "loss": 0.6144,
+      "step": 6382
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4949942176222496e-05,
+      "loss": 0.6928,
+      "step": 6383
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4947893245708394e-05,
+      "loss": 0.6426,
+      "step": 6384
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.494584417965133e-05,
+      "loss": 0.6638,
+      "step": 6385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4943794978145526e-05,
+      "loss": 0.6434,
+      "step": 6386
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.49417456412852e-05,
+      "loss": 0.6722,
+      "step": 6387
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4939696169164576e-05,
+      "loss": 0.7037,
+      "step": 6388
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4937646561877896e-05,
+      "loss": 0.6407,
+      "step": 6389
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.49355968195194e-05,
+      "loss": 0.693,
+      "step": 6390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4933546942183326e-05,
+      "loss": 0.5714,
+      "step": 6391
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4931496929963934e-05,
+      "loss": 0.6896,
+      "step": 6392
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4929446782955476e-05,
+      "loss": 0.6479,
+      "step": 6393
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4927396501252227e-05,
+      "loss": 0.6227,
+      "step": 6394
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4925346084948452e-05,
+      "loss": 0.6947,
+      "step": 6395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4923295534138424e-05,
+      "loss": 0.6541,
+      "step": 6396
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.492124484891643e-05,
+      "loss": 0.7526,
+      "step": 6397
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4919194029376762e-05,
+      "loss": 0.6338,
+      "step": 6398
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4917143075613717e-05,
+      "loss": 0.6755,
+      "step": 6399
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4915091987721595e-05,
+      "loss": 0.6504,
+      "step": 6400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4913040765794702e-05,
+      "loss": 0.7507,
+      "step": 6401
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4910989409927355e-05,
+      "loss": 0.6473,
+      "step": 6402
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4908937920213875e-05,
+      "loss": 0.7681,
+      "step": 6403
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4906886296748584e-05,
+      "loss": 0.6942,
+      "step": 6404
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4904834539625818e-05,
+      "loss": 0.6279,
+      "step": 6405
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.490278264893992e-05,
+      "loss": 0.653,
+      "step": 6406
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4900730624785229e-05,
+      "loss": 0.6731,
+      "step": 6407
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4898678467256098e-05,
+      "loss": 0.6588,
+      "step": 6408
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4896626176446887e-05,
+      "loss": 0.7245,
+      "step": 6409
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4894573752451955e-05,
+      "loss": 0.6209,
+      "step": 6410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4892521195365679e-05,
+      "loss": 0.7381,
+      "step": 6411
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4890468505282424e-05,
+      "loss": 0.7325,
+      "step": 6412
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.488841568229658e-05,
+      "loss": 0.6546,
+      "step": 6413
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4886362726502534e-05,
+      "loss": 0.6953,
+      "step": 6414
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4884309637994678e-05,
+      "loss": 0.6668,
+      "step": 6415
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4882256416867413e-05,
+      "loss": 0.7189,
+      "step": 6416
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4880203063215143e-05,
+      "loss": 0.7435,
+      "step": 6417
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4878149577132288e-05,
+      "loss": 0.7185,
+      "step": 6418
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4876095958713257e-05,
+      "loss": 0.6528,
+      "step": 6419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.487404220805248e-05,
+      "loss": 0.6342,
+      "step": 6420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4871988325244383e-05,
+      "loss": 0.6997,
+      "step": 6421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4869934310383408e-05,
+      "loss": 0.6381,
+      "step": 6422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4867880163563999e-05,
+      "loss": 0.7317,
+      "step": 6423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.48658258848806e-05,
+      "loss": 0.7577,
+      "step": 6424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4863771474427666e-05,
+      "loss": 0.7007,
+      "step": 6425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.486171693229966e-05,
+      "loss": 0.7095,
+      "step": 6426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.485966225859105e-05,
+      "loss": 0.6733,
+      "step": 6427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4857607453396307e-05,
+      "loss": 0.6797,
+      "step": 6428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4855552516809912e-05,
+      "loss": 0.6624,
+      "step": 6429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4853497448926345e-05,
+      "loss": 0.6049,
+      "step": 6430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4851442249840106e-05,
+      "loss": 0.6247,
+      "step": 6431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4849386919645686e-05,
+      "loss": 0.7802,
+      "step": 6432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.484733145843759e-05,
+      "loss": 0.681,
+      "step": 6433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4845275866310324e-05,
+      "loss": 0.6585,
+      "step": 6434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.484322014335841e-05,
+      "loss": 0.7159,
+      "step": 6435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4841164289676363e-05,
+      "loss": 0.6324,
+      "step": 6436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4839108305358715e-05,
+      "loss": 0.7329,
+      "step": 6437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4837052190499996e-05,
+      "loss": 0.7207,
+      "step": 6438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4834995945194746e-05,
+      "loss": 0.6608,
+      "step": 6439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4832939569537512e-05,
+      "loss": 0.6486,
+      "step": 6440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4830883063622844e-05,
+      "loss": 0.5915,
+      "step": 6441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.48288264275453e-05,
+      "loss": 0.6718,
+      "step": 6442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4826769661399445e-05,
+      "loss": 0.698,
+      "step": 6443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4824712765279843e-05,
+      "loss": 0.7061,
+      "step": 6444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4822655739281076e-05,
+      "loss": 0.5786,
+      "step": 6445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4820598583497715e-05,
+      "loss": 0.7968,
+      "step": 6446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4818541298024357e-05,
+      "loss": 0.6594,
+      "step": 6447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4816483882955597e-05,
+      "loss": 0.6632,
+      "step": 6448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4814426338386023e-05,
+      "loss": 0.7068,
+      "step": 6449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.481236866441025e-05,
+      "loss": 0.6839,
+      "step": 6450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4810310861122885e-05,
+      "loss": 0.6956,
+      "step": 6451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4808252928618545e-05,
+      "loss": 0.6114,
+      "step": 6452
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4806194866991851e-05,
+      "loss": 0.722,
+      "step": 6453
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4804136676337438e-05,
+      "loss": 0.6855,
+      "step": 6454
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4802078356749933e-05,
+      "loss": 0.6216,
+      "step": 6455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4800019908323981e-05,
+      "loss": 0.6818,
+      "step": 6456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4797961331154229e-05,
+      "loss": 0.6716,
+      "step": 6457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.479590262533533e-05,
+      "loss": 0.6308,
+      "step": 6458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.479384379096194e-05,
+      "loss": 0.7282,
+      "step": 6459
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4791784828128726e-05,
+      "loss": 0.6982,
+      "step": 6460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4789725736930355e-05,
+      "loss": 0.7428,
+      "step": 6461
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4787666517461507e-05,
+      "loss": 0.6407,
+      "step": 6462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.478560716981686e-05,
+      "loss": 0.6309,
+      "step": 6463
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4783547694091105e-05,
+      "loss": 0.7522,
+      "step": 6464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4781488090378937e-05,
+      "loss": 0.7309,
+      "step": 6465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4779428358775051e-05,
+      "loss": 0.654,
+      "step": 6466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4777368499374159e-05,
+      "loss": 0.6631,
+      "step": 6467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4775308512270965e-05,
+      "loss": 0.6609,
+      "step": 6468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4773248397560195e-05,
+      "loss": 0.5945,
+      "step": 6469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4771188155336565e-05,
+      "loss": 0.6658,
+      "step": 6470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4769127785694807e-05,
+      "loss": 0.6285,
+      "step": 6471
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4767067288729655e-05,
+      "loss": 0.6496,
+      "step": 6472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4765006664535848e-05,
+      "loss": 0.6419,
+      "step": 6473
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4762945913208137e-05,
+      "loss": 0.651,
+      "step": 6474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4760885034841277e-05,
+      "loss": 0.6476,
+      "step": 6475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4758824029530019e-05,
+      "loss": 0.5798,
+      "step": 6476
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.475676289736913e-05,
+      "loss": 0.636,
+      "step": 6477
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4754701638453382e-05,
+      "loss": 0.6399,
+      "step": 6478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4752640252877547e-05,
+      "loss": 0.6921,
+      "step": 6479
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.475057874073641e-05,
+      "loss": 0.6658,
+      "step": 6480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4748517102124754e-05,
+      "loss": 0.5892,
+      "step": 6481
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4746455337137382e-05,
+      "loss": 0.6111,
+      "step": 6482
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4744393445869083e-05,
+      "loss": 0.6486,
+      "step": 6483
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.474233142841467e-05,
+      "loss": 0.6847,
+      "step": 6484
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4740269284868944e-05,
+      "loss": 0.6508,
+      "step": 6485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.473820701532673e-05,
+      "loss": 0.6865,
+      "step": 6486
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4736144619882846e-05,
+      "loss": 0.6644,
+      "step": 6487
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4734082098632126e-05,
+      "loss": 0.7303,
+      "step": 6488
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4732019451669392e-05,
+      "loss": 0.7657,
+      "step": 6489
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4729956679089496e-05,
+      "loss": 0.6353,
+      "step": 6490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4727893780987277e-05,
+      "loss": 0.7165,
+      "step": 6491
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4725830757457588e-05,
+      "loss": 0.7381,
+      "step": 6492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4723767608595288e-05,
+      "loss": 0.6719,
+      "step": 6493
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4721704334495235e-05,
+      "loss": 0.626,
+      "step": 6494
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4719640935252301e-05,
+      "loss": 0.6644,
+      "step": 6495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4717577410961362e-05,
+      "loss": 0.6773,
+      "step": 6496
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.471551376171729e-05,
+      "loss": 0.6802,
+      "step": 6497
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.471344998761498e-05,
+      "loss": 0.7498,
+      "step": 6498
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.471138608874932e-05,
+      "loss": 0.6088,
+      "step": 6499
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4709322065215206e-05,
+      "loss": 0.6911,
+      "step": 6500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.470725791710754e-05,
+      "loss": 0.6096,
+      "step": 6501
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4705193644521234e-05,
+      "loss": 0.7214,
+      "step": 6502
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4703129247551204e-05,
+      "loss": 0.6839,
+      "step": 6503
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4701064726292363e-05,
+      "loss": 0.7002,
+      "step": 6504
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4699000080839643e-05,
+      "loss": 0.6718,
+      "step": 6505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4696935311287971e-05,
+      "loss": 0.6598,
+      "step": 6506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4694870417732292e-05,
+      "loss": 0.6069,
+      "step": 6507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.469280540026754e-05,
+      "loss": 0.6553,
+      "step": 6508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4690740258988671e-05,
+      "loss": 0.6794,
+      "step": 6509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4688674993990632e-05,
+      "loss": 0.7114,
+      "step": 6510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4686609605368392e-05,
+      "loss": 0.728,
+      "step": 6511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4684544093216908e-05,
+      "loss": 0.7192,
+      "step": 6512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4682478457631158e-05,
+      "loss": 0.6786,
+      "step": 6513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4680412698706111e-05,
+      "loss": 0.6875,
+      "step": 6514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.467834681653676e-05,
+      "loss": 0.6943,
+      "step": 6515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4676280811218086e-05,
+      "loss": 0.6622,
+      "step": 6516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4674214682845086e-05,
+      "loss": 0.6645,
+      "step": 6517
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4672148431512762e-05,
+      "loss": 0.6793,
+      "step": 6518
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4670082057316115e-05,
+      "loss": 0.6771,
+      "step": 6519
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4668015560350159e-05,
+      "loss": 0.704,
+      "step": 6520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4665948940709908e-05,
+      "loss": 0.645,
+      "step": 6521
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4663882198490387e-05,
+      "loss": 0.7063,
+      "step": 6522
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.466181533378662e-05,
+      "loss": 0.6977,
+      "step": 6523
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4659748346693647e-05,
+      "loss": 0.6194,
+      "step": 6524
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4657681237306502e-05,
+      "loss": 0.6618,
+      "step": 6525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4655614005720233e-05,
+      "loss": 0.6391,
+      "step": 6526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4653546652029888e-05,
+      "loss": 0.6927,
+      "step": 6527
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4651479176330528e-05,
+      "loss": 0.6583,
+      "step": 6528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4649411578717205e-05,
+      "loss": 0.6416,
+      "step": 6529
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4647343859284996e-05,
+      "loss": 0.6975,
+      "step": 6530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4645276018128968e-05,
+      "loss": 0.7194,
+      "step": 6531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4643208055344202e-05,
+      "loss": 0.6517,
+      "step": 6532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4641139971025783e-05,
+      "loss": 0.6258,
+      "step": 6533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.46390717652688e-05,
+      "loss": 0.7105,
+      "step": 6534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4637003438168348e-05,
+      "loss": 0.6676,
+      "step": 6535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4634934989819526e-05,
+      "loss": 0.7719,
+      "step": 6536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4632866420317444e-05,
+      "loss": 0.6593,
+      "step": 6537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.463079772975721e-05,
+      "loss": 0.6045,
+      "step": 6538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4628728918233943e-05,
+      "loss": 0.693,
+      "step": 6539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4626659985842768e-05,
+      "loss": 0.7255,
+      "step": 6540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4624590932678812e-05,
+      "loss": 0.6237,
+      "step": 6541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.462252175883721e-05,
+      "loss": 0.6713,
+      "step": 6542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4620452464413103e-05,
+      "loss": 0.6487,
+      "step": 6543
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4618383049501634e-05,
+      "loss": 0.7306,
+      "step": 6544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4616313514197956e-05,
+      "loss": 0.72,
+      "step": 6545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4614243858597221e-05,
+      "loss": 0.6693,
+      "step": 6546
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4612174082794595e-05,
+      "loss": 0.6908,
+      "step": 6547
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4610104186885245e-05,
+      "loss": 0.7609,
+      "step": 6548
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4608034170964342e-05,
+      "loss": 0.6872,
+      "step": 6549
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.460596403512707e-05,
+      "loss": 0.6901,
+      "step": 6550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4603893779468604e-05,
+      "loss": 0.6361,
+      "step": 6551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4601823404084143e-05,
+      "loss": 0.6396,
+      "step": 6552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4599752909068874e-05,
+      "loss": 0.6779,
+      "step": 6553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4597682294518003e-05,
+      "loss": 0.6803,
+      "step": 6554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4595611560526733e-05,
+      "loss": 0.6033,
+      "step": 6555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4593540707190277e-05,
+      "loss": 0.6875,
+      "step": 6556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4591469734603851e-05,
+      "loss": 0.739,
+      "step": 6557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4589398642862681e-05,
+      "loss": 0.6519,
+      "step": 6558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.458732743206199e-05,
+      "loss": 0.676,
+      "step": 6559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4585256102297013e-05,
+      "loss": 0.5932,
+      "step": 6560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4583184653662992e-05,
+      "loss": 0.6468,
+      "step": 6561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4581113086255168e-05,
+      "loss": 0.7118,
+      "step": 6562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4579041400168794e-05,
+      "loss": 0.683,
+      "step": 6563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4576969595499117e-05,
+      "loss": 0.6256,
+      "step": 6564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4574897672341408e-05,
+      "loss": 0.6984,
+      "step": 6565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4572825630790927e-05,
+      "loss": 0.637,
+      "step": 6566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.457075347094295e-05,
+      "loss": 0.6924,
+      "step": 6567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4568681192892751e-05,
+      "loss": 0.7407,
+      "step": 6568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4566608796735613e-05,
+      "loss": 0.6578,
+      "step": 6569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4564536282566823e-05,
+      "loss": 0.7333,
+      "step": 6570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4562463650481678e-05,
+      "loss": 0.6769,
+      "step": 6571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4560390900575472e-05,
+      "loss": 0.5887,
+      "step": 6572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4558318032943512e-05,
+      "loss": 0.7371,
+      "step": 6573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4556245047681107e-05,
+      "loss": 0.7014,
+      "step": 6574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4554171944883575e-05,
+      "loss": 0.681,
+      "step": 6575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4552098724646231e-05,
+      "loss": 0.6878,
+      "step": 6576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4550025387064405e-05,
+      "loss": 0.7013,
+      "step": 6577
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4547951932233426e-05,
+      "loss": 0.6441,
+      "step": 6578
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4545878360248632e-05,
+      "loss": 0.6463,
+      "step": 6579
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4543804671205367e-05,
+      "loss": 0.6162,
+      "step": 6580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4541730865198974e-05,
+      "loss": 0.8128,
+      "step": 6581
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4539656942324805e-05,
+      "loss": 0.6555,
+      "step": 6582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.453758290267822e-05,
+      "loss": 0.6562,
+      "step": 6583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4535508746354588e-05,
+      "loss": 0.7191,
+      "step": 6584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4533434473449271e-05,
+      "loss": 0.666,
+      "step": 6585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4531360084057644e-05,
+      "loss": 0.7089,
+      "step": 6586
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4529285578275089e-05,
+      "loss": 0.6512,
+      "step": 6587
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4527210956196991e-05,
+      "loss": 0.6979,
+      "step": 6588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4525136217918737e-05,
+      "loss": 0.7214,
+      "step": 6589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4523061363535726e-05,
+      "loss": 0.6788,
+      "step": 6590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4520986393143357e-05,
+      "loss": 0.6176,
+      "step": 6591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.451891130683704e-05,
+      "loss": 0.6748,
+      "step": 6592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4516836104712181e-05,
+      "loss": 0.7183,
+      "step": 6593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4514760786864203e-05,
+      "loss": 0.7071,
+      "step": 6594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4512685353388522e-05,
+      "loss": 0.6519,
+      "step": 6595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.451060980438057e-05,
+      "loss": 0.6841,
+      "step": 6596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4508534139935779e-05,
+      "loss": 0.6181,
+      "step": 6597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4506458360149584e-05,
+      "loss": 0.7209,
+      "step": 6598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4504382465117435e-05,
+      "loss": 0.6323,
+      "step": 6599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4502306454934775e-05,
+      "loss": 0.641,
+      "step": 6600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4500230329697062e-05,
+      "loss": 0.6582,
+      "step": 6601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4498154089499751e-05,
+      "loss": 0.6857,
+      "step": 6602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4496077734438313e-05,
+      "loss": 0.7246,
+      "step": 6603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.449400126460821e-05,
+      "loss": 0.6612,
+      "step": 6604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4491924680104927e-05,
+      "loss": 0.6914,
+      "step": 6605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4489847981023934e-05,
+      "loss": 0.635,
+      "step": 6606
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4487771167460724e-05,
+      "loss": 0.6845,
+      "step": 6607
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4485694239510786e-05,
+      "loss": 0.605,
+      "step": 6608
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4483617197269618e-05,
+      "loss": 0.6543,
+      "step": 6609
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4481540040832717e-05,
+      "loss": 0.733,
+      "step": 6610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4479462770295596e-05,
+      "loss": 0.6628,
+      "step": 6611
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4477385385753761e-05,
+      "loss": 0.6052,
+      "step": 6612
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4475307887302734e-05,
+      "loss": 0.6592,
+      "step": 6613
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4473230275038035e-05,
+      "loss": 0.6847,
+      "step": 6614
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4471152549055191e-05,
+      "loss": 0.7059,
+      "step": 6615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4469074709449739e-05,
+      "loss": 0.6568,
+      "step": 6616
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4466996756317212e-05,
+      "loss": 0.6825,
+      "step": 6617
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.446491868975316e-05,
+      "loss": 0.7172,
+      "step": 6618
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4462840509853125e-05,
+      "loss": 0.6254,
+      "step": 6619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4460762216712665e-05,
+      "loss": 0.7035,
+      "step": 6620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4458683810427335e-05,
+      "loss": 0.6669,
+      "step": 6621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4456605291092707e-05,
+      "loss": 0.6962,
+      "step": 6622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4454526658804342e-05,
+      "loss": 0.6686,
+      "step": 6623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4452447913657819e-05,
+      "loss": 0.6895,
+      "step": 6624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4450369055748717e-05,
+      "loss": 0.6832,
+      "step": 6625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4448290085172622e-05,
+      "loss": 0.6422,
+      "step": 6626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4446211002025123e-05,
+      "loss": 0.7488,
+      "step": 6627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4444131806401818e-05,
+      "loss": 0.5686,
+      "step": 6628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4442052498398302e-05,
+      "loss": 0.7211,
+      "step": 6629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4439973078110187e-05,
+      "loss": 0.6789,
+      "step": 6630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4437893545633078e-05,
+      "loss": 0.6902,
+      "step": 6631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4435813901062596e-05,
+      "loss": 0.7125,
+      "step": 6632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4433734144494363e-05,
+      "loss": 0.6793,
+      "step": 6633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4431654276023999e-05,
+      "loss": 0.6381,
+      "step": 6634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4429574295747141e-05,
+      "loss": 0.6234,
+      "step": 6635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4427494203759424e-05,
+      "loss": 0.6515,
+      "step": 6636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.442541400015649e-05,
+      "loss": 0.6552,
+      "step": 6637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4423333685033984e-05,
+      "loss": 0.6,
+      "step": 6638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4421253258487564e-05,
+      "loss": 0.6848,
+      "step": 6639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.441917272061288e-05,
+      "loss": 0.7208,
+      "step": 6640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4417092071505596e-05,
+      "loss": 0.6602,
+      "step": 6641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.441501131126138e-05,
+      "loss": 0.6858,
+      "step": 6642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4412930439975909e-05,
+      "loss": 0.6917,
+      "step": 6643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4410849457744855e-05,
+      "loss": 0.672,
+      "step": 6644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4408768364663905e-05,
+      "loss": 0.6797,
+      "step": 6645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4406687160828741e-05,
+      "loss": 0.7972,
+      "step": 6646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4404605846335063e-05,
+      "loss": 0.716,
+      "step": 6647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.440252442127856e-05,
+      "loss": 0.6953,
+      "step": 6648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4400442885754943e-05,
+      "loss": 0.6503,
+      "step": 6649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4398361239859921e-05,
+      "loss": 0.7678,
+      "step": 6650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4396279483689202e-05,
+      "loss": 0.6665,
+      "step": 6651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4394197617338505e-05,
+      "loss": 0.6957,
+      "step": 6652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4392115640903554e-05,
+      "loss": 0.6737,
+      "step": 6653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4390033554480081e-05,
+      "loss": 0.6514,
+      "step": 6654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4387951358163814e-05,
+      "loss": 0.6822,
+      "step": 6655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4385869052050496e-05,
+      "loss": 0.638,
+      "step": 6656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4383786636235867e-05,
+      "loss": 0.6666,
+      "step": 6657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4381704110815676e-05,
+      "loss": 0.5932,
+      "step": 6658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4379621475885681e-05,
+      "loss": 0.704,
+      "step": 6659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4377538731541638e-05,
+      "loss": 0.6452,
+      "step": 6660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4375455877879307e-05,
+      "loss": 0.6275,
+      "step": 6661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4373372914994465e-05,
+      "loss": 0.6816,
+      "step": 6662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4371289842982879e-05,
+      "loss": 0.655,
+      "step": 6663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4369206661940332e-05,
+      "loss": 0.7083,
+      "step": 6664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.43671233719626e-05,
+      "loss": 0.7004,
+      "step": 6665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4365039973145483e-05,
+      "loss": 0.6495,
+      "step": 6666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4362956465584767e-05,
+      "loss": 0.6263,
+      "step": 6667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4360872849376256e-05,
+      "loss": 0.663,
+      "step": 6668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4358789124615748e-05,
+      "loss": 0.66,
+      "step": 6669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4356705291399056e-05,
+      "loss": 0.7166,
+      "step": 6670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4354621349821996e-05,
+      "loss": 0.6109,
+      "step": 6671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4352537299980377e-05,
+      "loss": 0.6221,
+      "step": 6672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4350453141970036e-05,
+      "loss": 0.6533,
+      "step": 6673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4348368875886791e-05,
+      "loss": 0.6604,
+      "step": 6674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4346284501826482e-05,
+      "loss": 0.6189,
+      "step": 6675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4344200019884944e-05,
+      "loss": 0.6973,
+      "step": 6676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4342115430158024e-05,
+      "loss": 0.6868,
+      "step": 6677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4340030732741567e-05,
+      "loss": 0.7857,
+      "step": 6678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4337945927731431e-05,
+      "loss": 0.6353,
+      "step": 6679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4335861015223468e-05,
+      "loss": 0.6751,
+      "step": 6680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.433377599531355e-05,
+      "loss": 0.6329,
+      "step": 6681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4331690868097538e-05,
+      "loss": 0.669,
+      "step": 6682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4329605633671309e-05,
+      "loss": 0.6589,
+      "step": 6683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4327520292130741e-05,
+      "loss": 0.6465,
+      "step": 6684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4325434843571713e-05,
+      "loss": 0.6993,
+      "step": 6685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.432334928809012e-05,
+      "loss": 0.6852,
+      "step": 6686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.432126362578185e-05,
+      "loss": 0.6532,
+      "step": 6687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4319177856742805e-05,
+      "loss": 0.6046,
+      "step": 6688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4317091981068882e-05,
+      "loss": 0.6893,
+      "step": 6689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4315005998855993e-05,
+      "loss": 0.6235,
+      "step": 6690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.431291991020005e-05,
+      "loss": 0.6842,
+      "step": 6691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.431083371519697e-05,
+      "loss": 0.6398,
+      "step": 6692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4308747413942676e-05,
+      "loss": 0.6469,
+      "step": 6693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4306661006533095e-05,
+      "loss": 0.6524,
+      "step": 6694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4304574493064158e-05,
+      "loss": 0.6567,
+      "step": 6695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4302487873631806e-05,
+      "loss": 0.7044,
+      "step": 6696
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4300401148331977e-05,
+      "loss": 0.6737,
+      "step": 6697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.429831431726062e-05,
+      "loss": 0.6538,
+      "step": 6698
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4296227380513683e-05,
+      "loss": 0.7532,
+      "step": 6699
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4294140338187124e-05,
+      "loss": 0.6414,
+      "step": 6700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4292053190376909e-05,
+      "loss": 0.5923,
+      "step": 6701
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4289965937178998e-05,
+      "loss": 0.6428,
+      "step": 6702
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4287878578689366e-05,
+      "loss": 0.6779,
+      "step": 6703
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4285791115003988e-05,
+      "loss": 0.7354,
+      "step": 6704
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4283703546218845e-05,
+      "loss": 0.662,
+      "step": 6705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.428161587242992e-05,
+      "loss": 0.7526,
+      "step": 6706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4279528093733207e-05,
+      "loss": 0.6787,
+      "step": 6707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4277440210224698e-05,
+      "loss": 0.649,
+      "step": 6708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4275352222000398e-05,
+      "loss": 0.6643,
+      "step": 6709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4273264129156305e-05,
+      "loss": 0.6771,
+      "step": 6710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4271175931788436e-05,
+      "loss": 0.7265,
+      "step": 6711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4269087629992797e-05,
+      "loss": 0.6066,
+      "step": 6712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4266999223865415e-05,
+      "loss": 0.7478,
+      "step": 6713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4264910713502313e-05,
+      "loss": 0.6248,
+      "step": 6714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4262822098999515e-05,
+      "loss": 0.6327,
+      "step": 6715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4260733380453057e-05,
+      "loss": 0.639,
+      "step": 6716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4258644557958981e-05,
+      "loss": 0.6098,
+      "step": 6717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4256555631613326e-05,
+      "loss": 0.7292,
+      "step": 6718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.425446660151214e-05,
+      "loss": 0.6686,
+      "step": 6719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.425237746775148e-05,
+      "loss": 0.6133,
+      "step": 6720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4250288230427397e-05,
+      "loss": 0.6673,
+      "step": 6721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.424819888963596e-05,
+      "loss": 0.7285,
+      "step": 6722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4246109445473233e-05,
+      "loss": 0.6608,
+      "step": 6723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4244019898035284e-05,
+      "loss": 0.709,
+      "step": 6724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4241930247418196e-05,
+      "loss": 0.7406,
+      "step": 6725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4239840493718049e-05,
+      "loss": 0.6559,
+      "step": 6726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4237750637030924e-05,
+      "loss": 0.656,
+      "step": 6727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.423566067745292e-05,
+      "loss": 0.7023,
+      "step": 6728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4233570615080125e-05,
+      "loss": 0.6274,
+      "step": 6729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4231480450008645e-05,
+      "loss": 0.6847,
+      "step": 6730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4229390182334581e-05,
+      "loss": 0.6468,
+      "step": 6731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4227299812154042e-05,
+      "loss": 0.6172,
+      "step": 6732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4225209339563144e-05,
+      "loss": 0.6825,
+      "step": 6733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4223118764658008e-05,
+      "loss": 0.7109,
+      "step": 6734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4221028087534758e-05,
+      "loss": 0.6541,
+      "step": 6735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4218937308289518e-05,
+      "loss": 0.7707,
+      "step": 6736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4216846427018425e-05,
+      "loss": 0.6237,
+      "step": 6737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4214755443817614e-05,
+      "loss": 0.6247,
+      "step": 6738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4212664358783233e-05,
+      "loss": 0.6334,
+      "step": 6739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4210573172011421e-05,
+      "loss": 0.7235,
+      "step": 6740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4208481883598336e-05,
+      "loss": 0.6691,
+      "step": 6741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4206390493640132e-05,
+      "loss": 0.6581,
+      "step": 6742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4204299002232975e-05,
+      "loss": 0.6647,
+      "step": 6743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4202207409473025e-05,
+      "loss": 0.7083,
+      "step": 6744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4200115715456458e-05,
+      "loss": 0.6914,
+      "step": 6745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4198023920279442e-05,
+      "loss": 0.6594,
+      "step": 6746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4195932024038163e-05,
+      "loss": 0.7439,
+      "step": 6747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4193840026828804e-05,
+      "loss": 0.6715,
+      "step": 6748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4191747928747554e-05,
+      "loss": 0.6274,
+      "step": 6749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4189655729890606e-05,
+      "loss": 0.6684,
+      "step": 6750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4187563430354156e-05,
+      "loss": 0.6205,
+      "step": 6751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4185471030234416e-05,
+      "loss": 0.639,
+      "step": 6752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4183378529627584e-05,
+      "loss": 0.7107,
+      "step": 6753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4181285928629879e-05,
+      "loss": 0.6558,
+      "step": 6754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4179193227337513e-05,
+      "loss": 0.7232,
+      "step": 6755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4177100425846712e-05,
+      "loss": 0.7778,
+      "step": 6756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4175007524253699e-05,
+      "loss": 0.6423,
+      "step": 6757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4172914522654705e-05,
+      "loss": 0.7184,
+      "step": 6758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4170821421145966e-05,
+      "loss": 0.6313,
+      "step": 6759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4168728219823726e-05,
+      "loss": 0.7044,
+      "step": 6760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4166634918784221e-05,
+      "loss": 0.7395,
+      "step": 6761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4164541518123708e-05,
+      "loss": 0.6415,
+      "step": 6762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4162448017938436e-05,
+      "loss": 0.6931,
+      "step": 6763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4160354418324668e-05,
+      "loss": 0.7227,
+      "step": 6764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4158260719378662e-05,
+      "loss": 0.6402,
+      "step": 6765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4156166921196686e-05,
+      "loss": 0.7041,
+      "step": 6766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4154073023875013e-05,
+      "loss": 0.7289,
+      "step": 6767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4151979027509921e-05,
+      "loss": 0.6606,
+      "step": 6768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.414988493219769e-05,
+      "loss": 0.7202,
+      "step": 6769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4147790738034606e-05,
+      "loss": 0.6722,
+      "step": 6770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4145696445116959e-05,
+      "loss": 0.7581,
+      "step": 6771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.414360205354104e-05,
+      "loss": 0.6477,
+      "step": 6772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4141507563403158e-05,
+      "loss": 0.7064,
+      "step": 6773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4139412974799607e-05,
+      "loss": 0.6491,
+      "step": 6774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4137318287826698e-05,
+      "loss": 0.608,
+      "step": 6775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4135223502580746e-05,
+      "loss": 0.6851,
+      "step": 6776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4133128619158067e-05,
+      "loss": 0.6451,
+      "step": 6777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4131033637654984e-05,
+      "loss": 0.6526,
+      "step": 6778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4128938558167824e-05,
+      "loss": 0.6386,
+      "step": 6779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4126843380792914e-05,
+      "loss": 0.7408,
+      "step": 6780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4124748105626596e-05,
+      "loss": 0.6691,
+      "step": 6781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4122652732765202e-05,
+      "loss": 0.6634,
+      "step": 6782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.412055726230508e-05,
+      "loss": 0.6244,
+      "step": 6783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4118461694342582e-05,
+      "loss": 0.69,
+      "step": 6784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4116366028974056e-05,
+      "loss": 0.6998,
+      "step": 6785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4114270266295864e-05,
+      "loss": 0.6452,
+      "step": 6786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4112174406404366e-05,
+      "loss": 0.6592,
+      "step": 6787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4110078449395931e-05,
+      "loss": 0.7558,
+      "step": 6788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4107982395366928e-05,
+      "loss": 0.6908,
+      "step": 6789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4105886244413734e-05,
+      "loss": 0.729,
+      "step": 6790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4103789996632725e-05,
+      "loss": 0.6628,
+      "step": 6791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4101693652120293e-05,
+      "loss": 0.6767,
+      "step": 6792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4099597210972824e-05,
+      "loss": 0.7582,
+      "step": 6793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.409750067328671e-05,
+      "loss": 0.6834,
+      "step": 6794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4095404039158348e-05,
+      "loss": 0.6847,
+      "step": 6795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4093307308684148e-05,
+      "loss": 0.5982,
+      "step": 6796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4091210481960506e-05,
+      "loss": 0.6518,
+      "step": 6797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4089113559083842e-05,
+      "loss": 0.6261,
+      "step": 6798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4087016540150565e-05,
+      "loss": 0.7594,
+      "step": 6799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4084919425257099e-05,
+      "loss": 0.6567,
+      "step": 6800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.408282221449987e-05,
+      "loss": 0.7728,
+      "step": 6801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4080724907975302e-05,
+      "loss": 0.6576,
+      "step": 6802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4078627505779834e-05,
+      "loss": 0.7165,
+      "step": 6803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4076530008009898e-05,
+      "loss": 0.6549,
+      "step": 6804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.407443241476194e-05,
+      "loss": 0.6998,
+      "step": 6805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4072334726132406e-05,
+      "loss": 0.7147,
+      "step": 6806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4070236942217746e-05,
+      "loss": 0.7228,
+      "step": 6807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4068139063114414e-05,
+      "loss": 0.6399,
+      "step": 6808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4066041088918872e-05,
+      "loss": 0.6696,
+      "step": 6809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4063943019727582e-05,
+      "loss": 0.6684,
+      "step": 6810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4061844855637015e-05,
+      "loss": 0.6192,
+      "step": 6811
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4059746596743642e-05,
+      "loss": 0.7087,
+      "step": 6812
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4057648243143942e-05,
+      "loss": 0.6175,
+      "step": 6813
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.405554979493439e-05,
+      "loss": 0.6645,
+      "step": 6814
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4053451252211483e-05,
+      "loss": 0.6601,
+      "step": 6815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.40513526150717e-05,
+      "loss": 0.7223,
+      "step": 6816
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4049253883611543e-05,
+      "loss": 0.6075,
+      "step": 6817
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4047155057927506e-05,
+      "loss": 0.6243,
+      "step": 6818
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4045056138116095e-05,
+      "loss": 0.6726,
+      "step": 6819
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.404295712427382e-05,
+      "loss": 0.6932,
+      "step": 6820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4040858016497188e-05,
+      "loss": 0.5602,
+      "step": 6821
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4038758814882716e-05,
+      "loss": 0.661,
+      "step": 6822
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4036659519526923e-05,
+      "loss": 0.6653,
+      "step": 6823
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.403456013052634e-05,
+      "loss": 0.6816,
+      "step": 6824
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.403246064797749e-05,
+      "loss": 0.6834,
+      "step": 6825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4030361071976908e-05,
+      "loss": 0.6303,
+      "step": 6826
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4028261402621132e-05,
+      "loss": 0.6625,
+      "step": 6827
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4026161640006705e-05,
+      "loss": 0.6301,
+      "step": 6828
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4024061784230172e-05,
+      "loss": 0.6481,
+      "step": 6829
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4021961835388086e-05,
+      "loss": 0.6666,
+      "step": 6830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4019861793576995e-05,
+      "loss": 0.575,
+      "step": 6831
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4017761658893468e-05,
+      "loss": 0.6832,
+      "step": 6832
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4015661431434062e-05,
+      "loss": 0.6802,
+      "step": 6833
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4013561111295344e-05,
+      "loss": 0.6247,
+      "step": 6834
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.401146069857389e-05,
+      "loss": 0.6902,
+      "step": 6835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.400936019336627e-05,
+      "loss": 0.7067,
+      "step": 6836
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4007259595769072e-05,
+      "loss": 0.7413,
+      "step": 6837
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4005158905878873e-05,
+      "loss": 0.6167,
+      "step": 6838
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.400305812379227e-05,
+      "loss": 0.6869,
+      "step": 6839
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4000957249605848e-05,
+      "loss": 0.6663,
+      "step": 6840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3998856283416211e-05,
+      "loss": 0.6812,
+      "step": 6841
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3996755225319956e-05,
+      "loss": 0.6386,
+      "step": 6842
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3994654075413692e-05,
+      "loss": 0.6494,
+      "step": 6843
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3992552833794025e-05,
+      "loss": 0.6292,
+      "step": 6844
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3990451500557576e-05,
+      "loss": 0.6156,
+      "step": 6845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3988350075800956e-05,
+      "loss": 0.6744,
+      "step": 6846
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3986248559620792e-05,
+      "loss": 0.6462,
+      "step": 6847
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3984146952113708e-05,
+      "loss": 0.6452,
+      "step": 6848
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.398204525337634e-05,
+      "loss": 0.6446,
+      "step": 6849
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3979943463505318e-05,
+      "loss": 0.6632,
+      "step": 6850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3977841582597287e-05,
+      "loss": 0.6573,
+      "step": 6851
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3975739610748886e-05,
+      "loss": 0.7102,
+      "step": 6852
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3973637548056762e-05,
+      "loss": 0.717,
+      "step": 6853
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3971535394617572e-05,
+      "loss": 0.7019,
+      "step": 6854
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3969433150527967e-05,
+      "loss": 0.6271,
+      "step": 6855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3967330815884611e-05,
+      "loss": 0.7073,
+      "step": 6856
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3965228390784166e-05,
+      "loss": 0.6748,
+      "step": 6857
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3963125875323304e-05,
+      "loss": 0.6231,
+      "step": 6858
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3961023269598695e-05,
+      "loss": 0.6884,
+      "step": 6859
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3958920573707015e-05,
+      "loss": 0.6445,
+      "step": 6860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3956817787744947e-05,
+      "loss": 0.6777,
+      "step": 6861
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3954714911809177e-05,
+      "loss": 0.7088,
+      "step": 6862
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3952611945996395e-05,
+      "loss": 0.6322,
+      "step": 6863
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.395050889040329e-05,
+      "loss": 0.6981,
+      "step": 6864
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3948405745126568e-05,
+      "loss": 0.6835,
+      "step": 6865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3946302510262921e-05,
+      "loss": 0.6296,
+      "step": 6866
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.394419918590906e-05,
+      "loss": 0.6488,
+      "step": 6867
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3942095772161692e-05,
+      "loss": 0.6954,
+      "step": 6868
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3939992269117536e-05,
+      "loss": 0.65,
+      "step": 6869
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3937888676873308e-05,
+      "loss": 0.6508,
+      "step": 6870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3935784995525732e-05,
+      "loss": 0.709,
+      "step": 6871
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3933681225171528e-05,
+      "loss": 0.6382,
+      "step": 6872
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3931577365907433e-05,
+      "loss": 0.6073,
+      "step": 6873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.392947341783018e-05,
+      "loss": 0.7175,
+      "step": 6874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3927369381036508e-05,
+      "loss": 0.6561,
+      "step": 6875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3925265255623154e-05,
+      "loss": 0.7005,
+      "step": 6876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3923161041686876e-05,
+      "loss": 0.6161,
+      "step": 6877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3921056739324416e-05,
+      "loss": 0.6558,
+      "step": 6878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3918952348632534e-05,
+      "loss": 0.6392,
+      "step": 6879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3916847869707986e-05,
+      "loss": 0.6383,
+      "step": 6880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3914743302647537e-05,
+      "loss": 0.693,
+      "step": 6881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3912638647547954e-05,
+      "loss": 0.659,
+      "step": 6882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3910533904506003e-05,
+      "loss": 0.6717,
+      "step": 6883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3908429073618466e-05,
+      "loss": 0.6645,
+      "step": 6884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3906324154982118e-05,
+      "loss": 0.6864,
+      "step": 6885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3904219148693747e-05,
+      "loss": 0.7065,
+      "step": 6886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3902114054850135e-05,
+      "loss": 0.5932,
+      "step": 6887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3900008873548078e-05,
+      "loss": 0.6272,
+      "step": 6888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3897903604884369e-05,
+      "loss": 0.6305,
+      "step": 6889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.389579824895581e-05,
+      "loss": 0.6632,
+      "step": 6890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3893692805859198e-05,
+      "loss": 0.6296,
+      "step": 6891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3891587275691346e-05,
+      "loss": 0.6904,
+      "step": 6892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3889481658549063e-05,
+      "loss": 0.652,
+      "step": 6893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3887375954529167e-05,
+      "loss": 0.6645,
+      "step": 6894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3885270163728474e-05,
+      "loss": 0.7264,
+      "step": 6895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3883164286243813e-05,
+      "loss": 0.6559,
+      "step": 6896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3881058322172005e-05,
+      "loss": 0.7547,
+      "step": 6897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3878952271609887e-05,
+      "loss": 0.5883,
+      "step": 6898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3876846134654291e-05,
+      "loss": 0.6778,
+      "step": 6899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3874739911402053e-05,
+      "loss": 0.6834,
+      "step": 6900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3872633601950022e-05,
+      "loss": 0.6604,
+      "step": 6901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3870527206395044e-05,
+      "loss": 0.696,
+      "step": 6902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3868420724833971e-05,
+      "loss": 0.6754,
+      "step": 6903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3866314157363654e-05,
+      "loss": 0.6956,
+      "step": 6904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3864207504080958e-05,
+      "loss": 0.702,
+      "step": 6905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3862100765082739e-05,
+      "loss": 0.6779,
+      "step": 6906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3859993940465872e-05,
+      "loss": 0.7292,
+      "step": 6907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3857887030327221e-05,
+      "loss": 0.7098,
+      "step": 6908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3855780034763666e-05,
+      "loss": 0.623,
+      "step": 6909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3853672953872082e-05,
+      "loss": 0.6519,
+      "step": 6910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3851565787749358e-05,
+      "loss": 0.6444,
+      "step": 6911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.384945853649237e-05,
+      "loss": 0.7181,
+      "step": 6912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3847351200198017e-05,
+      "loss": 0.71,
+      "step": 6913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3845243778963193e-05,
+      "loss": 0.5904,
+      "step": 6914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3843136272884794e-05,
+      "loss": 0.7276,
+      "step": 6915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3841028682059723e-05,
+      "loss": 0.6719,
+      "step": 6916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3838921006584885e-05,
+      "loss": 0.6472,
+      "step": 6917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.383681324655719e-05,
+      "loss": 0.66,
+      "step": 6918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3834705402073554e-05,
+      "loss": 0.7412,
+      "step": 6919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3832597473230894e-05,
+      "loss": 0.7081,
+      "step": 6920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3830489460126133e-05,
+      "loss": 0.6783,
+      "step": 6921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3828381362856195e-05,
+      "loss": 0.6477,
+      "step": 6922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3826273181518008e-05,
+      "loss": 0.6378,
+      "step": 6923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3824164916208509e-05,
+      "loss": 0.7257,
+      "step": 6924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3822056567024633e-05,
+      "loss": 0.704,
+      "step": 6925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3819948134063319e-05,
+      "loss": 0.6472,
+      "step": 6926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3817839617421516e-05,
+      "loss": 0.73,
+      "step": 6927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3815731017196172e-05,
+      "loss": 0.6657,
+      "step": 6928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3813622333484235e-05,
+      "loss": 0.6941,
+      "step": 6929
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.381151356638267e-05,
+      "loss": 0.653,
+      "step": 6930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.380940471598843e-05,
+      "loss": 0.7079,
+      "step": 6931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3807295782398482e-05,
+      "loss": 0.6809,
+      "step": 6932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3805186765709795e-05,
+      "loss": 0.6868,
+      "step": 6933
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3803077666019335e-05,
+      "loss": 0.6158,
+      "step": 6934
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3800968483424085e-05,
+      "loss": 0.7188,
+      "step": 6935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.379885921802102e-05,
+      "loss": 0.6582,
+      "step": 6936
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3796749869907121e-05,
+      "loss": 0.665,
+      "step": 6937
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3794640439179383e-05,
+      "loss": 0.6659,
+      "step": 6938
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.379253092593479e-05,
+      "loss": 0.6972,
+      "step": 6939
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3790421330270338e-05,
+      "loss": 0.619,
+      "step": 6940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3788311652283029e-05,
+      "loss": 0.7213,
+      "step": 6941
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.378620189206986e-05,
+      "loss": 0.7039,
+      "step": 6942
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3784092049727838e-05,
+      "loss": 0.6818,
+      "step": 6943
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3781982125353971e-05,
+      "loss": 0.6432,
+      "step": 6944
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3779872119045282e-05,
+      "loss": 0.6764,
+      "step": 6945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3777762030898776e-05,
+      "loss": 0.7155,
+      "step": 6946
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3775651861011483e-05,
+      "loss": 0.6501,
+      "step": 6947
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3773541609480423e-05,
+      "loss": 0.7091,
+      "step": 6948
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3771431276402627e-05,
+      "loss": 0.6872,
+      "step": 6949
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3769320861875126e-05,
+      "loss": 0.6114,
+      "step": 6950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3767210365994954e-05,
+      "loss": 0.684,
+      "step": 6951
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3765099788859153e-05,
+      "loss": 0.7223,
+      "step": 6952
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3762989130564767e-05,
+      "loss": 0.6728,
+      "step": 6953
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3760878391208845e-05,
+      "loss": 0.6836,
+      "step": 6954
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3758767570888432e-05,
+      "loss": 0.661,
+      "step": 6955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3756656669700587e-05,
+      "loss": 0.6898,
+      "step": 6956
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3754545687742369e-05,
+      "loss": 0.6277,
+      "step": 6957
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3752434625110839e-05,
+      "loss": 0.7154,
+      "step": 6958
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.375032348190306e-05,
+      "loss": 0.6337,
+      "step": 6959
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3748212258216105e-05,
+      "loss": 0.706,
+      "step": 6960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3746100954147045e-05,
+      "loss": 0.7382,
+      "step": 6961
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.374398956979296e-05,
+      "loss": 0.6522,
+      "step": 6962
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.374187810525093e-05,
+      "loss": 0.6785,
+      "step": 6963
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3739766560618037e-05,
+      "loss": 0.6937,
+      "step": 6964
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.373765493599137e-05,
+      "loss": 0.7005,
+      "step": 6965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3735543231468023e-05,
+      "loss": 0.7166,
+      "step": 6966
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3733431447145087e-05,
+      "loss": 0.6962,
+      "step": 6967
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3731319583119662e-05,
+      "loss": 0.6139,
+      "step": 6968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.372920763948885e-05,
+      "loss": 0.6755,
+      "step": 6969
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.372709561634976e-05,
+      "loss": 0.6898,
+      "step": 6970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3724983513799505e-05,
+      "loss": 0.6011,
+      "step": 6971
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3722871331935193e-05,
+      "loss": 0.7028,
+      "step": 6972
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3720759070853944e-05,
+      "loss": 0.7514,
+      "step": 6973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3718646730652875e-05,
+      "loss": 0.7198,
+      "step": 6974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3716534311429117e-05,
+      "loss": 0.684,
+      "step": 6975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3714421813279792e-05,
+      "loss": 0.7025,
+      "step": 6976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3712309236302032e-05,
+      "loss": 0.6545,
+      "step": 6977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3710196580592978e-05,
+      "loss": 0.6822,
+      "step": 6978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3708083846249767e-05,
+      "loss": 0.6129,
+      "step": 6979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3705971033369538e-05,
+      "loss": 0.7146,
+      "step": 6980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3703858142049444e-05,
+      "loss": 0.6559,
+      "step": 6981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3701745172386627e-05,
+      "loss": 0.6962,
+      "step": 6982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3699632124478249e-05,
+      "loss": 0.6787,
+      "step": 6983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3697518998421457e-05,
+      "loss": 0.6779,
+      "step": 6984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3695405794313418e-05,
+      "loss": 0.6683,
+      "step": 6985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.36932925122513e-05,
+      "loss": 0.7238,
+      "step": 6986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3691179152332261e-05,
+      "loss": 0.6202,
+      "step": 6987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3689065714653482e-05,
+      "loss": 0.6668,
+      "step": 6988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3686952199312131e-05,
+      "loss": 0.7324,
+      "step": 6989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.368483860640539e-05,
+      "loss": 0.6283,
+      "step": 6990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3682724936030442e-05,
+      "loss": 0.6925,
+      "step": 6991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3680611188284471e-05,
+      "loss": 0.663,
+      "step": 6992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3678497363264665e-05,
+      "loss": 0.6321,
+      "step": 6993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.367638346106822e-05,
+      "loss": 0.6821,
+      "step": 6994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3674269481792328e-05,
+      "loss": 0.7023,
+      "step": 6995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3672155425534195e-05,
+      "loss": 0.7168,
+      "step": 6996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3670041292391018e-05,
+      "loss": 0.6792,
+      "step": 6997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.366792708246001e-05,
+      "loss": 0.7072,
+      "step": 6998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3665812795838378e-05,
+      "loss": 0.6636,
+      "step": 6999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3663698432623337e-05,
+      "loss": 0.6665,
+      "step": 7000
+    },
+    {
+      "epoch": 0.46,
+      "eval_text_loss": 0.6616618037223816,
+      "eval_text_runtime": 1.8052,
+      "eval_text_samples_per_second": 40.993,
+      "eval_text_steps_per_second": 1.108,
+      "step": 7000
+    },
+    {
+      "epoch": 0.46,
+      "eval_svit_loss": 0.8204585909843445,
+      "eval_svit_runtime": 6.6339,
+      "eval_svit_samples_per_second": 123.005,
+      "eval_svit_steps_per_second": 1.96,
+      "step": 7000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3661583992912104e-05,
+      "loss": 0.6266,
+      "step": 7001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3659469476801899e-05,
+      "loss": 0.6128,
+      "step": 7002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.365735488438995e-05,
+      "loss": 0.6458,
+      "step": 7003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3655240215773482e-05,
+      "loss": 0.6391,
+      "step": 7004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3653125471049728e-05,
+      "loss": 0.7594,
+      "step": 7005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3651010650315923e-05,
+      "loss": 0.6286,
+      "step": 7006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3648895753669307e-05,
+      "loss": 0.7507,
+      "step": 7007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3646780781207118e-05,
+      "loss": 0.6346,
+      "step": 7008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3644665733026608e-05,
+      "loss": 0.719,
+      "step": 7009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3642550609225017e-05,
+      "loss": 0.6508,
+      "step": 7010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3640435409899606e-05,
+      "loss": 0.6562,
+      "step": 7011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3638320135147626e-05,
+      "loss": 0.5991,
+      "step": 7012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3636204785066342e-05,
+      "loss": 0.7453,
+      "step": 7013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.363408935975301e-05,
+      "loss": 0.616,
+      "step": 7014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3631973859304902e-05,
+      "loss": 0.6969,
+      "step": 7015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3629858283819285e-05,
+      "loss": 0.6545,
+      "step": 7016
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3627742633393433e-05,
+      "loss": 0.6428,
+      "step": 7017
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.362562690812462e-05,
+      "loss": 0.6911,
+      "step": 7018
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.362351110811013e-05,
+      "loss": 0.6566,
+      "step": 7019
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3621395233447247e-05,
+      "loss": 0.6335,
+      "step": 7020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3619279284233255e-05,
+      "loss": 0.6446,
+      "step": 7021
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3617163260565449e-05,
+      "loss": 0.6895,
+      "step": 7022
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3615047162541118e-05,
+      "loss": 0.6912,
+      "step": 7023
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3612930990257562e-05,
+      "loss": 0.683,
+      "step": 7024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.361081474381208e-05,
+      "loss": 0.6851,
+      "step": 7025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.360869842330198e-05,
+      "loss": 0.6714,
+      "step": 7026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3606582028824564e-05,
+      "loss": 0.7487,
+      "step": 7027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3604465560477145e-05,
+      "loss": 0.6656,
+      "step": 7028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.360234901835704e-05,
+      "loss": 0.6876,
+      "step": 7029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3600232402561566e-05,
+      "loss": 0.6669,
+      "step": 7030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.359811571318804e-05,
+      "loss": 0.6681,
+      "step": 7031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3595998950333793e-05,
+      "loss": 0.7179,
+      "step": 7032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.359388211409615e-05,
+      "loss": 0.7367,
+      "step": 7033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.359176520457244e-05,
+      "loss": 0.6791,
+      "step": 7034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3589648221859998e-05,
+      "loss": 0.7292,
+      "step": 7035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3587531166056164e-05,
+      "loss": 0.6623,
+      "step": 7036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3585414037258281e-05,
+      "loss": 0.6777,
+      "step": 7037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.358329683556369e-05,
+      "loss": 0.687,
+      "step": 7038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3581179561069744e-05,
+      "loss": 0.6473,
+      "step": 7039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3579062213873788e-05,
+      "loss": 0.6436,
+      "step": 7040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.357694479407318e-05,
+      "loss": 0.6686,
+      "step": 7041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.357482730176528e-05,
+      "loss": 0.6736,
+      "step": 7042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3572709737047446e-05,
+      "loss": 0.6205,
+      "step": 7043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3570592100017046e-05,
+      "loss": 0.7024,
+      "step": 7044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3568474390771445e-05,
+      "loss": 0.7223,
+      "step": 7045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3566356609408014e-05,
+      "loss": 0.6571,
+      "step": 7046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3564238756024133e-05,
+      "loss": 0.7259,
+      "step": 7047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3562120830717175e-05,
+      "loss": 0.5964,
+      "step": 7048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3560002833584525e-05,
+      "loss": 0.7728,
+      "step": 7049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3557884764723565e-05,
+      "loss": 0.6079,
+      "step": 7050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3555766624231681e-05,
+      "loss": 0.6601,
+      "step": 7051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3553648412206268e-05,
+      "loss": 0.6807,
+      "step": 7052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3551530128744721e-05,
+      "loss": 0.6783,
+      "step": 7053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3549411773944435e-05,
+      "loss": 0.6128,
+      "step": 7054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3547293347902812e-05,
+      "loss": 0.5945,
+      "step": 7055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.354517485071726e-05,
+      "loss": 0.7045,
+      "step": 7056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.354305628248518e-05,
+      "loss": 0.6929,
+      "step": 7057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3540937643303989e-05,
+      "loss": 0.6931,
+      "step": 7058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3538818933271095e-05,
+      "loss": 0.705,
+      "step": 7059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3536700152483924e-05,
+      "loss": 0.6868,
+      "step": 7060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3534581301039889e-05,
+      "loss": 0.715,
+      "step": 7061
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3532462379036414e-05,
+      "loss": 0.7254,
+      "step": 7062
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.353034338657093e-05,
+      "loss": 0.6733,
+      "step": 7063
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.352822432374087e-05,
+      "loss": 0.6698,
+      "step": 7064
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3526105190643658e-05,
+      "loss": 0.6734,
+      "step": 7065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3523985987376744e-05,
+      "loss": 0.7075,
+      "step": 7066
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3521866714037559e-05,
+      "loss": 0.6146,
+      "step": 7067
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3519747370723547e-05,
+      "loss": 0.749,
+      "step": 7068
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3517627957532154e-05,
+      "loss": 0.6114,
+      "step": 7069
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.351550847456083e-05,
+      "loss": 0.6449,
+      "step": 7070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3513388921907035e-05,
+      "loss": 0.6343,
+      "step": 7071
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3511269299668216e-05,
+      "loss": 0.7089,
+      "step": 7072
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.350914960794184e-05,
+      "loss": 0.7198,
+      "step": 7073
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3507029846825364e-05,
+      "loss": 0.6731,
+      "step": 7074
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3504910016416255e-05,
+      "loss": 0.7287,
+      "step": 7075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3502790116811981e-05,
+      "loss": 0.7319,
+      "step": 7076
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3500670148110017e-05,
+      "loss": 0.6366,
+      "step": 7077
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3498550110407836e-05,
+      "loss": 0.671,
+      "step": 7078
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3496430003802921e-05,
+      "loss": 0.6336,
+      "step": 7079
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3494309828392746e-05,
+      "loss": 0.6148,
+      "step": 7080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3492189584274803e-05,
+      "loss": 0.6739,
+      "step": 7081
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3490069271546576e-05,
+      "loss": 0.7188,
+      "step": 7082
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3487948890305555e-05,
+      "loss": 0.6341,
+      "step": 7083
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.348582844064924e-05,
+      "loss": 0.6281,
+      "step": 7084
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3483707922675122e-05,
+      "loss": 0.6291,
+      "step": 7085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3481587336480704e-05,
+      "loss": 0.6573,
+      "step": 7086
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.347946668216349e-05,
+      "loss": 0.6937,
+      "step": 7087
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.347734595982099e-05,
+      "loss": 0.6926,
+      "step": 7088
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3475225169550706e-05,
+      "loss": 0.6717,
+      "step": 7089
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3473104311450161e-05,
+      "loss": 0.7132,
+      "step": 7090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3470983385616864e-05,
+      "loss": 0.6672,
+      "step": 7091
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3468862392148338e-05,
+      "loss": 0.6807,
+      "step": 7092
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3466741331142101e-05,
+      "loss": 0.6343,
+      "step": 7093
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3464620202695682e-05,
+      "loss": 0.624,
+      "step": 7094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3462499006906608e-05,
+      "loss": 0.6881,
+      "step": 7095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3460377743872416e-05,
+      "loss": 0.7091,
+      "step": 7096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3458256413690633e-05,
+      "loss": 0.6899,
+      "step": 7097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3456135016458804e-05,
+      "loss": 0.6682,
+      "step": 7098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3454013552274464e-05,
+      "loss": 0.7053,
+      "step": 7099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3451892021235162e-05,
+      "loss": 0.6703,
+      "step": 7100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3449770423438442e-05,
+      "loss": 0.6167,
+      "step": 7101
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3447648758981854e-05,
+      "loss": 0.6308,
+      "step": 7102
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.344552702796295e-05,
+      "loss": 0.6673,
+      "step": 7103
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.344340523047929e-05,
+      "loss": 0.719,
+      "step": 7104
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3441283366628436e-05,
+      "loss": 0.6681,
+      "step": 7105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3439161436507943e-05,
+      "loss": 0.6729,
+      "step": 7106
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3437039440215383e-05,
+      "loss": 0.7408,
+      "step": 7107
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3434917377848317e-05,
+      "loss": 0.6897,
+      "step": 7108
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3432795249504326e-05,
+      "loss": 0.6965,
+      "step": 7109
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3430673055280976e-05,
+      "loss": 0.6585,
+      "step": 7110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.342855079527585e-05,
+      "loss": 0.6445,
+      "step": 7111
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3426428469586525e-05,
+      "loss": 0.6478,
+      "step": 7112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3424306078310589e-05,
+      "loss": 0.6344,
+      "step": 7113
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3422183621545626e-05,
+      "loss": 0.7041,
+      "step": 7114
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3420061099389227e-05,
+      "loss": 0.6757,
+      "step": 7115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3417938511938983e-05,
+      "loss": 0.7339,
+      "step": 7116
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3415815859292491e-05,
+      "loss": 0.7195,
+      "step": 7117
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3413693141547353e-05,
+      "loss": 0.6878,
+      "step": 7118
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3411570358801163e-05,
+      "loss": 0.6563,
+      "step": 7119
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3409447511151528e-05,
+      "loss": 0.6523,
+      "step": 7120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3407324598696059e-05,
+      "loss": 0.6815,
+      "step": 7121
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3405201621532366e-05,
+      "loss": 0.684,
+      "step": 7122
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3403078579758062e-05,
+      "loss": 0.7008,
+      "step": 7123
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3400955473470765e-05,
+      "loss": 0.6405,
+      "step": 7124
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.339883230276809e-05,
+      "loss": 0.6516,
+      "step": 7125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3396709067747663e-05,
+      "loss": 0.6432,
+      "step": 7126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.339458576850711e-05,
+      "loss": 0.6745,
+      "step": 7127
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3392462405144058e-05,
+      "loss": 0.7256,
+      "step": 7128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3390338977756138e-05,
+      "loss": 0.7379,
+      "step": 7129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3388215486440987e-05,
+      "loss": 0.6077,
+      "step": 7130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3386091931296241e-05,
+      "loss": 0.6938,
+      "step": 7131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3383968312419539e-05,
+      "loss": 0.6546,
+      "step": 7132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3381844629908525e-05,
+      "loss": 0.6807,
+      "step": 7133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3379720883860845e-05,
+      "loss": 0.6415,
+      "step": 7134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3377597074374149e-05,
+      "loss": 0.6666,
+      "step": 7135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3375473201546084e-05,
+      "loss": 0.6554,
+      "step": 7136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.337334926547431e-05,
+      "loss": 0.6512,
+      "step": 7137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3371225266256486e-05,
+      "loss": 0.66,
+      "step": 7138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.336910120399027e-05,
+      "loss": 0.6369,
+      "step": 7139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3366977078773323e-05,
+      "loss": 0.7272,
+      "step": 7140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3364852890703317e-05,
+      "loss": 0.6285,
+      "step": 7141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3362728639877918e-05,
+      "loss": 0.6363,
+      "step": 7142
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3360604326394796e-05,
+      "loss": 0.7038,
+      "step": 7143
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3358479950351631e-05,
+      "loss": 0.7475,
+      "step": 7144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3356355511846098e-05,
+      "loss": 0.6035,
+      "step": 7145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3354231010975875e-05,
+      "loss": 0.6497,
+      "step": 7146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3352106447838656e-05,
+      "loss": 0.7038,
+      "step": 7147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3349981822532117e-05,
+      "loss": 0.6364,
+      "step": 7148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3347857135153951e-05,
+      "loss": 0.6587,
+      "step": 7149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.334573238580185e-05,
+      "loss": 0.6293,
+      "step": 7150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3343607574573512e-05,
+      "loss": 0.6694,
+      "step": 7151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3341482701566633e-05,
+      "loss": 0.6362,
+      "step": 7152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3339357766878911e-05,
+      "loss": 0.6248,
+      "step": 7153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3337232770608052e-05,
+      "loss": 0.6804,
+      "step": 7154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3335107712851762e-05,
+      "loss": 0.6695,
+      "step": 7155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3332982593707753e-05,
+      "loss": 0.6609,
+      "step": 7156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3330857413273732e-05,
+      "loss": 0.6345,
+      "step": 7157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3328732171647421e-05,
+      "loss": 0.6417,
+      "step": 7158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.332660686892653e-05,
+      "loss": 0.7066,
+      "step": 7159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3324481505208785e-05,
+      "loss": 0.6771,
+      "step": 7160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.332235608059191e-05,
+      "loss": 0.7508,
+      "step": 7161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3320230595173625e-05,
+      "loss": 0.6912,
+      "step": 7162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3318105049051662e-05,
+      "loss": 0.7255,
+      "step": 7163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3315979442323756e-05,
+      "loss": 0.7677,
+      "step": 7164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3313853775087639e-05,
+      "loss": 0.6782,
+      "step": 7165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3311728047441048e-05,
+      "loss": 0.6746,
+      "step": 7166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3309602259481726e-05,
+      "loss": 0.6554,
+      "step": 7167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3307476411307409e-05,
+      "loss": 0.5944,
+      "step": 7168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3305350503015848e-05,
+      "loss": 0.6853,
+      "step": 7169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3303224534704792e-05,
+      "loss": 0.6396,
+      "step": 7170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3301098506471992e-05,
+      "loss": 0.6907,
+      "step": 7171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3298972418415197e-05,
+      "loss": 0.6382,
+      "step": 7172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.329684627063217e-05,
+      "loss": 0.6873,
+      "step": 7173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3294720063220667e-05,
+      "loss": 0.6223,
+      "step": 7174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3292593796278451e-05,
+      "loss": 0.7094,
+      "step": 7175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3290467469903285e-05,
+      "loss": 0.6591,
+      "step": 7176
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3288341084192944e-05,
+      "loss": 0.6704,
+      "step": 7177
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3286214639245187e-05,
+      "loss": 0.7,
+      "step": 7178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3284088135157792e-05,
+      "loss": 0.6751,
+      "step": 7179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3281961572028538e-05,
+      "loss": 0.6371,
+      "step": 7180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3279834949955203e-05,
+      "loss": 0.6612,
+      "step": 7181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3277708269035565e-05,
+      "loss": 0.5473,
+      "step": 7182
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3275581529367411e-05,
+      "loss": 0.6803,
+      "step": 7183
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3273454731048524e-05,
+      "loss": 0.7047,
+      "step": 7184
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3271327874176696e-05,
+      "loss": 0.6836,
+      "step": 7185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3269200958849717e-05,
+      "loss": 0.6374,
+      "step": 7186
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3267073985165384e-05,
+      "loss": 0.6356,
+      "step": 7187
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3264946953221495e-05,
+      "loss": 0.7011,
+      "step": 7188
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3262819863115848e-05,
+      "loss": 0.6535,
+      "step": 7189
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3260692714946248e-05,
+      "loss": 0.6475,
+      "step": 7190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3258565508810494e-05,
+      "loss": 0.612,
+      "step": 7191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3256438244806404e-05,
+      "loss": 0.6308,
+      "step": 7192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3254310923031781e-05,
+      "loss": 0.6814,
+      "step": 7193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3252183543584443e-05,
+      "loss": 0.6361,
+      "step": 7194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3250056106562201e-05,
+      "loss": 0.7184,
+      "step": 7195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3247928612062876e-05,
+      "loss": 0.6298,
+      "step": 7196
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3245801060184294e-05,
+      "loss": 0.6424,
+      "step": 7197
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3243673451024275e-05,
+      "loss": 0.6435,
+      "step": 7198
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3241545784680643e-05,
+      "loss": 0.6071,
+      "step": 7199
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3239418061251233e-05,
+      "loss": 0.6509,
+      "step": 7200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3237290280833874e-05,
+      "loss": 0.7556,
+      "step": 7201
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.32351624435264e-05,
+      "loss": 0.6427,
+      "step": 7202
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3233034549426644e-05,
+      "loss": 0.6528,
+      "step": 7203
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3230906598632453e-05,
+      "loss": 0.6135,
+      "step": 7204
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3228778591241668e-05,
+      "loss": 0.6166,
+      "step": 7205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3226650527352132e-05,
+      "loss": 0.6085,
+      "step": 7206
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3224522407061695e-05,
+      "loss": 0.6342,
+      "step": 7207
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3222394230468201e-05,
+      "loss": 0.6409,
+      "step": 7208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.322026599766951e-05,
+      "loss": 0.6214,
+      "step": 7209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3218137708763471e-05,
+      "loss": 0.7515,
+      "step": 7210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.321600936384795e-05,
+      "loss": 0.7153,
+      "step": 7211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.32138809630208e-05,
+      "loss": 0.6542,
+      "step": 7212
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3211752506379884e-05,
+      "loss": 0.6298,
+      "step": 7213
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3209623994023073e-05,
+      "loss": 0.69,
+      "step": 7214
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3207495426048231e-05,
+      "loss": 0.644,
+      "step": 7215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.320536680255323e-05,
+      "loss": 0.6621,
+      "step": 7216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3203238123635946e-05,
+      "loss": 0.7248,
+      "step": 7217
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.320110938939425e-05,
+      "loss": 0.7326,
+      "step": 7218
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3198980599926024e-05,
+      "loss": 0.6648,
+      "step": 7219
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3196851755329144e-05,
+      "loss": 0.6213,
+      "step": 7220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3194722855701495e-05,
+      "loss": 0.6075,
+      "step": 7221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.319259390114097e-05,
+      "loss": 0.7146,
+      "step": 7222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3190464891745448e-05,
+      "loss": 0.7087,
+      "step": 7223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3188335827612831e-05,
+      "loss": 0.6965,
+      "step": 7224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3186206708840998e-05,
+      "loss": 0.6676,
+      "step": 7225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3184077535527857e-05,
+      "loss": 0.6322,
+      "step": 7226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3181948307771301e-05,
+      "loss": 0.7342,
+      "step": 7227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3179819025669234e-05,
+      "loss": 0.6892,
+      "step": 7228
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3177689689319555e-05,
+      "loss": 0.6914,
+      "step": 7229
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3175560298820173e-05,
+      "loss": 0.6747,
+      "step": 7230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3173430854268996e-05,
+      "loss": 0.7338,
+      "step": 7231
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.317130135576394e-05,
+      "loss": 0.7005,
+      "step": 7232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3169171803402909e-05,
+      "loss": 0.6904,
+      "step": 7233
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3167042197283825e-05,
+      "loss": 0.6082,
+      "step": 7234
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3164912537504609e-05,
+      "loss": 0.6433,
+      "step": 7235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3162782824163174e-05,
+      "loss": 0.5662,
+      "step": 7236
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3160653057357448e-05,
+      "loss": 0.6926,
+      "step": 7237
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3158523237185357e-05,
+      "loss": 0.6776,
+      "step": 7238
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.315639336374483e-05,
+      "loss": 0.6723,
+      "step": 7239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3154263437133795e-05,
+      "loss": 0.6506,
+      "step": 7240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3152133457450188e-05,
+      "loss": 0.7122,
+      "step": 7241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3150003424791942e-05,
+      "loss": 0.6643,
+      "step": 7242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3147873339256998e-05,
+      "loss": 0.6005,
+      "step": 7243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3145743200943293e-05,
+      "loss": 0.6624,
+      "step": 7244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3143613009948773e-05,
+      "loss": 0.6884,
+      "step": 7245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.314148276637138e-05,
+      "loss": 0.6791,
+      "step": 7246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3139352470309066e-05,
+      "loss": 0.6809,
+      "step": 7247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3137222121859776e-05,
+      "loss": 0.7276,
+      "step": 7248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3135091721121469e-05,
+      "loss": 0.7526,
+      "step": 7249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3132961268192094e-05,
+      "loss": 0.7365,
+      "step": 7250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3130830763169613e-05,
+      "loss": 0.6975,
+      "step": 7251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3128700206151985e-05,
+      "loss": 0.6684,
+      "step": 7252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3126569597237167e-05,
+      "loss": 0.6876,
+      "step": 7253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3124438936523127e-05,
+      "loss": 0.6698,
+      "step": 7254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3122308224107833e-05,
+      "loss": 0.6762,
+      "step": 7255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3120177460089256e-05,
+      "loss": 0.702,
+      "step": 7256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3118046644565362e-05,
+      "loss": 0.6606,
+      "step": 7257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3115915777634134e-05,
+      "loss": 0.6531,
+      "step": 7258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3113784859393539e-05,
+      "loss": 0.7087,
+      "step": 7259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.311165388994156e-05,
+      "loss": 0.6108,
+      "step": 7260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3109522869376176e-05,
+      "loss": 0.6443,
+      "step": 7261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3107391797795372e-05,
+      "loss": 0.6836,
+      "step": 7262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3105260675297136e-05,
+      "loss": 0.7165,
+      "step": 7263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3103129501979454e-05,
+      "loss": 0.6988,
+      "step": 7264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3100998277940316e-05,
+      "loss": 0.7209,
+      "step": 7265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3098867003277717e-05,
+      "loss": 0.688,
+      "step": 7266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3096735678089646e-05,
+      "loss": 0.6967,
+      "step": 7267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3094604302474111e-05,
+      "loss": 0.5907,
+      "step": 7268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3092472876529105e-05,
+      "loss": 0.6484,
+      "step": 7269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3090341400352628e-05,
+      "loss": 0.6532,
+      "step": 7270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3088209874042687e-05,
+      "loss": 0.7046,
+      "step": 7271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3086078297697291e-05,
+      "loss": 0.6385,
+      "step": 7272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.308394667141445e-05,
+      "loss": 0.644,
+      "step": 7273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3081814995292167e-05,
+      "loss": 0.6587,
+      "step": 7274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3079683269428464e-05,
+      "loss": 0.6093,
+      "step": 7275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3077551493921353e-05,
+      "loss": 0.6351,
+      "step": 7276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3075419668868856e-05,
+      "loss": 0.6573,
+      "step": 7277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3073287794368989e-05,
+      "loss": 0.7345,
+      "step": 7278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3071155870519776e-05,
+      "loss": 0.7105,
+      "step": 7279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3069023897419242e-05,
+      "loss": 0.6591,
+      "step": 7280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3066891875165414e-05,
+      "loss": 0.7111,
+      "step": 7281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3064759803856324e-05,
+      "loss": 0.6629,
+      "step": 7282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3062627683590003e-05,
+      "loss": 0.6638,
+      "step": 7283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3060495514464484e-05,
+      "loss": 0.6321,
+      "step": 7284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3058363296577803e-05,
+      "loss": 0.6521,
+      "step": 7285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3056231030028e-05,
+      "loss": 0.6595,
+      "step": 7286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.305409871491311e-05,
+      "loss": 0.5689,
+      "step": 7287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3051966351331184e-05,
+      "loss": 0.6576,
+      "step": 7288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3049833939380262e-05,
+      "loss": 0.6489,
+      "step": 7289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3047701479158396e-05,
+      "loss": 0.6197,
+      "step": 7290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.304556897076363e-05,
+      "loss": 0.6479,
+      "step": 7291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3043436414294024e-05,
+      "loss": 0.663,
+      "step": 7292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3041303809847623e-05,
+      "loss": 0.6268,
+      "step": 7293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.303917115752249e-05,
+      "loss": 0.6588,
+      "step": 7294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3037038457416677e-05,
+      "loss": 0.6802,
+      "step": 7295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3034905709628252e-05,
+      "loss": 0.6604,
+      "step": 7296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3032772914255271e-05,
+      "loss": 0.6423,
+      "step": 7297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3030640071395805e-05,
+      "loss": 0.637,
+      "step": 7298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3028507181147918e-05,
+      "loss": 0.6789,
+      "step": 7299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3026374243609683e-05,
+      "loss": 0.6224,
+      "step": 7300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3024241258879166e-05,
+      "loss": 0.7218,
+      "step": 7301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3022108227054446e-05,
+      "loss": 0.6803,
+      "step": 7302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3019975148233594e-05,
+      "loss": 0.6297,
+      "step": 7303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3017842022514692e-05,
+      "loss": 0.6734,
+      "step": 7304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3015708849995818e-05,
+      "loss": 0.7369,
+      "step": 7305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3013575630775055e-05,
+      "loss": 0.6868,
+      "step": 7306
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.301144236495049e-05,
+      "loss": 0.6033,
+      "step": 7307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3009309052620205e-05,
+      "loss": 0.688,
+      "step": 7308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3007175693882294e-05,
+      "loss": 0.6569,
+      "step": 7309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3005042288834843e-05,
+      "loss": 0.6771,
+      "step": 7310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3002908837575952e-05,
+      "loss": 0.7247,
+      "step": 7311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3000775340203708e-05,
+      "loss": 0.6326,
+      "step": 7312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2998641796816211e-05,
+      "loss": 0.6746,
+      "step": 7313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2996508207511565e-05,
+      "loss": 0.6775,
+      "step": 7314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2994374572387868e-05,
+      "loss": 0.6409,
+      "step": 7315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2992240891543221e-05,
+      "loss": 0.7284,
+      "step": 7316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2990107165075737e-05,
+      "loss": 0.6728,
+      "step": 7317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.298797339308352e-05,
+      "loss": 0.6914,
+      "step": 7318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2985839575664676e-05,
+      "loss": 0.6783,
+      "step": 7319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2983705712917324e-05,
+      "loss": 0.6945,
+      "step": 7320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2981571804939572e-05,
+      "loss": 0.6758,
+      "step": 7321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.297943785182954e-05,
+      "loss": 0.71,
+      "step": 7322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2977303853685343e-05,
+      "loss": 0.6624,
+      "step": 7323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2975169810605109e-05,
+      "loss": 0.7408,
+      "step": 7324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2973035722686952e-05,
+      "loss": 0.6077,
+      "step": 7325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2970901590029001e-05,
+      "loss": 0.7243,
+      "step": 7326
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2968767412729382e-05,
+      "loss": 0.6592,
+      "step": 7327
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2966633190886221e-05,
+      "loss": 0.7359,
+      "step": 7328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2964498924597652e-05,
+      "loss": 0.6891,
+      "step": 7329
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2962364613961806e-05,
+      "loss": 0.6956,
+      "step": 7330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2960230259076818e-05,
+      "loss": 0.7336,
+      "step": 7331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2958095860040827e-05,
+      "loss": 0.6474,
+      "step": 7332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2955961416951966e-05,
+      "loss": 0.6666,
+      "step": 7333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2953826929908381e-05,
+      "loss": 0.6707,
+      "step": 7334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2951692399008217e-05,
+      "loss": 0.7182,
+      "step": 7335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.294955782434961e-05,
+      "loss": 0.7123,
+      "step": 7336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2947423206030717e-05,
+      "loss": 0.6861,
+      "step": 7337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.294528854414968e-05,
+      "loss": 0.7521,
+      "step": 7338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2943153838804649e-05,
+      "loss": 0.5872,
+      "step": 7339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2941019090093782e-05,
+      "loss": 0.6622,
+      "step": 7340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2938884298115235e-05,
+      "loss": 0.675,
+      "step": 7341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2936749462967157e-05,
+      "loss": 0.6654,
+      "step": 7342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2934614584747716e-05,
+      "loss": 0.7275,
+      "step": 7343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2932479663555065e-05,
+      "loss": 0.6237,
+      "step": 7344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2930344699487371e-05,
+      "loss": 0.6904,
+      "step": 7345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2928209692642799e-05,
+      "loss": 0.6916,
+      "step": 7346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2926074643119513e-05,
+      "loss": 0.6829,
+      "step": 7347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2923939551015685e-05,
+      "loss": 0.6762,
+      "step": 7348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2921804416429484e-05,
+      "loss": 0.662,
+      "step": 7349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2919669239459082e-05,
+      "loss": 0.6683,
+      "step": 7350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2917534020202658e-05,
+      "loss": 0.6514,
+      "step": 7351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2915398758758383e-05,
+      "loss": 0.6234,
+      "step": 7352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2913263455224436e-05,
+      "loss": 0.6699,
+      "step": 7353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2911128109699002e-05,
+      "loss": 0.642,
+      "step": 7354
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2908992722280254e-05,
+      "loss": 0.74,
+      "step": 7355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2906857293066389e-05,
+      "loss": 0.6601,
+      "step": 7356
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2904721822155584e-05,
+      "loss": 0.6036,
+      "step": 7357
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.290258630964603e-05,
+      "loss": 0.6723,
+      "step": 7358
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2900450755635917e-05,
+      "loss": 0.6558,
+      "step": 7359
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2898315160223435e-05,
+      "loss": 0.7144,
+      "step": 7360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2896179523506781e-05,
+      "loss": 0.622,
+      "step": 7361
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.289404384558415e-05,
+      "loss": 0.6756,
+      "step": 7362
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2891908126553739e-05,
+      "loss": 0.7004,
+      "step": 7363
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2889772366513745e-05,
+      "loss": 0.7058,
+      "step": 7364
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2887636565562371e-05,
+      "loss": 0.6324,
+      "step": 7365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2885500723797826e-05,
+      "loss": 0.6796,
+      "step": 7366
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2883364841318307e-05,
+      "loss": 0.6301,
+      "step": 7367
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2881228918222024e-05,
+      "loss": 0.6801,
+      "step": 7368
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2879092954607189e-05,
+      "loss": 0.6415,
+      "step": 7369
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2876956950572005e-05,
+      "loss": 0.7029,
+      "step": 7370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2874820906214693e-05,
+      "loss": 0.7235,
+      "step": 7371
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2872684821633462e-05,
+      "loss": 0.6333,
+      "step": 7372
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2870548696926532e-05,
+      "loss": 0.6799,
+      "step": 7373
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.286841253219212e-05,
+      "loss": 0.7007,
+      "step": 7374
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2866276327528444e-05,
+      "loss": 0.658,
+      "step": 7375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2864140083033727e-05,
+      "loss": 0.682,
+      "step": 7376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2862003798806194e-05,
+      "loss": 0.6361,
+      "step": 7377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2859867474944069e-05,
+      "loss": 0.6674,
+      "step": 7378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2857731111545581e-05,
+      "loss": 0.7048,
+      "step": 7379
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2855594708708958e-05,
+      "loss": 0.6949,
+      "step": 7380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2853458266532429e-05,
+      "loss": 0.7233,
+      "step": 7381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2851321785114228e-05,
+      "loss": 0.5617,
+      "step": 7382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2849185264552592e-05,
+      "loss": 0.6399,
+      "step": 7383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2847048704945754e-05,
+      "loss": 0.6564,
+      "step": 7384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2844912106391957e-05,
+      "loss": 0.6575,
+      "step": 7385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2842775468989437e-05,
+      "loss": 0.6958,
+      "step": 7386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2840638792836434e-05,
+      "loss": 0.7084,
+      "step": 7387
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2838502078031194e-05,
+      "loss": 0.6286,
+      "step": 7388
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2836365324671963e-05,
+      "loss": 0.6447,
+      "step": 7389
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2834228532856989e-05,
+      "loss": 0.6317,
+      "step": 7390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2832091702684517e-05,
+      "loss": 0.6288,
+      "step": 7391
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.28299548342528e-05,
+      "loss": 0.6047,
+      "step": 7392
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2827817927660092e-05,
+      "loss": 0.6543,
+      "step": 7393
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2825680983004643e-05,
+      "loss": 0.6863,
+      "step": 7394
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2823544000384711e-05,
+      "loss": 0.6049,
+      "step": 7395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2821406979898556e-05,
+      "loss": 0.6617,
+      "step": 7396
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2819269921644432e-05,
+      "loss": 0.6705,
+      "step": 7397
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2817132825720604e-05,
+      "loss": 0.7262,
+      "step": 7398
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2814995692225333e-05,
+      "loss": 0.7018,
+      "step": 7399
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2812858521256887e-05,
+      "loss": 0.5966,
+      "step": 7400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.281072131291353e-05,
+      "loss": 0.6886,
+      "step": 7401
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.280858406729353e-05,
+      "loss": 0.6697,
+      "step": 7402
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2806446784495157e-05,
+      "loss": 0.7012,
+      "step": 7403
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2804309464616677e-05,
+      "loss": 0.6371,
+      "step": 7404
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2802172107756371e-05,
+      "loss": 0.7037,
+      "step": 7405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.280003471401251e-05,
+      "loss": 0.6924,
+      "step": 7406
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2797897283483374e-05,
+      "loss": 0.7612,
+      "step": 7407
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2795759816267235e-05,
+      "loss": 0.6777,
+      "step": 7408
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.279362231246238e-05,
+      "loss": 0.6911,
+      "step": 7409
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2791484772167084e-05,
+      "loss": 0.6581,
+      "step": 7410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2789347195479636e-05,
+      "loss": 0.6976,
+      "step": 7411
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2787209582498315e-05,
+      "loss": 0.6852,
+      "step": 7412
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2785071933321413e-05,
+      "loss": 0.7817,
+      "step": 7413
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2782934248047214e-05,
+      "loss": 0.6113,
+      "step": 7414
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.278079652677401e-05,
+      "loss": 0.6942,
+      "step": 7415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2778658769600092e-05,
+      "loss": 0.682,
+      "step": 7416
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2776520976623759e-05,
+      "loss": 0.6763,
+      "step": 7417
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2774383147943294e-05,
+      "loss": 0.7794,
+      "step": 7418
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2772245283657005e-05,
+      "loss": 0.6896,
+      "step": 7419
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2770107383863184e-05,
+      "loss": 0.7174,
+      "step": 7420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.276796944866013e-05,
+      "loss": 0.6513,
+      "step": 7421
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2765831478146149e-05,
+      "loss": 0.6492,
+      "step": 7422
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2763693472419538e-05,
+      "loss": 0.6588,
+      "step": 7423
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.276155543157861e-05,
+      "loss": 0.6301,
+      "step": 7424
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.275941735572166e-05,
+      "loss": 0.7134,
+      "step": 7425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2757279244947009e-05,
+      "loss": 0.655,
+      "step": 7426
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2755141099352956e-05,
+      "loss": 0.7105,
+      "step": 7427
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.275300291903782e-05,
+      "loss": 0.6357,
+      "step": 7428
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2750864704099905e-05,
+      "loss": 0.6909,
+      "step": 7429
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2748726454637531e-05,
+      "loss": 0.7371,
+      "step": 7430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2746588170749013e-05,
+      "loss": 0.5858,
+      "step": 7431
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2744449852532668e-05,
+      "loss": 0.6654,
+      "step": 7432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2742311500086812e-05,
+      "loss": 0.6088,
+      "step": 7433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2740173113509776e-05,
+      "loss": 0.6535,
+      "step": 7434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2738034692899868e-05,
+      "loss": 0.6805,
+      "step": 7435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2735896238355422e-05,
+      "loss": 0.6321,
+      "step": 7436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2733757749974761e-05,
+      "loss": 0.6414,
+      "step": 7437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2731619227856209e-05,
+      "loss": 0.6636,
+      "step": 7438
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2729480672098093e-05,
+      "loss": 0.6002,
+      "step": 7439
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.272734208279875e-05,
+      "loss": 0.641,
+      "step": 7440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2725203460056507e-05,
+      "loss": 0.6574,
+      "step": 7441
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2723064803969697e-05,
+      "loss": 0.5791,
+      "step": 7442
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2720926114636656e-05,
+      "loss": 0.7415,
+      "step": 7443
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2718787392155718e-05,
+      "loss": 0.7113,
+      "step": 7444
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2716648636625224e-05,
+      "loss": 0.6894,
+      "step": 7445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2714509848143509e-05,
+      "loss": 0.6364,
+      "step": 7446
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2712371026808916e-05,
+      "loss": 0.6192,
+      "step": 7447
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2710232172719786e-05,
+      "loss": 0.6605,
+      "step": 7448
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2708093285974464e-05,
+      "loss": 0.6831,
+      "step": 7449
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2705954366671294e-05,
+      "loss": 0.7346,
+      "step": 7450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2703815414908626e-05,
+      "loss": 0.6827,
+      "step": 7451
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.27016764307848e-05,
+      "loss": 0.6554,
+      "step": 7452
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2699537414398178e-05,
+      "loss": 0.6584,
+      "step": 7453
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2697398365847102e-05,
+      "loss": 0.6787,
+      "step": 7454
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2695259285229924e-05,
+      "loss": 0.695,
+      "step": 7455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2693120172645002e-05,
+      "loss": 0.6702,
+      "step": 7456
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.269098102819069e-05,
+      "loss": 0.611,
+      "step": 7457
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2688841851965347e-05,
+      "loss": 0.6774,
+      "step": 7458
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2686702644067328e-05,
+      "loss": 0.6615,
+      "step": 7459
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2684563404594995e-05,
+      "loss": 0.648,
+      "step": 7460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2682424133646711e-05,
+      "loss": 0.6424,
+      "step": 7461
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2680284831320838e-05,
+      "loss": 0.6605,
+      "step": 7462
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2678145497715734e-05,
+      "loss": 0.661,
+      "step": 7463
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2676006132929773e-05,
+      "loss": 0.6836,
+      "step": 7464
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2673866737061317e-05,
+      "loss": 0.6819,
+      "step": 7465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.267172731020874e-05,
+      "loss": 0.6477,
+      "step": 7466
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2669587852470408e-05,
+      "loss": 0.5791,
+      "step": 7467
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2667448363944694e-05,
+      "loss": 0.6446,
+      "step": 7468
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2665308844729967e-05,
+      "loss": 0.6217,
+      "step": 7469
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2663169294924607e-05,
+      "loss": 0.6897,
+      "step": 7470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2661029714626987e-05,
+      "loss": 0.7092,
+      "step": 7471
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2658890103935482e-05,
+      "loss": 0.6574,
+      "step": 7472
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.265675046294847e-05,
+      "loss": 0.7123,
+      "step": 7473
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2654610791764334e-05,
+      "loss": 0.7323,
+      "step": 7474
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2652471090481459e-05,
+      "loss": 0.6903,
+      "step": 7475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.265033135919822e-05,
+      "loss": 0.5953,
+      "step": 7476
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2648191598013003e-05,
+      "loss": 0.7046,
+      "step": 7477
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2646051807024195e-05,
+      "loss": 0.6916,
+      "step": 7478
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2643911986330183e-05,
+      "loss": 0.7058,
+      "step": 7479
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2641772136029356e-05,
+      "loss": 0.6061,
+      "step": 7480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2639632256220098e-05,
+      "loss": 0.6404,
+      "step": 7481
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2637492347000803e-05,
+      "loss": 0.6471,
+      "step": 7482
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2635352408469868e-05,
+      "loss": 0.6605,
+      "step": 7483
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2633212440725677e-05,
+      "loss": 0.6481,
+      "step": 7484
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2631072443866633e-05,
+      "loss": 0.7085,
+      "step": 7485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2628932417991132e-05,
+      "loss": 0.6213,
+      "step": 7486
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2626792363197564e-05,
+      "loss": 0.5905,
+      "step": 7487
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2624652279584337e-05,
+      "loss": 0.6614,
+      "step": 7488
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2622512167249843e-05,
+      "loss": 0.6755,
+      "step": 7489
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2620372026292486e-05,
+      "loss": 0.608,
+      "step": 7490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2618231856810674e-05,
+      "loss": 0.6234,
+      "step": 7491
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2616091658902806e-05,
+      "loss": 0.6681,
+      "step": 7492
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2613951432667288e-05,
+      "loss": 0.6441,
+      "step": 7493
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.261181117820253e-05,
+      "loss": 0.6686,
+      "step": 7494
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2609670895606936e-05,
+      "loss": 0.6273,
+      "step": 7495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2607530584978917e-05,
+      "loss": 0.6453,
+      "step": 7496
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2605390246416884e-05,
+      "loss": 0.6815,
+      "step": 7497
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2603249880019245e-05,
+      "loss": 0.6394,
+      "step": 7498
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.260110948588442e-05,
+      "loss": 0.664,
+      "step": 7499
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.259896906411082e-05,
+      "loss": 0.7044,
+      "step": 7500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2596828614796862e-05,
+      "loss": 0.6714,
+      "step": 7501
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.259468813804096e-05,
+      "loss": 0.6589,
+      "step": 7502
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2592547633941537e-05,
+      "loss": 0.6918,
+      "step": 7503
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2590407102597004e-05,
+      "loss": 0.6879,
+      "step": 7504
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2588266544105794e-05,
+      "loss": 0.6258,
+      "step": 7505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2586125958566317e-05,
+      "loss": 0.6318,
+      "step": 7506
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2583985346077004e-05,
+      "loss": 0.5914,
+      "step": 7507
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2581844706736274e-05,
+      "loss": 0.6632,
+      "step": 7508
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2579704040642562e-05,
+      "loss": 0.6643,
+      "step": 7509
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2577563347894285e-05,
+      "loss": 0.6827,
+      "step": 7510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2575422628589877e-05,
+      "loss": 0.6663,
+      "step": 7511
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2573281882827763e-05,
+      "loss": 0.7322,
+      "step": 7512
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2571141110706382e-05,
+      "loss": 0.654,
+      "step": 7513
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2569000312324152e-05,
+      "loss": 0.6085,
+      "step": 7514
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.256685948777952e-05,
+      "loss": 0.5945,
+      "step": 7515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2564718637170912e-05,
+      "loss": 0.7115,
+      "step": 7516
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2562577760596768e-05,
+      "loss": 0.6785,
+      "step": 7517
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.256043685815552e-05,
+      "loss": 0.6747,
+      "step": 7518
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.255829592994561e-05,
+      "loss": 0.6756,
+      "step": 7519
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2556154976065475e-05,
+      "loss": 0.6863,
+      "step": 7520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2554013996613556e-05,
+      "loss": 0.6355,
+      "step": 7521
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2551872991688295e-05,
+      "loss": 0.6714,
+      "step": 7522
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2549731961388129e-05,
+      "loss": 0.6798,
+      "step": 7523
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.254759090581151e-05,
+      "loss": 0.609,
+      "step": 7524
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2545449825056877e-05,
+      "loss": 0.7321,
+      "step": 7525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.254330871922268e-05,
+      "loss": 0.6123,
+      "step": 7526
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2541167588407364e-05,
+      "loss": 0.7053,
+      "step": 7527
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2539026432709377e-05,
+      "loss": 0.7295,
+      "step": 7528
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2536885252227171e-05,
+      "loss": 0.6165,
+      "step": 7529
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2534744047059195e-05,
+      "loss": 0.6977,
+      "step": 7530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2532602817303898e-05,
+      "loss": 0.6883,
+      "step": 7531
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2530461563059734e-05,
+      "loss": 0.6948,
+      "step": 7532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2528320284425161e-05,
+      "loss": 0.6144,
+      "step": 7533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2526178981498633e-05,
+      "loss": 0.6571,
+      "step": 7534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2524037654378602e-05,
+      "loss": 0.6609,
+      "step": 7535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2521896303163533e-05,
+      "loss": 0.6567,
+      "step": 7536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2519754927951877e-05,
+      "loss": 0.7033,
+      "step": 7537
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2517613528842096e-05,
+      "loss": 0.711,
+      "step": 7538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2515472105932652e-05,
+      "loss": 0.7177,
+      "step": 7539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2513330659322003e-05,
+      "loss": 0.7217,
+      "step": 7540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2511189189108616e-05,
+      "loss": 0.6233,
+      "step": 7541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2509047695390951e-05,
+      "loss": 0.7206,
+      "step": 7542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2506906178267481e-05,
+      "loss": 0.6327,
+      "step": 7543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2504764637836664e-05,
+      "loss": 0.6611,
+      "step": 7544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2502623074196972e-05,
+      "loss": 0.7256,
+      "step": 7545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2500481487446868e-05,
+      "loss": 0.6405,
+      "step": 7546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2498339877684827e-05,
+      "loss": 0.6322,
+      "step": 7547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2496198245009315e-05,
+      "loss": 0.7094,
+      "step": 7548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2494056589518803e-05,
+      "loss": 0.6654,
+      "step": 7549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2491914911311768e-05,
+      "loss": 0.7207,
+      "step": 7550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2489773210486684e-05,
+      "loss": 0.6427,
+      "step": 7551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2487631487142018e-05,
+      "loss": 0.6641,
+      "step": 7552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2485489741376256e-05,
+      "loss": 0.6162,
+      "step": 7553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2483347973287865e-05,
+      "loss": 0.6951,
+      "step": 7554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2481206182975327e-05,
+      "loss": 0.6908,
+      "step": 7555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.247906437053712e-05,
+      "loss": 0.6664,
+      "step": 7556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2476922536071722e-05,
+      "loss": 0.6767,
+      "step": 7557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.247478067967762e-05,
+      "loss": 0.6512,
+      "step": 7558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2472638801453287e-05,
+      "loss": 0.6097,
+      "step": 7559
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2470496901497216e-05,
+      "loss": 0.6037,
+      "step": 7560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.246835497990788e-05,
+      "loss": 0.6716,
+      "step": 7561
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.246621303678377e-05,
+      "loss": 0.7103,
+      "step": 7562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.246407107222337e-05,
+      "loss": 0.6383,
+      "step": 7563
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2461929086325166e-05,
+      "loss": 0.6799,
+      "step": 7564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2459787079187647e-05,
+      "loss": 0.6452,
+      "step": 7565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2457645050909301e-05,
+      "loss": 0.6529,
+      "step": 7566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2455503001588615e-05,
+      "loss": 0.6807,
+      "step": 7567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2453360931324088e-05,
+      "loss": 0.6865,
+      "step": 7568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2451218840214203e-05,
+      "loss": 0.6947,
+      "step": 7569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2449076728357457e-05,
+      "loss": 0.6478,
+      "step": 7570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.244693459585234e-05,
+      "loss": 0.6359,
+      "step": 7571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2444792442797349e-05,
+      "loss": 0.655,
+      "step": 7572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2442650269290976e-05,
+      "loss": 0.7173,
+      "step": 7573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.244050807543172e-05,
+      "loss": 0.665,
+      "step": 7574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2438365861318082e-05,
+      "loss": 0.6772,
+      "step": 7575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2436223627048553e-05,
+      "loss": 0.6854,
+      "step": 7576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2434081372721637e-05,
+      "loss": 0.6439,
+      "step": 7577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2431939098435832e-05,
+      "loss": 0.7273,
+      "step": 7578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.242979680428964e-05,
+      "loss": 0.6993,
+      "step": 7579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.242765449038156e-05,
+      "loss": 0.6772,
+      "step": 7580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.24255121568101e-05,
+      "loss": 0.5904,
+      "step": 7581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2423369803673756e-05,
+      "loss": 0.6787,
+      "step": 7582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2421227431071038e-05,
+      "loss": 0.6688,
+      "step": 7583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.241908503910045e-05,
+      "loss": 0.6641,
+      "step": 7584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2416942627860502e-05,
+      "loss": 0.6517,
+      "step": 7585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2414800197449694e-05,
+      "loss": 0.6521,
+      "step": 7586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2412657747966543e-05,
+      "loss": 0.6892,
+      "step": 7587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2410515279509551e-05,
+      "loss": 0.6232,
+      "step": 7588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2408372792177228e-05,
+      "loss": 0.7073,
+      "step": 7589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2406230286068085e-05,
+      "loss": 0.6029,
+      "step": 7590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.240408776128064e-05,
+      "loss": 0.6665,
+      "step": 7591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2401945217913399e-05,
+      "loss": 0.6831,
+      "step": 7592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2399802656064875e-05,
+      "loss": 0.6321,
+      "step": 7593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2397660075833586e-05,
+      "loss": 0.7223,
+      "step": 7594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2395517477318045e-05,
+      "loss": 0.6526,
+      "step": 7595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2393374860616767e-05,
+      "loss": 0.6522,
+      "step": 7596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2391232225828269e-05,
+      "loss": 0.6385,
+      "step": 7597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.238908957305107e-05,
+      "loss": 0.6939,
+      "step": 7598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2386946902383685e-05,
+      "loss": 0.7375,
+      "step": 7599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2384804213924637e-05,
+      "loss": 0.7194,
+      "step": 7600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2382661507772444e-05,
+      "loss": 0.6569,
+      "step": 7601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.238051878402563e-05,
+      "loss": 0.6878,
+      "step": 7602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2378376042782712e-05,
+      "loss": 0.6328,
+      "step": 7603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2376233284142214e-05,
+      "loss": 0.6488,
+      "step": 7604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.237409050820266e-05,
+      "loss": 0.6698,
+      "step": 7605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.237194771506257e-05,
+      "loss": 0.6519,
+      "step": 7606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2369804904820475e-05,
+      "loss": 0.6692,
+      "step": 7607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2367662077574897e-05,
+      "loss": 0.7009,
+      "step": 7608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2365519233424366e-05,
+      "loss": 0.6582,
+      "step": 7609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.23633763724674e-05,
+      "loss": 0.6498,
+      "step": 7610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2361233494802541e-05,
+      "loss": 0.5887,
+      "step": 7611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2359090600528302e-05,
+      "loss": 0.671,
+      "step": 7612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2356947689743226e-05,
+      "loss": 0.6782,
+      "step": 7613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2354804762545834e-05,
+      "loss": 0.6669,
+      "step": 7614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.235266181903466e-05,
+      "loss": 0.618,
+      "step": 7615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2350518859308237e-05,
+      "loss": 0.6312,
+      "step": 7616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2348375883465098e-05,
+      "loss": 0.694,
+      "step": 7617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2346232891603771e-05,
+      "loss": 0.6327,
+      "step": 7618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2344089883822796e-05,
+      "loss": 0.5984,
+      "step": 7619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2341946860220703e-05,
+      "loss": 0.6293,
+      "step": 7620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2339803820896033e-05,
+      "loss": 0.6498,
+      "step": 7621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.233766076594732e-05,
+      "loss": 0.7549,
+      "step": 7622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2335517695473094e-05,
+      "loss": 0.675,
+      "step": 7623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2333374609571897e-05,
+      "loss": 0.7147,
+      "step": 7624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2331231508342271e-05,
+      "loss": 0.6338,
+      "step": 7625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2329088391882753e-05,
+      "loss": 0.7101,
+      "step": 7626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.232694526029188e-05,
+      "loss": 0.6315,
+      "step": 7627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2324802113668198e-05,
+      "loss": 0.704,
+      "step": 7628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.232265895211024e-05,
+      "loss": 0.6969,
+      "step": 7629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2320515775716555e-05,
+      "loss": 0.7486,
+      "step": 7630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2318372584585678e-05,
+      "loss": 0.6935,
+      "step": 7631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.231622937881616e-05,
+      "loss": 0.6705,
+      "step": 7632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2314086158506537e-05,
+      "loss": 0.7316,
+      "step": 7633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.231194292375536e-05,
+      "loss": 0.6729,
+      "step": 7634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.230979967466117e-05,
+      "loss": 0.6788,
+      "step": 7635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2307656411322517e-05,
+      "loss": 0.7454,
+      "step": 7636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2305513133837945e-05,
+      "loss": 0.6231,
+      "step": 7637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2303369842305999e-05,
+      "loss": 0.6611,
+      "step": 7638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2301226536825231e-05,
+      "loss": 0.6522,
+      "step": 7639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2299083217494184e-05,
+      "loss": 0.6369,
+      "step": 7640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2296939884411408e-05,
+      "loss": 0.6148,
+      "step": 7641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2294796537675455e-05,
+      "loss": 0.6095,
+      "step": 7642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2292653177384877e-05,
+      "loss": 0.7036,
+      "step": 7643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.229050980363822e-05,
+      "loss": 0.6672,
+      "step": 7644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2288366416534042e-05,
+      "loss": 0.6949,
+      "step": 7645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2286223016170886e-05,
+      "loss": 0.6778,
+      "step": 7646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2284079602647314e-05,
+      "loss": 0.6063,
+      "step": 7647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.228193617606187e-05,
+      "loss": 0.6086,
+      "step": 7648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2279792736513117e-05,
+      "loss": 0.7107,
+      "step": 7649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2277649284099602e-05,
+      "loss": 0.6412,
+      "step": 7650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2275505818919886e-05,
+      "loss": 0.6909,
+      "step": 7651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.227336234107252e-05,
+      "loss": 0.6768,
+      "step": 7652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2271218850656066e-05,
+      "loss": 0.6088,
+      "step": 7653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2269075347769078e-05,
+      "loss": 0.7242,
+      "step": 7654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2266931832510107e-05,
+      "loss": 0.6143,
+      "step": 7655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2264788304977721e-05,
+      "loss": 0.6306,
+      "step": 7656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2262644765270472e-05,
+      "loss": 0.6614,
+      "step": 7657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2260501213486921e-05,
+      "loss": 0.6253,
+      "step": 7658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2258357649725628e-05,
+      "loss": 0.5541,
+      "step": 7659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2256214074085155e-05,
+      "loss": 0.6404,
+      "step": 7660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2254070486664059e-05,
+      "loss": 0.7026,
+      "step": 7661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2251926887560904e-05,
+      "loss": 0.6661,
+      "step": 7662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.224978327687425e-05,
+      "loss": 0.5801,
+      "step": 7663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2247639654702662e-05,
+      "loss": 0.7037,
+      "step": 7664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2245496021144701e-05,
+      "loss": 0.6627,
+      "step": 7665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2243352376298926e-05,
+      "loss": 0.6751,
+      "step": 7666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.224120872026391e-05,
+      "loss": 0.6412,
+      "step": 7667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2239065053138214e-05,
+      "loss": 0.6104,
+      "step": 7668
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.22369213750204e-05,
+      "loss": 0.7041,
+      "step": 7669
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2234777686009039e-05,
+      "loss": 0.5938,
+      "step": 7670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.223263398620269e-05,
+      "loss": 0.6781,
+      "step": 7671
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2230490275699924e-05,
+      "loss": 0.6065,
+      "step": 7672
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2228346554599307e-05,
+      "loss": 0.6572,
+      "step": 7673
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2226202822999404e-05,
+      "loss": 0.6563,
+      "step": 7674
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2224059080998786e-05,
+      "loss": 0.6816,
+      "step": 7675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2221915328696021e-05,
+      "loss": 0.7088,
+      "step": 7676
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.221977156618968e-05,
+      "loss": 0.7103,
+      "step": 7677
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2217627793578328e-05,
+      "loss": 0.6689,
+      "step": 7678
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2215484010960537e-05,
+      "loss": 0.5874,
+      "step": 7679
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2213340218434877e-05,
+      "loss": 0.645,
+      "step": 7680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.221119641609992e-05,
+      "loss": 0.7672,
+      "step": 7681
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2209052604054231e-05,
+      "loss": 0.6949,
+      "step": 7682
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2206908782396388e-05,
+      "loss": 0.7066,
+      "step": 7683
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2204764951224966e-05,
+      "loss": 0.6107,
+      "step": 7684
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.220262111063853e-05,
+      "loss": 0.6534,
+      "step": 7685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2200477260735655e-05,
+      "loss": 0.6338,
+      "step": 7686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2198333401614919e-05,
+      "loss": 0.7079,
+      "step": 7687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2196189533374892e-05,
+      "loss": 0.6259,
+      "step": 7688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2194045656114144e-05,
+      "loss": 0.7214,
+      "step": 7689
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2191901769931259e-05,
+      "loss": 0.7427,
+      "step": 7690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.21897578749248e-05,
+      "loss": 0.6618,
+      "step": 7691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2187613971193355e-05,
+      "loss": 0.712,
+      "step": 7692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2185470058835491e-05,
+      "loss": 0.6503,
+      "step": 7693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2183326137949792e-05,
+      "loss": 0.6269,
+      "step": 7694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2181182208634825e-05,
+      "loss": 0.6626,
+      "step": 7695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2179038270989176e-05,
+      "loss": 0.6631,
+      "step": 7696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2176894325111417e-05,
+      "loss": 0.6887,
+      "step": 7697
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2174750371100127e-05,
+      "loss": 0.6841,
+      "step": 7698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2172606409053887e-05,
+      "loss": 0.6681,
+      "step": 7699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2170462439071269e-05,
+      "loss": 0.6965,
+      "step": 7700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2168318461250859e-05,
+      "loss": 0.6774,
+      "step": 7701
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2166174475691233e-05,
+      "loss": 0.5927,
+      "step": 7702
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.216403048249097e-05,
+      "loss": 0.6956,
+      "step": 7703
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.216188648174865e-05,
+      "loss": 0.6818,
+      "step": 7704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2159742473562856e-05,
+      "loss": 0.6408,
+      "step": 7705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2157598458032165e-05,
+      "loss": 0.636,
+      "step": 7706
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2155454435255161e-05,
+      "loss": 0.5983,
+      "step": 7707
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2153310405330422e-05,
+      "loss": 0.6863,
+      "step": 7708
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2151166368356533e-05,
+      "loss": 0.6681,
+      "step": 7709
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.214902232443207e-05,
+      "loss": 0.6626,
+      "step": 7710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2146878273655627e-05,
+      "loss": 0.6757,
+      "step": 7711
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2144734216125775e-05,
+      "loss": 0.6786,
+      "step": 7712
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2142590151941104e-05,
+      "loss": 0.677,
+      "step": 7713
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2140446081200192e-05,
+      "loss": 0.6745,
+      "step": 7714
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.213830200400163e-05,
+      "loss": 0.7016,
+      "step": 7715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2136157920443992e-05,
+      "loss": 0.6487,
+      "step": 7716
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2134013830625866e-05,
+      "loss": 0.63,
+      "step": 7717
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.213186973464584e-05,
+      "loss": 0.5989,
+      "step": 7718
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2129725632602496e-05,
+      "loss": 0.7011,
+      "step": 7719
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2127581524594418e-05,
+      "loss": 0.6284,
+      "step": 7720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2125437410720195e-05,
+      "loss": 0.6674,
+      "step": 7721
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.212329329107841e-05,
+      "loss": 0.6544,
+      "step": 7722
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2121149165767645e-05,
+      "loss": 0.708,
+      "step": 7723
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2119005034886493e-05,
+      "loss": 0.6947,
+      "step": 7724
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2116860898533533e-05,
+      "loss": 0.7016,
+      "step": 7725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2114716756807357e-05,
+      "loss": 0.6522,
+      "step": 7726
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2112572609806549e-05,
+      "loss": 0.654,
+      "step": 7727
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.21104284576297e-05,
+      "loss": 0.6142,
+      "step": 7728
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2108284300375391e-05,
+      "loss": 0.63,
+      "step": 7729
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2106140138142219e-05,
+      "loss": 0.6758,
+      "step": 7730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2103995971028761e-05,
+      "loss": 0.6838,
+      "step": 7731
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2101851799133613e-05,
+      "loss": 0.6669,
+      "step": 7732
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2099707622555355e-05,
+      "loss": 0.7045,
+      "step": 7733
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2097563441392582e-05,
+      "loss": 0.617,
+      "step": 7734
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.209541925574388e-05,
+      "loss": 0.7227,
+      "step": 7735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.209327506570784e-05,
+      "loss": 0.5945,
+      "step": 7736
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.209113087138305e-05,
+      "loss": 0.6795,
+      "step": 7737
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2088986672868098e-05,
+      "loss": 0.6667,
+      "step": 7738
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.208684247026158e-05,
+      "loss": 0.6347,
+      "step": 7739
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.208469826366207e-05,
+      "loss": 0.6997,
+      "step": 7740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2082554053168174e-05,
+      "loss": 0.6378,
+      "step": 7741
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2080409838878471e-05,
+      "loss": 0.7001,
+      "step": 7742
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2078265620891561e-05,
+      "loss": 0.7022,
+      "step": 7743
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2076121399306024e-05,
+      "loss": 0.6611,
+      "step": 7744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.207397717422046e-05,
+      "loss": 0.6726,
+      "step": 7745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.207183294573345e-05,
+      "loss": 0.6543,
+      "step": 7746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2069688713943593e-05,
+      "loss": 0.6658,
+      "step": 7747
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2067544478949476e-05,
+      "loss": 0.7479,
+      "step": 7748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2065400240849692e-05,
+      "loss": 0.6374,
+      "step": 7749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2063255999742829e-05,
+      "loss": 0.7094,
+      "step": 7750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2061111755727478e-05,
+      "loss": 0.6773,
+      "step": 7751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2058967508902235e-05,
+      "loss": 0.6118,
+      "step": 7752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.205682325936569e-05,
+      "loss": 0.6217,
+      "step": 7753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2054679007216434e-05,
+      "loss": 0.6484,
+      "step": 7754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2052534752553063e-05,
+      "loss": 0.6708,
+      "step": 7755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.205039049547416e-05,
+      "loss": 0.6892,
+      "step": 7756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2048246236078323e-05,
+      "loss": 0.6007,
+      "step": 7757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2046101974464144e-05,
+      "loss": 0.6921,
+      "step": 7758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2043957710730212e-05,
+      "loss": 0.6551,
+      "step": 7759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2041813444975126e-05,
+      "loss": 0.6356,
+      "step": 7760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.203966917729747e-05,
+      "loss": 0.6634,
+      "step": 7761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2037524907795845e-05,
+      "loss": 0.6349,
+      "step": 7762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2035380636568838e-05,
+      "loss": 0.721,
+      "step": 7763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2033236363715043e-05,
+      "loss": 0.7063,
+      "step": 7764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2031092089333054e-05,
+      "loss": 0.7067,
+      "step": 7765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2028947813521466e-05,
+      "loss": 0.6472,
+      "step": 7766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2026803536378863e-05,
+      "loss": 0.6656,
+      "step": 7767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2024659258003847e-05,
+      "loss": 0.6521,
+      "step": 7768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2022514978495007e-05,
+      "loss": 0.6647,
+      "step": 7769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2020370697950941e-05,
+      "loss": 0.6801,
+      "step": 7770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2018226416470233e-05,
+      "loss": 0.6823,
+      "step": 7771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2016082134151488e-05,
+      "loss": 0.6857,
+      "step": 7772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.201393785109329e-05,
+      "loss": 0.6988,
+      "step": 7773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2011793567394234e-05,
+      "loss": 0.6661,
+      "step": 7774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2009649283152913e-05,
+      "loss": 0.7413,
+      "step": 7775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2007504998467923e-05,
+      "loss": 0.7012,
+      "step": 7776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.200536071343786e-05,
+      "loss": 0.6233,
+      "step": 7777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2003216428161309e-05,
+      "loss": 0.6831,
+      "step": 7778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2001072142736869e-05,
+      "loss": 0.7043,
+      "step": 7779
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1998927857263135e-05,
+      "loss": 0.651,
+      "step": 7780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1996783571838694e-05,
+      "loss": 0.7067,
+      "step": 7781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1994639286562144e-05,
+      "loss": 0.6577,
+      "step": 7782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1992495001532076e-05,
+      "loss": 0.6822,
+      "step": 7783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1990350716847088e-05,
+      "loss": 0.6553,
+      "step": 7784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1988206432605769e-05,
+      "loss": 0.6482,
+      "step": 7785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1986062148906713e-05,
+      "loss": 0.6666,
+      "step": 7786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1983917865848512e-05,
+      "loss": 0.5873,
+      "step": 7787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1981773583529768e-05,
+      "loss": 0.5891,
+      "step": 7788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1979629302049061e-05,
+      "loss": 0.7031,
+      "step": 7789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1977485021504992e-05,
+      "loss": 0.6539,
+      "step": 7790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1975340741996155e-05,
+      "loss": 0.6369,
+      "step": 7791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1973196463621139e-05,
+      "loss": 0.6466,
+      "step": 7792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1971052186478537e-05,
+      "loss": 0.7129,
+      "step": 7793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1968907910666945e-05,
+      "loss": 0.5825,
+      "step": 7794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1966763636284959e-05,
+      "loss": 0.6945,
+      "step": 7795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1964619363431164e-05,
+      "loss": 0.7092,
+      "step": 7796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1962475092204157e-05,
+      "loss": 0.6182,
+      "step": 7797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1960330822702529e-05,
+      "loss": 0.6689,
+      "step": 7798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1958186555024878e-05,
+      "loss": 0.652,
+      "step": 7799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.195604228926979e-05,
+      "loss": 0.6777,
+      "step": 7800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1953898025535858e-05,
+      "loss": 0.7236,
+      "step": 7801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1951753763921677e-05,
+      "loss": 0.6556,
+      "step": 7802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1949609504525844e-05,
+      "loss": 0.5996,
+      "step": 7803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1947465247446941e-05,
+      "loss": 0.6369,
+      "step": 7804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1945320992783565e-05,
+      "loss": 0.6654,
+      "step": 7805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1943176740634312e-05,
+      "loss": 0.6138,
+      "step": 7806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1941032491097767e-05,
+      "loss": 0.5861,
+      "step": 7807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1938888244272524e-05,
+      "loss": 0.7566,
+      "step": 7808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1936744000257174e-05,
+      "loss": 0.6402,
+      "step": 7809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1934599759150312e-05,
+      "loss": 0.6497,
+      "step": 7810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1932455521050527e-05,
+      "loss": 0.7529,
+      "step": 7811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1930311286056408e-05,
+      "loss": 0.6788,
+      "step": 7812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.192816705426655e-05,
+      "loss": 0.5978,
+      "step": 7813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1926022825779546e-05,
+      "loss": 0.6273,
+      "step": 7814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1923878600693978e-05,
+      "loss": 0.7089,
+      "step": 7815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1921734379108443e-05,
+      "loss": 0.6443,
+      "step": 7816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1919590161121528e-05,
+      "loss": 0.5918,
+      "step": 7817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.191744594683183e-05,
+      "loss": 0.6732,
+      "step": 7818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.191530173633793e-05,
+      "loss": 0.6764,
+      "step": 7819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1913157529738424e-05,
+      "loss": 0.7058,
+      "step": 7820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1911013327131899e-05,
+      "loss": 0.6966,
+      "step": 7821
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1908869128616951e-05,
+      "loss": 0.6447,
+      "step": 7822
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.190672493429216e-05,
+      "loss": 0.6385,
+      "step": 7823
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1904580744256119e-05,
+      "loss": 0.6802,
+      "step": 7824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1902436558607419e-05,
+      "loss": 0.6449,
+      "step": 7825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1900292377444646e-05,
+      "loss": 0.691,
+      "step": 7826
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1898148200866391e-05,
+      "loss": 0.6748,
+      "step": 7827
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.189600402897124e-05,
+      "loss": 0.6533,
+      "step": 7828
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1893859861857787e-05,
+      "loss": 0.5965,
+      "step": 7829
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.189171569962461e-05,
+      "loss": 0.6546,
+      "step": 7830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1889571542370302e-05,
+      "loss": 0.6964,
+      "step": 7831
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.188742739019345e-05,
+      "loss": 0.6322,
+      "step": 7832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1885283243192644e-05,
+      "loss": 0.6948,
+      "step": 7833
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1883139101466468e-05,
+      "loss": 0.6303,
+      "step": 7834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.188099496511351e-05,
+      "loss": 0.6057,
+      "step": 7835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1878850834232354e-05,
+      "loss": 0.6595,
+      "step": 7836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1876706708921595e-05,
+      "loss": 0.6618,
+      "step": 7837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1874562589279806e-05,
+      "loss": 0.6398,
+      "step": 7838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.187241847540558e-05,
+      "loss": 0.586,
+      "step": 7839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1870274367397508e-05,
+      "loss": 0.6894,
+      "step": 7840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1868130265354163e-05,
+      "loss": 0.6901,
+      "step": 7841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1865986169374134e-05,
+      "loss": 0.6911,
+      "step": 7842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.186384207955601e-05,
+      "loss": 0.6572,
+      "step": 7843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1861697995998376e-05,
+      "loss": 0.6665,
+      "step": 7844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1859553918799808e-05,
+      "loss": 0.6794,
+      "step": 7845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1857409848058896e-05,
+      "loss": 0.7203,
+      "step": 7846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1855265783874224e-05,
+      "loss": 0.6242,
+      "step": 7847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1853121726344377e-05,
+      "loss": 0.6483,
+      "step": 7848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.185097767556793e-05,
+      "loss": 0.6008,
+      "step": 7849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.184883363164347e-05,
+      "loss": 0.6376,
+      "step": 7850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.184668959466958e-05,
+      "loss": 0.6211,
+      "step": 7851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1844545564744843e-05,
+      "loss": 0.616,
+      "step": 7852
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1842401541967838e-05,
+      "loss": 0.6541,
+      "step": 7853
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1840257526437145e-05,
+      "loss": 0.6376,
+      "step": 7854
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1838113518251349e-05,
+      "loss": 0.685,
+      "step": 7855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1835969517509033e-05,
+      "loss": 0.6874,
+      "step": 7856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.183382552430877e-05,
+      "loss": 0.668,
+      "step": 7857
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1831681538749142e-05,
+      "loss": 0.6751,
+      "step": 7858
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1829537560928732e-05,
+      "loss": 0.6485,
+      "step": 7859
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1827393590946116e-05,
+      "loss": 0.6561,
+      "step": 7860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1825249628899873e-05,
+      "loss": 0.7028,
+      "step": 7861
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1823105674888582e-05,
+      "loss": 0.6745,
+      "step": 7862
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1820961729010828e-05,
+      "loss": 0.7007,
+      "step": 7863
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1818817791365177e-05,
+      "loss": 0.6331,
+      "step": 7864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.181667386205021e-05,
+      "loss": 0.6413,
+      "step": 7865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.181452994116451e-05,
+      "loss": 0.634,
+      "step": 7866
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1812386028806648e-05,
+      "loss": 0.6591,
+      "step": 7867
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.18102421250752e-05,
+      "loss": 0.6781,
+      "step": 7868
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1808098230068745e-05,
+      "loss": 0.6822,
+      "step": 7869
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1805954343885855e-05,
+      "loss": 0.6443,
+      "step": 7870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1803810466625114e-05,
+      "loss": 0.6547,
+      "step": 7871
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1801666598385083e-05,
+      "loss": 0.6555,
+      "step": 7872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1799522739264344e-05,
+      "loss": 0.7176,
+      "step": 7873
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1797378889361474e-05,
+      "loss": 0.6858,
+      "step": 7874
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1795235048775038e-05,
+      "loss": 0.6508,
+      "step": 7875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1793091217603612e-05,
+      "loss": 0.6769,
+      "step": 7876
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.179094739594577e-05,
+      "loss": 0.6848,
+      "step": 7877
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1788803583900085e-05,
+      "loss": 0.6714,
+      "step": 7878
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1786659781565127e-05,
+      "loss": 0.6665,
+      "step": 7879
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1784515989039464e-05,
+      "loss": 0.678,
+      "step": 7880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1782372206421673e-05,
+      "loss": 0.6655,
+      "step": 7881
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1780228433810324e-05,
+      "loss": 0.6797,
+      "step": 7882
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.177808467130398e-05,
+      "loss": 0.6872,
+      "step": 7883
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1775940919001215e-05,
+      "loss": 0.6489,
+      "step": 7884
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1773797177000598e-05,
+      "loss": 0.7254,
+      "step": 7885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1771653445400697e-05,
+      "loss": 0.5746,
+      "step": 7886
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1769509724300078e-05,
+      "loss": 0.6461,
+      "step": 7887
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1767366013797312e-05,
+      "loss": 0.6127,
+      "step": 7888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1765222313990967e-05,
+      "loss": 0.6713,
+      "step": 7889
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1763078624979602e-05,
+      "loss": 0.6535,
+      "step": 7890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1760934946861786e-05,
+      "loss": 0.6582,
+      "step": 7891
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1758791279736089e-05,
+      "loss": 0.679,
+      "step": 7892
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1756647623701075e-05,
+      "loss": 0.7033,
+      "step": 7893
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1754503978855303e-05,
+      "loss": 0.6226,
+      "step": 7894
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.175236034529734e-05,
+      "loss": 0.7348,
+      "step": 7895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.175021672312575e-05,
+      "loss": 0.6563,
+      "step": 7896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.17480731124391e-05,
+      "loss": 0.6755,
+      "step": 7897
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1745929513335944e-05,
+      "loss": 0.6595,
+      "step": 7898
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1743785925914847e-05,
+      "loss": 0.6815,
+      "step": 7899
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1741642350274373e-05,
+      "loss": 0.6869,
+      "step": 7900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1739498786513081e-05,
+      "loss": 0.6505,
+      "step": 7901
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.173735523472953e-05,
+      "loss": 0.6524,
+      "step": 7902
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1735211695022281e-05,
+      "loss": 0.6517,
+      "step": 7903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1733068167489893e-05,
+      "loss": 0.6789,
+      "step": 7904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1730924652230926e-05,
+      "loss": 0.6319,
+      "step": 7905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1728781149343936e-05,
+      "loss": 0.6555,
+      "step": 7906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1726637658927479e-05,
+      "loss": 0.6508,
+      "step": 7907
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1724494181080118e-05,
+      "loss": 0.6243,
+      "step": 7908
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.17223507159004e-05,
+      "loss": 0.6797,
+      "step": 7909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1720207263486887e-05,
+      "loss": 0.6945,
+      "step": 7910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.171806382393813e-05,
+      "loss": 0.6044,
+      "step": 7911
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1715920397352692e-05,
+      "loss": 0.6696,
+      "step": 7912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1713776983829116e-05,
+      "loss": 0.685,
+      "step": 7913
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.171163358346596e-05,
+      "loss": 0.6881,
+      "step": 7914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.170949019636178e-05,
+      "loss": 0.6952,
+      "step": 7915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1707346822615127e-05,
+      "loss": 0.6436,
+      "step": 7916
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1705203462324546e-05,
+      "loss": 0.6877,
+      "step": 7917
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1703060115588595e-05,
+      "loss": 0.7086,
+      "step": 7918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.170091678250582e-05,
+      "loss": 0.5679,
+      "step": 7919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1698773463174773e-05,
+      "loss": 0.6821,
+      "step": 7920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1696630157694004e-05,
+      "loss": 0.6369,
+      "step": 7921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1694486866162056e-05,
+      "loss": 0.6489,
+      "step": 7922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1692343588677486e-05,
+      "loss": 0.5739,
+      "step": 7923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1690200325338832e-05,
+      "loss": 0.7012,
+      "step": 7924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1688057076244642e-05,
+      "loss": 0.6121,
+      "step": 7925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1685913841493464e-05,
+      "loss": 0.617,
+      "step": 7926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1683770621183844e-05,
+      "loss": 0.6218,
+      "step": 7927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1681627415414323e-05,
+      "loss": 0.6771,
+      "step": 7928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1679484224283449e-05,
+      "loss": 0.726,
+      "step": 7929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.167734104788976e-05,
+      "loss": 0.7097,
+      "step": 7930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1675197886331806e-05,
+      "loss": 0.632,
+      "step": 7931
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.167305473970812e-05,
+      "loss": 0.6644,
+      "step": 7932
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1670911608117247e-05,
+      "loss": 0.5919,
+      "step": 7933
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1668768491657728e-05,
+      "loss": 0.6887,
+      "step": 7934
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1666625390428103e-05,
+      "loss": 0.6861,
+      "step": 7935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1664482304526909e-05,
+      "loss": 0.7159,
+      "step": 7936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1662339234052685e-05,
+      "loss": 0.6879,
+      "step": 7937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1660196179103966e-05,
+      "loss": 0.636,
+      "step": 7938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1658053139779299e-05,
+      "loss": 0.7252,
+      "step": 7939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1655910116177206e-05,
+      "loss": 0.6433,
+      "step": 7940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1653767108396229e-05,
+      "loss": 0.6464,
+      "step": 7941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1651624116534906e-05,
+      "loss": 0.6557,
+      "step": 7942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1649481140691763e-05,
+      "loss": 0.6967,
+      "step": 7943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.164733818096534e-05,
+      "loss": 0.6659,
+      "step": 7944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1645195237454167e-05,
+      "loss": 0.685,
+      "step": 7945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1643052310256778e-05,
+      "loss": 0.6896,
+      "step": 7946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1640909399471699e-05,
+      "loss": 0.6449,
+      "step": 7947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1638766505197463e-05,
+      "loss": 0.6911,
+      "step": 7948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1636623627532598e-05,
+      "loss": 0.6756,
+      "step": 7949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.163448076657564e-05,
+      "loss": 0.6305,
+      "step": 7950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1632337922425105e-05,
+      "loss": 0.6854,
+      "step": 7951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1630195095179526e-05,
+      "loss": 0.6406,
+      "step": 7952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.162805228493743e-05,
+      "loss": 0.6506,
+      "step": 7953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1625909491797344e-05,
+      "loss": 0.6293,
+      "step": 7954
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1623766715857789e-05,
+      "loss": 0.6929,
+      "step": 7955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1621623957217289e-05,
+      "loss": 0.6382,
+      "step": 7956
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1619481215974375e-05,
+      "loss": 0.6644,
+      "step": 7957
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1617338492227558e-05,
+      "loss": 0.5878,
+      "step": 7958
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1615195786075366e-05,
+      "loss": 0.6655,
+      "step": 7959
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1613053097616316e-05,
+      "loss": 0.7114,
+      "step": 7960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1610910426948934e-05,
+      "loss": 0.6458,
+      "step": 7961
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1608767774171733e-05,
+      "loss": 0.689,
+      "step": 7962
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1606625139383236e-05,
+      "loss": 0.6747,
+      "step": 7963
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1604482522681956e-05,
+      "loss": 0.6494,
+      "step": 7964
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1602339924166418e-05,
+      "loss": 0.6513,
+      "step": 7965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1600197343935127e-05,
+      "loss": 0.6459,
+      "step": 7966
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1598054782086604e-05,
+      "loss": 0.7004,
+      "step": 7967
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1595912238719361e-05,
+      "loss": 0.6337,
+      "step": 7968
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1593769713931915e-05,
+      "loss": 0.6264,
+      "step": 7969
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1591627207822774e-05,
+      "loss": 0.6406,
+      "step": 7970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1589484720490451e-05,
+      "loss": 0.6409,
+      "step": 7971
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1587342252033457e-05,
+      "loss": 0.7169,
+      "step": 7972
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1585199802550307e-05,
+      "loss": 0.6645,
+      "step": 7973
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.15830573721395e-05,
+      "loss": 0.6732,
+      "step": 7974
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.158091496089955e-05,
+      "loss": 0.669,
+      "step": 7975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1578772568928964e-05,
+      "loss": 0.6483,
+      "step": 7976
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1576630196326246e-05,
+      "loss": 0.6417,
+      "step": 7977
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1574487843189903e-05,
+      "loss": 0.7042,
+      "step": 7978
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1572345509618439e-05,
+      "loss": 0.6427,
+      "step": 7979
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1570203195710364e-05,
+      "loss": 0.7245,
+      "step": 7980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.156806090156417e-05,
+      "loss": 0.7339,
+      "step": 7981
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1565918627278363e-05,
+      "loss": 0.6905,
+      "step": 7982
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1563776372951446e-05,
+      "loss": 0.6525,
+      "step": 7983
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1561634138681922e-05,
+      "loss": 0.7527,
+      "step": 7984
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.155949192456828e-05,
+      "loss": 0.6001,
+      "step": 7985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1557349730709026e-05,
+      "loss": 0.6446,
+      "step": 7986
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1555207557202654e-05,
+      "loss": 0.6561,
+      "step": 7987
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1553065404147665e-05,
+      "loss": 0.656,
+      "step": 7988
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1550923271642546e-05,
+      "loss": 0.6949,
+      "step": 7989
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1548781159785798e-05,
+      "loss": 0.6089,
+      "step": 7990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1546639068675916e-05,
+      "loss": 0.6759,
+      "step": 7991
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1544496998411386e-05,
+      "loss": 0.6566,
+      "step": 7992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1542354949090701e-05,
+      "loss": 0.6647,
+      "step": 7993
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1540212920812354e-05,
+      "loss": 0.6332,
+      "step": 7994
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1538070913674837e-05,
+      "loss": 0.5934,
+      "step": 7995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1535928927776633e-05,
+      "loss": 0.5997,
+      "step": 7996
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1533786963216233e-05,
+      "loss": 0.7218,
+      "step": 7997
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1531645020092121e-05,
+      "loss": 0.6395,
+      "step": 7998
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.152950309850279e-05,
+      "loss": 0.6407,
+      "step": 7999
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1527361198546714e-05,
+      "loss": 0.6743,
+      "step": 8000
+    },
+    {
+      "epoch": 0.53,
+      "eval_text_loss": 0.6594743728637695,
+      "eval_text_runtime": 1.8388,
+      "eval_text_samples_per_second": 40.243,
+      "eval_text_steps_per_second": 1.088,
+      "step": 8000
+    },
+    {
+      "epoch": 0.53,
+      "eval_svit_loss": 0.8068998456001282,
+      "eval_svit_runtime": 6.5509,
+      "eval_svit_samples_per_second": 124.563,
+      "eval_svit_steps_per_second": 1.984,
+      "step": 8000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1525219320322383e-05,
+      "loss": 0.612,
+      "step": 8001
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1523077463928277e-05,
+      "loss": 0.6497,
+      "step": 8002
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1520935629462883e-05,
+      "loss": 0.62,
+      "step": 8003
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1518793817024675e-05,
+      "loss": 0.6292,
+      "step": 8004
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1516652026712137e-05,
+      "loss": 0.5849,
+      "step": 8005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1514510258623745e-05,
+      "loss": 0.6067,
+      "step": 8006
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1512368512857983e-05,
+      "loss": 0.5775,
+      "step": 8007
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1510226789513318e-05,
+      "loss": 0.6393,
+      "step": 8008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1508085088688232e-05,
+      "loss": 0.6788,
+      "step": 8009
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1505943410481198e-05,
+      "loss": 0.5959,
+      "step": 8010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1503801754990687e-05,
+      "loss": 0.6361,
+      "step": 8011
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1501660122315175e-05,
+      "loss": 0.6664,
+      "step": 8012
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1499518512553133e-05,
+      "loss": 0.6415,
+      "step": 8013
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1497376925803032e-05,
+      "loss": 0.6775,
+      "step": 8014
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1495235362163338e-05,
+      "loss": 0.6019,
+      "step": 8015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.149309382173252e-05,
+      "loss": 0.5653,
+      "step": 8016
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1490952304609048e-05,
+      "loss": 0.6549,
+      "step": 8017
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1488810810891388e-05,
+      "loss": 0.7088,
+      "step": 8018
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1486669340677998e-05,
+      "loss": 0.6689,
+      "step": 8019
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.148452789406735e-05,
+      "loss": 0.6103,
+      "step": 8020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1482386471157905e-05,
+      "loss": 0.6484,
+      "step": 8021
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1480245072048127e-05,
+      "loss": 0.5807,
+      "step": 8022
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.147810369683647e-05,
+      "loss": 0.5892,
+      "step": 8023
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1475962345621397e-05,
+      "loss": 0.6812,
+      "step": 8024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.147382101850137e-05,
+      "loss": 0.6383,
+      "step": 8025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.147167971557484e-05,
+      "loss": 0.6608,
+      "step": 8026
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1469538436940267e-05,
+      "loss": 0.7135,
+      "step": 8027
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1467397182696104e-05,
+      "loss": 0.6032,
+      "step": 8028
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.146525595294081e-05,
+      "loss": 0.6263,
+      "step": 8029
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1463114747772833e-05,
+      "loss": 0.5827,
+      "step": 8030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1460973567290623e-05,
+      "loss": 0.6929,
+      "step": 8031
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1458832411592635e-05,
+      "loss": 0.6157,
+      "step": 8032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1456691280777323e-05,
+      "loss": 0.6872,
+      "step": 8033
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1454550174943126e-05,
+      "loss": 0.7842,
+      "step": 8034
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1452409094188492e-05,
+      "loss": 0.6887,
+      "step": 8035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1450268038611872e-05,
+      "loss": 0.6688,
+      "step": 8036
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.144812700831171e-05,
+      "loss": 0.7016,
+      "step": 8037
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1445986003386446e-05,
+      "loss": 0.7338,
+      "step": 8038
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1443845023934525e-05,
+      "loss": 0.6865,
+      "step": 8039
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1441704070054394e-05,
+      "loss": 0.7197,
+      "step": 8040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1439563141844483e-05,
+      "loss": 0.656,
+      "step": 8041
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1437422239403236e-05,
+      "loss": 0.7134,
+      "step": 8042
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1435281362829089e-05,
+      "loss": 0.6458,
+      "step": 8043
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1433140512220483e-05,
+      "loss": 0.6419,
+      "step": 8044
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1430999687675848e-05,
+      "loss": 0.5674,
+      "step": 8045
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1428858889293622e-05,
+      "loss": 0.7266,
+      "step": 8046
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1426718117172235e-05,
+      "loss": 0.5869,
+      "step": 8047
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1424577371410127e-05,
+      "loss": 0.6596,
+      "step": 8048
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1422436652105717e-05,
+      "loss": 0.6276,
+      "step": 8049
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.142029595935744e-05,
+      "loss": 0.7657,
+      "step": 8050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1418155293263725e-05,
+      "loss": 0.6635,
+      "step": 8051
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1416014653922998e-05,
+      "loss": 0.7079,
+      "step": 8052
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1413874041433684e-05,
+      "loss": 0.6191,
+      "step": 8053
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.141173345589421e-05,
+      "loss": 0.6257,
+      "step": 8054
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1409592897402995e-05,
+      "loss": 0.6262,
+      "step": 8055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1407452366058469e-05,
+      "loss": 0.649,
+      "step": 8056
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1405311861959042e-05,
+      "loss": 0.6626,
+      "step": 8057
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1403171385203139e-05,
+      "loss": 0.6952,
+      "step": 8058
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1401030935889184e-05,
+      "loss": 0.7082,
+      "step": 8059
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1398890514115583e-05,
+      "loss": 0.6033,
+      "step": 8060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1396750119980756e-05,
+      "loss": 0.6894,
+      "step": 8061
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1394609753583118e-05,
+      "loss": 0.6316,
+      "step": 8062
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1392469415021087e-05,
+      "loss": 0.6137,
+      "step": 8063
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1390329104393066e-05,
+      "loss": 0.5495,
+      "step": 8064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1388188821797472e-05,
+      "loss": 0.6692,
+      "step": 8065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.138604856733271e-05,
+      "loss": 0.5917,
+      "step": 8066
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1383908341097198e-05,
+      "loss": 0.6107,
+      "step": 8067
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1381768143189328e-05,
+      "loss": 0.6196,
+      "step": 8068
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1379627973707515e-05,
+      "loss": 0.6207,
+      "step": 8069
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.137748783275016e-05,
+      "loss": 0.6662,
+      "step": 8070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1375347720415668e-05,
+      "loss": 0.5936,
+      "step": 8071
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1373207636802437e-05,
+      "loss": 0.7222,
+      "step": 8072
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1371067582008869e-05,
+      "loss": 0.6476,
+      "step": 8073
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.136892755613337e-05,
+      "loss": 0.7142,
+      "step": 8074
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1366787559274324e-05,
+      "loss": 0.6807,
+      "step": 8075
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1364647591530136e-05,
+      "loss": 0.6143,
+      "step": 8076
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1362507652999198e-05,
+      "loss": 0.6022,
+      "step": 8077
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1360367743779906e-05,
+      "loss": 0.6711,
+      "step": 8078
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1358227863970648e-05,
+      "loss": 0.6904,
+      "step": 8079
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1356088013669818e-05,
+      "loss": 0.632,
+      "step": 8080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1353948192975802e-05,
+      "loss": 0.6347,
+      "step": 8081
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1351808401987e-05,
+      "loss": 0.6785,
+      "step": 8082
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1349668640801783e-05,
+      "loss": 0.7564,
+      "step": 8083
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1347528909518543e-05,
+      "loss": 0.6673,
+      "step": 8084
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1345389208235664e-05,
+      "loss": 0.656,
+      "step": 8085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.134324953705153e-05,
+      "loss": 0.6331,
+      "step": 8086
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1341109896064522e-05,
+      "loss": 0.6832,
+      "step": 8087
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1338970285373015e-05,
+      "loss": 0.6866,
+      "step": 8088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1336830705075394e-05,
+      "loss": 0.7096,
+      "step": 8089
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1334691155270035e-05,
+      "loss": 0.7461,
+      "step": 8090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1332551636055309e-05,
+      "loss": 0.7142,
+      "step": 8091
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1330412147529592e-05,
+      "loss": 0.5882,
+      "step": 8092
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1328272689791263e-05,
+      "loss": 0.6891,
+      "step": 8093
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1326133262938685e-05,
+      "loss": 0.6535,
+      "step": 8094
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1323993867070228e-05,
+      "loss": 0.6339,
+      "step": 8095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1321854502284268e-05,
+      "loss": 0.6099,
+      "step": 8096
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1319715168679166e-05,
+      "loss": 0.62,
+      "step": 8097
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1317575866353293e-05,
+      "loss": 0.6744,
+      "step": 8098
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1315436595405005e-05,
+      "loss": 0.6142,
+      "step": 8099
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1313297355932671e-05,
+      "loss": 0.6187,
+      "step": 8100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1311158148034657e-05,
+      "loss": 0.6085,
+      "step": 8101
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1309018971809312e-05,
+      "loss": 0.6853,
+      "step": 8102
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1306879827355e-05,
+      "loss": 0.5759,
+      "step": 8103
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1304740714770077e-05,
+      "loss": 0.6168,
+      "step": 8104
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1302601634152902e-05,
+      "loss": 0.6136,
+      "step": 8105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1300462585601823e-05,
+      "loss": 0.6492,
+      "step": 8106
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1298323569215196e-05,
+      "loss": 0.6375,
+      "step": 8107
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1296184585091378e-05,
+      "loss": 0.6492,
+      "step": 8108
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1294045633328707e-05,
+      "loss": 0.6796,
+      "step": 8109
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1291906714025538e-05,
+      "loss": 0.6989,
+      "step": 8110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1289767827280215e-05,
+      "loss": 0.6634,
+      "step": 8111
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1287628973191087e-05,
+      "loss": 0.6903,
+      "step": 8112
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1285490151856492e-05,
+      "loss": 0.577,
+      "step": 8113
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1283351363374778e-05,
+      "loss": 0.7149,
+      "step": 8114
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1281212607844281e-05,
+      "loss": 0.6641,
+      "step": 8115
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1279073885363348e-05,
+      "loss": 0.659,
+      "step": 8116
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1276935196030305e-05,
+      "loss": 0.6775,
+      "step": 8117
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1274796539943495e-05,
+      "loss": 0.6565,
+      "step": 8118
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.127265791720125e-05,
+      "loss": 0.6131,
+      "step": 8119
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1270519327901907e-05,
+      "loss": 0.6291,
+      "step": 8120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1268380772143794e-05,
+      "loss": 0.6636,
+      "step": 8121
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1266242250025241e-05,
+      "loss": 0.6461,
+      "step": 8122
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1264103761644577e-05,
+      "loss": 0.6231,
+      "step": 8123
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1261965307100134e-05,
+      "loss": 0.6537,
+      "step": 8124
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1259826886490228e-05,
+      "loss": 0.6453,
+      "step": 8125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1257688499913187e-05,
+      "loss": 0.6104,
+      "step": 8126
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1255550147467336e-05,
+      "loss": 0.6518,
+      "step": 8127
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1253411829250988e-05,
+      "loss": 0.727,
+      "step": 8128
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.125127354536247e-05,
+      "loss": 0.6207,
+      "step": 8129
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1249135295900096e-05,
+      "loss": 0.6662,
+      "step": 8130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1246997080962185e-05,
+      "loss": 0.6568,
+      "step": 8131
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1244858900647046e-05,
+      "loss": 0.6127,
+      "step": 8132
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1242720755052992e-05,
+      "loss": 0.6991,
+      "step": 8133
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1240582644278336e-05,
+      "loss": 0.613,
+      "step": 8134
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1238444568421394e-05,
+      "loss": 0.6721,
+      "step": 8135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1236306527580464e-05,
+      "loss": 0.6138,
+      "step": 8136
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1234168521853854e-05,
+      "loss": 0.6409,
+      "step": 8137
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.123203055133987e-05,
+      "loss": 0.658,
+      "step": 8138
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.122989261613682e-05,
+      "loss": 0.6464,
+      "step": 8139
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1227754716342996e-05,
+      "loss": 0.7161,
+      "step": 8140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1225616852056705e-05,
+      "loss": 0.6897,
+      "step": 8141
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1223479023376247e-05,
+      "loss": 0.6024,
+      "step": 8142
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1221341230399908e-05,
+      "loss": 0.72,
+      "step": 8143
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1219203473225992e-05,
+      "loss": 0.7116,
+      "step": 8144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1217065751952788e-05,
+      "loss": 0.6343,
+      "step": 8145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1214928066678591e-05,
+      "loss": 0.6329,
+      "step": 8146
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1212790417501687e-05,
+      "loss": 0.6352,
+      "step": 8147
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1210652804520368e-05,
+      "loss": 0.6336,
+      "step": 8148
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1208515227832917e-05,
+      "loss": 0.6855,
+      "step": 8149
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1206377687537625e-05,
+      "loss": 0.6357,
+      "step": 8150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1204240183732767e-05,
+      "loss": 0.6904,
+      "step": 8151
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.120210271651663e-05,
+      "loss": 0.6057,
+      "step": 8152
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.119996528598749e-05,
+      "loss": 0.5996,
+      "step": 8153
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1197827892243631e-05,
+      "loss": 0.7014,
+      "step": 8154
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1195690535383325e-05,
+      "loss": 0.6253,
+      "step": 8155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1193553215504847e-05,
+      "loss": 0.6947,
+      "step": 8156
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1191415932706471e-05,
+      "loss": 0.6157,
+      "step": 8157
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1189278687086474e-05,
+      "loss": 0.733,
+      "step": 8158
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1187141478743114e-05,
+      "loss": 0.6268,
+      "step": 8159
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1185004307774666e-05,
+      "loss": 0.6459,
+      "step": 8160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1182867174279398e-05,
+      "loss": 0.7026,
+      "step": 8161
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1180730078355569e-05,
+      "loss": 0.6226,
+      "step": 8162
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1178593020101446e-05,
+      "loss": 0.6947,
+      "step": 8163
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1176455999615288e-05,
+      "loss": 0.6956,
+      "step": 8164
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1174319016995361e-05,
+      "loss": 0.6634,
+      "step": 8165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1172182072339912e-05,
+      "loss": 0.6926,
+      "step": 8166
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.11700451657472e-05,
+      "loss": 0.6051,
+      "step": 8167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1167908297315484e-05,
+      "loss": 0.6161,
+      "step": 8168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1165771467143015e-05,
+      "loss": 0.6604,
+      "step": 8169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.116363467532804e-05,
+      "loss": 0.65,
+      "step": 8170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1161497921968809e-05,
+      "loss": 0.6755,
+      "step": 8171
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1159361207163568e-05,
+      "loss": 0.6401,
+      "step": 8172
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1157224531010567e-05,
+      "loss": 0.6513,
+      "step": 8173
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1155087893608045e-05,
+      "loss": 0.6317,
+      "step": 8174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1152951295054245e-05,
+      "loss": 0.6866,
+      "step": 8175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1150814735447412e-05,
+      "loss": 0.648,
+      "step": 8176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1148678214885774e-05,
+      "loss": 0.6118,
+      "step": 8177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1146541733467574e-05,
+      "loss": 0.7376,
+      "step": 8178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1144405291291044e-05,
+      "loss": 0.6559,
+      "step": 8179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1142268888454421e-05,
+      "loss": 0.6033,
+      "step": 8180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1140132525055932e-05,
+      "loss": 0.7232,
+      "step": 8181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1137996201193806e-05,
+      "loss": 0.6855,
+      "step": 8182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1135859916966272e-05,
+      "loss": 0.6087,
+      "step": 8183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1133723672471559e-05,
+      "loss": 0.6372,
+      "step": 8184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1131587467807882e-05,
+      "loss": 0.6303,
+      "step": 8185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1129451303073469e-05,
+      "loss": 0.6109,
+      "step": 8186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1127315178366538e-05,
+      "loss": 0.6165,
+      "step": 8187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1125179093785308e-05,
+      "loss": 0.6478,
+      "step": 8188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1123043049427995e-05,
+      "loss": 0.6777,
+      "step": 8189
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1120907045392812e-05,
+      "loss": 0.6469,
+      "step": 8190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1118771081777978e-05,
+      "loss": 0.6392,
+      "step": 8191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1116635158681696e-05,
+      "loss": 0.6377,
+      "step": 8192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1114499276202176e-05,
+      "loss": 0.6788,
+      "step": 8193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1112363434437628e-05,
+      "loss": 0.5572,
+      "step": 8194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1110227633486258e-05,
+      "loss": 0.6349,
+      "step": 8195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1108091873446264e-05,
+      "loss": 0.6141,
+      "step": 8196
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1105956154415851e-05,
+      "loss": 0.6551,
+      "step": 8197
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.110382047649322e-05,
+      "loss": 0.7115,
+      "step": 8198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1101684839776567e-05,
+      "loss": 0.6497,
+      "step": 8199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1099549244364086e-05,
+      "loss": 0.735,
+      "step": 8200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1097413690353972e-05,
+      "loss": 0.6674,
+      "step": 8201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1095278177844417e-05,
+      "loss": 0.6253,
+      "step": 8202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1093142706933615e-05,
+      "loss": 0.5817,
+      "step": 8203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1091007277719747e-05,
+      "loss": 0.6265,
+      "step": 8204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1088871890301e-05,
+      "loss": 0.6783,
+      "step": 8205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1086736544775564e-05,
+      "loss": 0.6543,
+      "step": 8206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.108460124124162e-05,
+      "loss": 0.5937,
+      "step": 8207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1082465979797344e-05,
+      "loss": 0.6256,
+      "step": 8208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1080330760540917e-05,
+      "loss": 0.6292,
+      "step": 8209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1078195583570518e-05,
+      "loss": 0.6429,
+      "step": 8210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1076060448984318e-05,
+      "loss": 0.6304,
+      "step": 8211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1073925356880487e-05,
+      "loss": 0.6822,
+      "step": 8212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1071790307357203e-05,
+      "loss": 0.6071,
+      "step": 8213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1069655300512631e-05,
+      "loss": 0.6643,
+      "step": 8214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1067520336444938e-05,
+      "loss": 0.7069,
+      "step": 8215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1065385415252286e-05,
+      "loss": 0.6748,
+      "step": 8216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1063250537032842e-05,
+      "loss": 0.63,
+      "step": 8217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.106111570188477e-05,
+      "loss": 0.7044,
+      "step": 8218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.105898090990622e-05,
+      "loss": 0.6485,
+      "step": 8219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1056846161195352e-05,
+      "loss": 0.6756,
+      "step": 8220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1054711455850323e-05,
+      "loss": 0.6611,
+      "step": 8221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1052576793969287e-05,
+      "loss": 0.618,
+      "step": 8222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1050442175650391e-05,
+      "loss": 0.6524,
+      "step": 8223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1048307600991785e-05,
+      "loss": 0.6748,
+      "step": 8224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1046173070091621e-05,
+      "loss": 0.6841,
+      "step": 8225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1044038583048036e-05,
+      "loss": 0.7009,
+      "step": 8226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1041904139959177e-05,
+      "loss": 0.6609,
+      "step": 8227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1039769740923183e-05,
+      "loss": 0.627,
+      "step": 8228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1037635386038196e-05,
+      "loss": 0.6615,
+      "step": 8229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1035501075402349e-05,
+      "loss": 0.6038,
+      "step": 8230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.103336680911378e-05,
+      "loss": 0.6957,
+      "step": 8231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1031232587270619e-05,
+      "loss": 0.6577,
+      "step": 8232
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1029098409971001e-05,
+      "loss": 0.6539,
+      "step": 8233
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.102696427731305e-05,
+      "loss": 0.6979,
+      "step": 8234
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1024830189394892e-05,
+      "loss": 0.6727,
+      "step": 8235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1022696146314655e-05,
+      "loss": 0.5875,
+      "step": 8236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1020562148170462e-05,
+      "loss": 0.6462,
+      "step": 8237
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.101842819506043e-05,
+      "loss": 0.6901,
+      "step": 8238
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1016294287082678e-05,
+      "loss": 0.6976,
+      "step": 8239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1014160424335323e-05,
+      "loss": 0.6822,
+      "step": 8240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1012026606916483e-05,
+      "loss": 0.7283,
+      "step": 8241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1009892834924263e-05,
+      "loss": 0.6323,
+      "step": 8242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1007759108456776e-05,
+      "loss": 0.708,
+      "step": 8243
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1005625427612136e-05,
+      "loss": 0.665,
+      "step": 8244
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1003491792488438e-05,
+      "loss": 0.6394,
+      "step": 8245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.100135820318379e-05,
+      "loss": 0.6292,
+      "step": 8246
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0999224659796294e-05,
+      "loss": 0.6197,
+      "step": 8247
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0997091162424052e-05,
+      "loss": 0.6834,
+      "step": 8248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0994957711165159e-05,
+      "loss": 0.6422,
+      "step": 8249
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0992824306117707e-05,
+      "loss": 0.6843,
+      "step": 8250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0990690947379794e-05,
+      "loss": 0.6611,
+      "step": 8251
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0988557635049514e-05,
+      "loss": 0.6731,
+      "step": 8252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0986424369224948e-05,
+      "loss": 0.6721,
+      "step": 8253
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0984291150004185e-05,
+      "loss": 0.7327,
+      "step": 8254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.098215797748531e-05,
+      "loss": 0.6023,
+      "step": 8255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0980024851766408e-05,
+      "loss": 0.741,
+      "step": 8256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0977891772945556e-05,
+      "loss": 0.6206,
+      "step": 8257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0975758741120834e-05,
+      "loss": 0.7057,
+      "step": 8258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0973625756390321e-05,
+      "loss": 0.6272,
+      "step": 8259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0971492818852083e-05,
+      "loss": 0.6621,
+      "step": 8260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0969359928604195e-05,
+      "loss": 0.6539,
+      "step": 8261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0967227085744728e-05,
+      "loss": 0.6631,
+      "step": 8262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.096509429037175e-05,
+      "loss": 0.7253,
+      "step": 8263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0962961542583324e-05,
+      "loss": 0.6008,
+      "step": 8264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0960828842477513e-05,
+      "loss": 0.6528,
+      "step": 8265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0958696190152376e-05,
+      "loss": 0.6541,
+      "step": 8266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.095656358570598e-05,
+      "loss": 0.5671,
+      "step": 8267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.095443102923637e-05,
+      "loss": 0.5928,
+      "step": 8268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0952298520841606e-05,
+      "loss": 0.6962,
+      "step": 8269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0950166060619737e-05,
+      "loss": 0.6265,
+      "step": 8270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0948033648668818e-05,
+      "loss": 0.6872,
+      "step": 8271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0945901285086891e-05,
+      "loss": 0.665,
+      "step": 8272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0943768969972003e-05,
+      "loss": 0.6181,
+      "step": 8273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0941636703422196e-05,
+      "loss": 0.6361,
+      "step": 8274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.093950448553552e-05,
+      "loss": 0.6855,
+      "step": 8275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0937372316409997e-05,
+      "loss": 0.6404,
+      "step": 8276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0935240196143675e-05,
+      "loss": 0.6908,
+      "step": 8277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0933108124834588e-05,
+      "loss": 0.641,
+      "step": 8278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.093097610258076e-05,
+      "loss": 0.6763,
+      "step": 8279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0928844129480228e-05,
+      "loss": 0.6236,
+      "step": 8280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0926712205631013e-05,
+      "loss": 0.7152,
+      "step": 8281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0924580331131146e-05,
+      "loss": 0.6448,
+      "step": 8282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0922448506078647e-05,
+      "loss": 0.7055,
+      "step": 8283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0920316730571536e-05,
+      "loss": 0.6131,
+      "step": 8284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0918185004707833e-05,
+      "loss": 0.6285,
+      "step": 8285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0916053328585554e-05,
+      "loss": 0.6895,
+      "step": 8286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0913921702302711e-05,
+      "loss": 0.6069,
+      "step": 8287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0911790125957313e-05,
+      "loss": 0.6175,
+      "step": 8288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0909658599647373e-05,
+      "loss": 0.6987,
+      "step": 8289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0907527123470899e-05,
+      "loss": 0.6589,
+      "step": 8290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0905395697525891e-05,
+      "loss": 0.6315,
+      "step": 8291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0903264321910351e-05,
+      "loss": 0.689,
+      "step": 8292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0901132996722287e-05,
+      "loss": 0.5244,
+      "step": 8293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0899001722059686e-05,
+      "loss": 0.6101,
+      "step": 8294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0896870498020547e-05,
+      "loss": 0.6872,
+      "step": 8295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0894739324702865e-05,
+      "loss": 0.6625,
+      "step": 8296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0892608202204628e-05,
+      "loss": 0.6706,
+      "step": 8297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0890477130623825e-05,
+      "loss": 0.6954,
+      "step": 8298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0888346110058441e-05,
+      "loss": 0.6236,
+      "step": 8299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0886215140606462e-05,
+      "loss": 0.6848,
+      "step": 8300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.088408422236587e-05,
+      "loss": 0.6286,
+      "step": 8301
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0881953355434638e-05,
+      "loss": 0.7868,
+      "step": 8302
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0879822539910744e-05,
+      "loss": 0.6962,
+      "step": 8303
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0877691775892166e-05,
+      "loss": 0.5881,
+      "step": 8304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0875561063476874e-05,
+      "loss": 0.6647,
+      "step": 8305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0873430402762835e-05,
+      "loss": 0.6396,
+      "step": 8306
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0871299793848017e-05,
+      "loss": 0.6579,
+      "step": 8307
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0869169236830386e-05,
+      "loss": 0.6911,
+      "step": 8308
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0867038731807908e-05,
+      "loss": 0.6493,
+      "step": 8309
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0864908278878532e-05,
+      "loss": 0.6393,
+      "step": 8310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0862777878140225e-05,
+      "loss": 0.6177,
+      "step": 8311
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0860647529690938e-05,
+      "loss": 0.7152,
+      "step": 8312
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0858517233628622e-05,
+      "loss": 0.6267,
+      "step": 8313
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0856386990051228e-05,
+      "loss": 0.6682,
+      "step": 8314
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0854256799056706e-05,
+      "loss": 0.6485,
+      "step": 8315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0852126660743006e-05,
+      "loss": 0.6477,
+      "step": 8316
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.084999657520806e-05,
+      "loss": 0.7365,
+      "step": 8317
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0847866542549814e-05,
+      "loss": 0.6996,
+      "step": 8318
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0845736562866204e-05,
+      "loss": 0.6793,
+      "step": 8319
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0843606636255174e-05,
+      "loss": 0.6037,
+      "step": 8320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0841476762814645e-05,
+      "loss": 0.6209,
+      "step": 8321
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0839346942642554e-05,
+      "loss": 0.6575,
+      "step": 8322
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0837217175836828e-05,
+      "loss": 0.6914,
+      "step": 8323
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0835087462495395e-05,
+      "loss": 0.6912,
+      "step": 8324
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0832957802716175e-05,
+      "loss": 0.6359,
+      "step": 8325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0830828196597092e-05,
+      "loss": 0.7205,
+      "step": 8326
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0828698644236065e-05,
+      "loss": 0.6446,
+      "step": 8327
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0826569145731006e-05,
+      "loss": 0.6706,
+      "step": 8328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.082443970117983e-05,
+      "loss": 0.6446,
+      "step": 8329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0822310310680447e-05,
+      "loss": 0.6233,
+      "step": 8330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.082018097433077e-05,
+      "loss": 0.6137,
+      "step": 8331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0818051692228701e-05,
+      "loss": 0.6293,
+      "step": 8332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0815922464472144e-05,
+      "loss": 0.6556,
+      "step": 8333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0813793291159e-05,
+      "loss": 0.6475,
+      "step": 8334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0811664172387175e-05,
+      "loss": 0.6385,
+      "step": 8335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0809535108254552e-05,
+      "loss": 0.6967,
+      "step": 8336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.080740609885903e-05,
+      "loss": 0.6742,
+      "step": 8337
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0805277144298504e-05,
+      "loss": 0.6872,
+      "step": 8338
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0803148244670859e-05,
+      "loss": 0.6912,
+      "step": 8339
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.080101940007398e-05,
+      "loss": 0.6184,
+      "step": 8340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.079889061060575e-05,
+      "loss": 0.6804,
+      "step": 8341
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.079676187636406e-05,
+      "loss": 0.5857,
+      "step": 8342
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.079463319744677e-05,
+      "loss": 0.6398,
+      "step": 8343
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.079250457395177e-05,
+      "loss": 0.6831,
+      "step": 8344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0790376005976927e-05,
+      "loss": 0.6593,
+      "step": 8345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0788247493620117e-05,
+      "loss": 0.6196,
+      "step": 8346
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0786119036979204e-05,
+      "loss": 0.6495,
+      "step": 8347
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0783990636152052e-05,
+      "loss": 0.6739,
+      "step": 8348
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0781862291236526e-05,
+      "loss": 0.6767,
+      "step": 8349
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0779734002330493e-05,
+      "loss": 0.6425,
+      "step": 8350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.07776057695318e-05,
+      "loss": 0.7257,
+      "step": 8351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0775477592938307e-05,
+      "loss": 0.581,
+      "step": 8352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0773349472647867e-05,
+      "loss": 0.6441,
+      "step": 8353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0771221408758336e-05,
+      "loss": 0.6626,
+      "step": 8354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0769093401367549e-05,
+      "loss": 0.6365,
+      "step": 8355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0766965450573357e-05,
+      "loss": 0.6491,
+      "step": 8356
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0764837556473604e-05,
+      "loss": 0.6763,
+      "step": 8357
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.076270971916613e-05,
+      "loss": 0.677,
+      "step": 8358
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0760581938748768e-05,
+      "loss": 0.6621,
+      "step": 8359
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0758454215319358e-05,
+      "loss": 0.6672,
+      "step": 8360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.075632654897573e-05,
+      "loss": 0.6394,
+      "step": 8361
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0754198939815709e-05,
+      "loss": 0.6711,
+      "step": 8362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0752071387937125e-05,
+      "loss": 0.5582,
+      "step": 8363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0749943893437801e-05,
+      "loss": 0.624,
+      "step": 8364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0747816456415561e-05,
+      "loss": 0.5845,
+      "step": 8365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0745689076968221e-05,
+      "loss": 0.6797,
+      "step": 8366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0743561755193598e-05,
+      "loss": 0.7076,
+      "step": 8367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0741434491189505e-05,
+      "loss": 0.6503,
+      "step": 8368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0739307285053758e-05,
+      "loss": 0.6266,
+      "step": 8369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0737180136884154e-05,
+      "loss": 0.6094,
+      "step": 8370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0735053046778506e-05,
+      "loss": 0.5499,
+      "step": 8371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0732926014834614e-05,
+      "loss": 0.6309,
+      "step": 8372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0730799041150283e-05,
+      "loss": 0.6324,
+      "step": 8373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0728672125823305e-05,
+      "loss": 0.6224,
+      "step": 8374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0726545268951476e-05,
+      "loss": 0.6473,
+      "step": 8375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0724418470632594e-05,
+      "loss": 0.6104,
+      "step": 8376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0722291730964437e-05,
+      "loss": 0.7305,
+      "step": 8377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0720165050044797e-05,
+      "loss": 0.6327,
+      "step": 8378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.071803842797146e-05,
+      "loss": 0.6583,
+      "step": 8379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0715911864842208e-05,
+      "loss": 0.6489,
+      "step": 8380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0713785360754815e-05,
+      "loss": 0.7304,
+      "step": 8381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0711658915807059e-05,
+      "loss": 0.6394,
+      "step": 8382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0709532530096714e-05,
+      "loss": 0.6172,
+      "step": 8383
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0707406203721553e-05,
+      "loss": 0.622,
+      "step": 8384
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0705279936779335e-05,
+      "loss": 0.6404,
+      "step": 8385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.070315372936783e-05,
+      "loss": 0.63,
+      "step": 8386
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0701027581584802e-05,
+      "loss": 0.6497,
+      "step": 8387
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0698901493528014e-05,
+      "loss": 0.6043,
+      "step": 8388
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.069677546529521e-05,
+      "loss": 0.6799,
+      "step": 8389
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0694649496984152e-05,
+      "loss": 0.5886,
+      "step": 8390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0692523588692592e-05,
+      "loss": 0.6462,
+      "step": 8391
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0690397740518278e-05,
+      "loss": 0.6823,
+      "step": 8392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0688271952558953e-05,
+      "loss": 0.6332,
+      "step": 8393
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0686146224912362e-05,
+      "loss": 0.6195,
+      "step": 8394
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0684020557676248e-05,
+      "loss": 0.7581,
+      "step": 8395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.068189495094834e-05,
+      "loss": 0.7073,
+      "step": 8396
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0679769404826379e-05,
+      "loss": 0.6574,
+      "step": 8397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0677643919408094e-05,
+      "loss": 0.7317,
+      "step": 8398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0675518494791217e-05,
+      "loss": 0.5703,
+      "step": 8399
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.067339313107347e-05,
+      "loss": 0.6303,
+      "step": 8400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.067126782835258e-05,
+      "loss": 0.7239,
+      "step": 8401
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0669142586726266e-05,
+      "loss": 0.6847,
+      "step": 8402
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0667017406292251e-05,
+      "loss": 0.6246,
+      "step": 8403
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.066489228714824e-05,
+      "loss": 0.69,
+      "step": 8404
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.066276722939195e-05,
+      "loss": 0.6325,
+      "step": 8405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0660642233121091e-05,
+      "loss": 0.6299,
+      "step": 8406
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0658517298433371e-05,
+      "loss": 0.7004,
+      "step": 8407
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0656392425426489e-05,
+      "loss": 0.6497,
+      "step": 8408
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.065426761419815e-05,
+      "loss": 0.6747,
+      "step": 8409
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0652142864846053e-05,
+      "loss": 0.6777,
+      "step": 8410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0650018177467887e-05,
+      "loss": 0.6457,
+      "step": 8411
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0647893552161347e-05,
+      "loss": 0.6437,
+      "step": 8412
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0645768989024122e-05,
+      "loss": 0.732,
+      "step": 8413
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0643644488153906e-05,
+      "loss": 0.5871,
+      "step": 8414
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.064152004964837e-05,
+      "loss": 0.6372,
+      "step": 8415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0639395673605204e-05,
+      "loss": 0.6437,
+      "step": 8416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0637271360122083e-05,
+      "loss": 0.7576,
+      "step": 8417
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0635147109296686e-05,
+      "loss": 0.7135,
+      "step": 8418
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0633022921226679e-05,
+      "loss": 0.6694,
+      "step": 8419
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0630898796009732e-05,
+      "loss": 0.5765,
+      "step": 8420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0628774733743515e-05,
+      "loss": 0.6489,
+      "step": 8421
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.062665073452569e-05,
+      "loss": 0.6206,
+      "step": 8422
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0624526798453916e-05,
+      "loss": 0.6494,
+      "step": 8423
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0622402925625854e-05,
+      "loss": 0.7345,
+      "step": 8424
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0620279116139155e-05,
+      "loss": 0.6511,
+      "step": 8425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0618155370091478e-05,
+      "loss": 0.6252,
+      "step": 8426
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0616031687580463e-05,
+      "loss": 0.6191,
+      "step": 8427
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.061390806870376e-05,
+      "loss": 0.5975,
+      "step": 8428
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0611784513559015e-05,
+      "loss": 0.6936,
+      "step": 8429
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0609661022243864e-05,
+      "loss": 0.5941,
+      "step": 8430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0607537594855944e-05,
+      "loss": 0.6866,
+      "step": 8431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0605414231492892e-05,
+      "loss": 0.6472,
+      "step": 8432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0603290932252339e-05,
+      "loss": 0.6585,
+      "step": 8433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0601167697231912e-05,
+      "loss": 0.5532,
+      "step": 8434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0599044526529238e-05,
+      "loss": 0.6591,
+      "step": 8435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0596921420241938e-05,
+      "loss": 0.7029,
+      "step": 8436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0594798378467636e-05,
+      "loss": 0.6391,
+      "step": 8437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0592675401303944e-05,
+      "loss": 0.6897,
+      "step": 8438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0590552488848474e-05,
+      "loss": 0.6823,
+      "step": 8439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.058842964119884e-05,
+      "loss": 0.6511,
+      "step": 8440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0586306858452653e-05,
+      "loss": 0.6481,
+      "step": 8441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.058418414070751e-05,
+      "loss": 0.6165,
+      "step": 8442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0582061488061018e-05,
+      "loss": 0.6648,
+      "step": 8443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0579938900610777e-05,
+      "loss": 0.633,
+      "step": 8444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0577816378454376e-05,
+      "loss": 0.651,
+      "step": 8445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0575693921689413e-05,
+      "loss": 0.6573,
+      "step": 8446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0573571530413475e-05,
+      "loss": 0.643,
+      "step": 8447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0571449204724152e-05,
+      "loss": 0.6649,
+      "step": 8448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0569326944719026e-05,
+      "loss": 0.6042,
+      "step": 8449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0567204750495676e-05,
+      "loss": 0.6002,
+      "step": 8450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0565082622151682e-05,
+      "loss": 0.65,
+      "step": 8451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0562960559784623e-05,
+      "loss": 0.6429,
+      "step": 8452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.056083856349206e-05,
+      "loss": 0.657,
+      "step": 8453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0558716633371566e-05,
+      "loss": 0.6738,
+      "step": 8454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0556594769520708e-05,
+      "loss": 0.6729,
+      "step": 8455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.055447297203705e-05,
+      "loss": 0.6297,
+      "step": 8456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0552351241018148e-05,
+      "loss": 0.6246,
+      "step": 8457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0550229576561558e-05,
+      "loss": 0.6511,
+      "step": 8458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0548107978764837e-05,
+      "loss": 0.6222,
+      "step": 8459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0545986447725538e-05,
+      "loss": 0.6655,
+      "step": 8460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0543864983541198e-05,
+      "loss": 0.5975,
+      "step": 8461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0541743586309366e-05,
+      "loss": 0.6943,
+      "step": 8462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0539622256127588e-05,
+      "loss": 0.6304,
+      "step": 8463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0537500993093393e-05,
+      "loss": 0.6667,
+      "step": 8464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0535379797304319e-05,
+      "loss": 0.5982,
+      "step": 8465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0533258668857901e-05,
+      "loss": 0.6329,
+      "step": 8466
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0531137607851666e-05,
+      "loss": 0.6257,
+      "step": 8467
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0529016614383138e-05,
+      "loss": 0.6558,
+      "step": 8468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0526895688549839e-05,
+      "loss": 0.6919,
+      "step": 8469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.052477483044929e-05,
+      "loss": 0.639,
+      "step": 8470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0522654040179014e-05,
+      "loss": 0.6658,
+      "step": 8471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.052053331783651e-05,
+      "loss": 0.6564,
+      "step": 8472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0518412663519297e-05,
+      "loss": 0.6194,
+      "step": 8473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.051629207732488e-05,
+      "loss": 0.6895,
+      "step": 8474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0514171559350762e-05,
+      "loss": 0.6558,
+      "step": 8475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0512051109694446e-05,
+      "loss": 0.6368,
+      "step": 8476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0509930728453425e-05,
+      "loss": 0.5895,
+      "step": 8477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.05078104157252e-05,
+      "loss": 0.6559,
+      "step": 8478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0505690171607257e-05,
+      "loss": 0.5894,
+      "step": 8479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0503569996197083e-05,
+      "loss": 0.6495,
+      "step": 8480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0501449889592163e-05,
+      "loss": 0.6583,
+      "step": 8481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0499329851889986e-05,
+      "loss": 0.6268,
+      "step": 8482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0497209883188021e-05,
+      "loss": 0.6751,
+      "step": 8483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0495089983583747e-05,
+      "loss": 0.6693,
+      "step": 8484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0492970153174637e-05,
+      "loss": 0.663,
+      "step": 8485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0490850392058164e-05,
+      "loss": 0.6499,
+      "step": 8486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0488730700331786e-05,
+      "loss": 0.65,
+      "step": 8487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0486611078092967e-05,
+      "loss": 0.7105,
+      "step": 8488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0484491525439169e-05,
+      "loss": 0.6491,
+      "step": 8489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.048237204246785e-05,
+      "loss": 0.6735,
+      "step": 8490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0480252629276457e-05,
+      "loss": 0.5519,
+      "step": 8491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0478133285962444e-05,
+      "loss": 0.7303,
+      "step": 8492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0476014012623262e-05,
+      "loss": 0.6629,
+      "step": 8493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0473894809356342e-05,
+      "loss": 0.6895,
+      "step": 8494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0471775676259132e-05,
+      "loss": 0.6344,
+      "step": 8495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0469656613429069e-05,
+      "loss": 0.6555,
+      "step": 8496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.046753762096359e-05,
+      "loss": 0.6364,
+      "step": 8497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0465418698960115e-05,
+      "loss": 0.6859,
+      "step": 8498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0463299847516079e-05,
+      "loss": 0.7169,
+      "step": 8499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0461181066728904e-05,
+      "loss": 0.6781,
+      "step": 8500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0459062356696015e-05,
+      "loss": 0.7162,
+      "step": 8501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0456943717514822e-05,
+      "loss": 0.6358,
+      "step": 8502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0454825149282742e-05,
+      "loss": 0.7555,
+      "step": 8503
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0452706652097187e-05,
+      "loss": 0.6607,
+      "step": 8504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0450588226055569e-05,
+      "loss": 0.6343,
+      "step": 8505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0448469871255283e-05,
+      "loss": 0.739,
+      "step": 8506
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0446351587793733e-05,
+      "loss": 0.6832,
+      "step": 8507
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.044423337576832e-05,
+      "loss": 0.6166,
+      "step": 8508
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.044211523527644e-05,
+      "loss": 0.6524,
+      "step": 8509
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0439997166415477e-05,
+      "loss": 0.6513,
+      "step": 8510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0437879169282826e-05,
+      "loss": 0.6482,
+      "step": 8511
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0435761243975871e-05,
+      "loss": 0.6266,
+      "step": 8512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0433643390591987e-05,
+      "loss": 0.652,
+      "step": 8513
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.043152560922856e-05,
+      "loss": 0.6659,
+      "step": 8514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0429407899982957e-05,
+      "loss": 0.6215,
+      "step": 8515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0427290262952556e-05,
+      "loss": 0.7165,
+      "step": 8516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0425172698234722e-05,
+      "loss": 0.6348,
+      "step": 8517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.042305520592682e-05,
+      "loss": 0.6653,
+      "step": 8518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0420937786126213e-05,
+      "loss": 0.6282,
+      "step": 8519
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.041882043893026e-05,
+      "loss": 0.6211,
+      "step": 8520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0416703164436312e-05,
+      "loss": 0.6312,
+      "step": 8521
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.041458596274172e-05,
+      "loss": 0.6768,
+      "step": 8522
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0412468833943835e-05,
+      "loss": 0.7235,
+      "step": 8523
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0410351778140004e-05,
+      "loss": 0.6497,
+      "step": 8524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0408234795427563e-05,
+      "loss": 0.6795,
+      "step": 8525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0406117885903852e-05,
+      "loss": 0.6642,
+      "step": 8526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.040400104966621e-05,
+      "loss": 0.6579,
+      "step": 8527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0401884286811962e-05,
+      "loss": 0.6731,
+      "step": 8528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0399767597438436e-05,
+      "loss": 0.6111,
+      "step": 8529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0397650981642961e-05,
+      "loss": 0.653,
+      "step": 8530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0395534439522856e-05,
+      "loss": 0.7191,
+      "step": 8531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0393417971175439e-05,
+      "loss": 0.6545,
+      "step": 8532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0391301576698023e-05,
+      "loss": 0.6962,
+      "step": 8533
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.038918525618792e-05,
+      "loss": 0.6297,
+      "step": 8534
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0387069009742442e-05,
+      "loss": 0.6959,
+      "step": 8535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0384952837458883e-05,
+      "loss": 0.6294,
+      "step": 8536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0382836739434552e-05,
+      "loss": 0.6888,
+      "step": 8537
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0380720715766744e-05,
+      "loss": 0.6258,
+      "step": 8538
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0378604766552756e-05,
+      "loss": 0.6763,
+      "step": 8539
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0376488891889873e-05,
+      "loss": 0.5856,
+      "step": 8540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0374373091875381e-05,
+      "loss": 0.711,
+      "step": 8541
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.037225736660657e-05,
+      "loss": 0.64,
+      "step": 8542
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0370141716180719e-05,
+      "loss": 0.6078,
+      "step": 8543
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0368026140695099e-05,
+      "loss": 0.6688,
+      "step": 8544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.036591064024699e-05,
+      "loss": 0.6416,
+      "step": 8545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0363795214933662e-05,
+      "loss": 0.6341,
+      "step": 8546
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0361679864852374e-05,
+      "loss": 0.6281,
+      "step": 8547
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0359564590100397e-05,
+      "loss": 0.6079,
+      "step": 8548
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0357449390774983e-05,
+      "loss": 0.6651,
+      "step": 8549
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0355334266973396e-05,
+      "loss": 0.6044,
+      "step": 8550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0353219218792884e-05,
+      "loss": 0.6871,
+      "step": 8551
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0351104246330694e-05,
+      "loss": 0.6269,
+      "step": 8552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0348989349684076e-05,
+      "loss": 0.6192,
+      "step": 8553
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0346874528950274e-05,
+      "loss": 0.6861,
+      "step": 8554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.034475978422652e-05,
+      "loss": 0.6309,
+      "step": 8555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0342645115610053e-05,
+      "loss": 0.6268,
+      "step": 8556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0340530523198102e-05,
+      "loss": 0.681,
+      "step": 8557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0338416007087898e-05,
+      "loss": 0.6328,
+      "step": 8558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0336301567376665e-05,
+      "loss": 0.5977,
+      "step": 8559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0334187204161623e-05,
+      "loss": 0.7051,
+      "step": 8560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0332072917539994e-05,
+      "loss": 0.7438,
+      "step": 8561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0329958707608984e-05,
+      "loss": 0.6068,
+      "step": 8562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0327844574465809e-05,
+      "loss": 0.6022,
+      "step": 8563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0325730518207673e-05,
+      "loss": 0.7176,
+      "step": 8564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0323616538931785e-05,
+      "loss": 0.6511,
+      "step": 8565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0321502636735336e-05,
+      "loss": 0.6831,
+      "step": 8566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0319388811715531e-05,
+      "loss": 0.6004,
+      "step": 8567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0317275063969558e-05,
+      "loss": 0.6775,
+      "step": 8568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0315161393594612e-05,
+      "loss": 0.6327,
+      "step": 8569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0313047800687871e-05,
+      "loss": 0.5993,
+      "step": 8570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0310934285346519e-05,
+      "loss": 0.6991,
+      "step": 8571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.030882084766774e-05,
+      "loss": 0.6565,
+      "step": 8572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0306707487748705e-05,
+      "loss": 0.6603,
+      "step": 8573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0304594205686582e-05,
+      "loss": 0.6235,
+      "step": 8574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0302481001578544e-05,
+      "loss": 0.6235,
+      "step": 8575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0300367875521755e-05,
+      "loss": 0.6187,
+      "step": 8576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0298254827613374e-05,
+      "loss": 0.6865,
+      "step": 8577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0296141857950558e-05,
+      "loss": 0.7762,
+      "step": 8578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.029402896663046e-05,
+      "loss": 0.6159,
+      "step": 8579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0291916153750237e-05,
+      "loss": 0.6074,
+      "step": 8580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0289803419407023e-05,
+      "loss": 0.6108,
+      "step": 8581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0287690763697969e-05,
+      "loss": 0.6181,
+      "step": 8582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.028557818672021e-05,
+      "loss": 0.6767,
+      "step": 8583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0283465688570887e-05,
+      "loss": 0.6815,
+      "step": 8584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0281353269347127e-05,
+      "loss": 0.629,
+      "step": 8585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0279240929146058e-05,
+      "loss": 0.7004,
+      "step": 8586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0277128668064806e-05,
+      "loss": 0.6431,
+      "step": 8587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0275016486200498e-05,
+      "loss": 0.7124,
+      "step": 8588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.027290438365024e-05,
+      "loss": 0.5869,
+      "step": 8589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.027079236051115e-05,
+      "loss": 0.7217,
+      "step": 8590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.026868041688034e-05,
+      "loss": 0.561,
+      "step": 8591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0266568552854917e-05,
+      "loss": 0.6356,
+      "step": 8592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0264456768531981e-05,
+      "loss": 0.6696,
+      "step": 8593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.026234506400863e-05,
+      "loss": 0.6783,
+      "step": 8594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0260233439381967e-05,
+      "loss": 0.6491,
+      "step": 8595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0258121894749073e-05,
+      "loss": 0.6705,
+      "step": 8596
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.025601043020704e-05,
+      "loss": 0.6759,
+      "step": 8597
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0253899045852954e-05,
+      "loss": 0.5803,
+      "step": 8598
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0251787741783898e-05,
+      "loss": 0.6522,
+      "step": 8599
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0249676518096942e-05,
+      "loss": 0.7011,
+      "step": 8600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0247565374889163e-05,
+      "loss": 0.6158,
+      "step": 8601
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0245454312257632e-05,
+      "loss": 0.5683,
+      "step": 8602
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0243343330299415e-05,
+      "loss": 0.6617,
+      "step": 8603
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.024123242911157e-05,
+      "loss": 0.6799,
+      "step": 8604
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0239121608791159e-05,
+      "loss": 0.6596,
+      "step": 8605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0237010869435233e-05,
+      "loss": 0.6131,
+      "step": 8606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0234900211140847e-05,
+      "loss": 0.6583,
+      "step": 8607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0232789634005047e-05,
+      "loss": 0.6688,
+      "step": 8608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0230679138124874e-05,
+      "loss": 0.6429,
+      "step": 8609
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0228568723597373e-05,
+      "loss": 0.6744,
+      "step": 8610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.022645839051958e-05,
+      "loss": 0.593,
+      "step": 8611
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0224348138988518e-05,
+      "loss": 0.6594,
+      "step": 8612
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0222237969101223e-05,
+      "loss": 0.7021,
+      "step": 8613
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0220127880954722e-05,
+      "loss": 0.6713,
+      "step": 8614
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0218017874646029e-05,
+      "loss": 0.709,
+      "step": 8615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0215907950272166e-05,
+      "loss": 0.706,
+      "step": 8616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0213798107930144e-05,
+      "loss": 0.6024,
+      "step": 8617
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0211688347716975e-05,
+      "loss": 0.6106,
+      "step": 8618
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0209578669729662e-05,
+      "loss": 0.6723,
+      "step": 8619
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0207469074065212e-05,
+      "loss": 0.6769,
+      "step": 8620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0205359560820618e-05,
+      "loss": 0.6546,
+      "step": 8621
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0203250130092881e-05,
+      "loss": 0.7035,
+      "step": 8622
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0201140781978985e-05,
+      "loss": 0.7068,
+      "step": 8623
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0199031516575918e-05,
+      "loss": 0.6788,
+      "step": 8624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0196922333980665e-05,
+      "loss": 0.6305,
+      "step": 8625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.019481323429021e-05,
+      "loss": 0.6919,
+      "step": 8626
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0192704217601518e-05,
+      "loss": 0.6729,
+      "step": 8627
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.019059528401157e-05,
+      "loss": 0.6368,
+      "step": 8628
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0188486433617334e-05,
+      "loss": 0.6695,
+      "step": 8629
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0186377666515766e-05,
+      "loss": 0.6817,
+      "step": 8630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.018426898280383e-05,
+      "loss": 0.6025,
+      "step": 8631
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0182160382578485e-05,
+      "loss": 0.6358,
+      "step": 8632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0180051865936683e-05,
+      "loss": 0.5756,
+      "step": 8633
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.017794343297537e-05,
+      "loss": 0.6281,
+      "step": 8634
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0175835083791492e-05,
+      "loss": 0.6498,
+      "step": 8635
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0173726818481992e-05,
+      "loss": 0.7053,
+      "step": 8636
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0171618637143809e-05,
+      "loss": 0.6276,
+      "step": 8637
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.016951053987387e-05,
+      "loss": 0.6365,
+      "step": 8638
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0167402526769108e-05,
+      "loss": 0.6147,
+      "step": 8639
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0165294597926447e-05,
+      "loss": 0.5716,
+      "step": 8640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0163186753442813e-05,
+      "loss": 0.6338,
+      "step": 8641
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0161078993415118e-05,
+      "loss": 0.6484,
+      "step": 8642
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0158971317940279e-05,
+      "loss": 0.6354,
+      "step": 8643
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.015686372711521e-05,
+      "loss": 0.683,
+      "step": 8644
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.015475622103681e-05,
+      "loss": 0.6375,
+      "step": 8645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0152648799801983e-05,
+      "loss": 0.6567,
+      "step": 8646
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0150541463507631e-05,
+      "loss": 0.6367,
+      "step": 8647
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0148434212250648e-05,
+      "loss": 0.6501,
+      "step": 8648
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0146327046127919e-05,
+      "loss": 0.667,
+      "step": 8649
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0144219965236336e-05,
+      "loss": 0.652,
+      "step": 8650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.014211296967278e-05,
+      "loss": 0.6502,
+      "step": 8651
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.014000605953413e-05,
+      "loss": 0.6508,
+      "step": 8652
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0137899234917262e-05,
+      "loss": 0.5888,
+      "step": 8653
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0135792495919044e-05,
+      "loss": 0.6608,
+      "step": 8654
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0133685842636346e-05,
+      "loss": 0.6698,
+      "step": 8655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0131579275166033e-05,
+      "loss": 0.6614,
+      "step": 8656
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0129472793604959e-05,
+      "loss": 0.556,
+      "step": 8657
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0127366398049978e-05,
+      "loss": 0.6006,
+      "step": 8658
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0125260088597947e-05,
+      "loss": 0.6628,
+      "step": 8659
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0123153865345715e-05,
+      "loss": 0.6434,
+      "step": 8660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0121047728390115e-05,
+      "loss": 0.6754,
+      "step": 8661
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0118941677827993e-05,
+      "loss": 0.7009,
+      "step": 8662
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.011683571375619e-05,
+      "loss": 0.6847,
+      "step": 8663
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0114729836271527e-05,
+      "loss": 0.6803,
+      "step": 8664
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0112624045470834e-05,
+      "loss": 0.6035,
+      "step": 8665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0110518341450938e-05,
+      "loss": 0.6313,
+      "step": 8666
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0108412724308657e-05,
+      "loss": 0.6257,
+      "step": 8667
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0106307194140805e-05,
+      "loss": 0.6363,
+      "step": 8668
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0104201751044194e-05,
+      "loss": 0.6341,
+      "step": 8669
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.010209639511563e-05,
+      "loss": 0.6872,
+      "step": 8670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0099991126451924e-05,
+      "loss": 0.5886,
+      "step": 8671
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0097885945149865e-05,
+      "loss": 0.6851,
+      "step": 8672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0095780851306255e-05,
+      "loss": 0.6091,
+      "step": 8673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.009367584501788e-05,
+      "loss": 0.6414,
+      "step": 8674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0091570926381537e-05,
+      "loss": 0.6997,
+      "step": 8675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0089466095493997e-05,
+      "loss": 0.6613,
+      "step": 8676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.008736135245205e-05,
+      "loss": 0.7012,
+      "step": 8677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0085256697352467e-05,
+      "loss": 0.6204,
+      "step": 8678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0083152130292015e-05,
+      "loss": 0.6734,
+      "step": 8679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0081047651367466e-05,
+      "loss": 0.6224,
+      "step": 8680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0078943260675582e-05,
+      "loss": 0.6651,
+      "step": 8681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0076838958313128e-05,
+      "loss": 0.6086,
+      "step": 8682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0074734744376847e-05,
+      "loss": 0.5795,
+      "step": 8683
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0072630618963493e-05,
+      "loss": 0.7162,
+      "step": 8684
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.007052658216982e-05,
+      "loss": 0.6375,
+      "step": 8685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.006842263409257e-05,
+      "loss": 0.6302,
+      "step": 8686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0066318774828474e-05,
+      "loss": 0.6929,
+      "step": 8687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.006421500447427e-05,
+      "loss": 0.7019,
+      "step": 8688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0062111323126691e-05,
+      "loss": 0.6463,
+      "step": 8689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0060007730882468e-05,
+      "loss": 0.6198,
+      "step": 8690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.005790422783831e-05,
+      "loss": 0.6372,
+      "step": 8691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0055800814090943e-05,
+      "loss": 0.63,
+      "step": 8692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0053697489737083e-05,
+      "loss": 0.5711,
+      "step": 8693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0051594254873436e-05,
+      "loss": 0.6473,
+      "step": 8694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.004949110959671e-05,
+      "loss": 0.5931,
+      "step": 8695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0047388054003606e-05,
+      "loss": 0.6642,
+      "step": 8696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0045285088190825e-05,
+      "loss": 0.6383,
+      "step": 8697
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0043182212255055e-05,
+      "loss": 0.6576,
+      "step": 8698
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0041079426292987e-05,
+      "loss": 0.6118,
+      "step": 8699
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0038976730401307e-05,
+      "loss": 0.6765,
+      "step": 8700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0036874124676698e-05,
+      "loss": 0.6453,
+      "step": 8701
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0034771609215835e-05,
+      "loss": 0.6395,
+      "step": 8702
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.003266918411539e-05,
+      "loss": 0.5664,
+      "step": 8703
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0030566849472033e-05,
+      "loss": 0.5902,
+      "step": 8704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0028464605382432e-05,
+      "loss": 0.6902,
+      "step": 8705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.002636245194324e-05,
+      "loss": 0.6463,
+      "step": 8706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0024260389251117e-05,
+      "loss": 0.6418,
+      "step": 8707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0022158417402714e-05,
+      "loss": 0.7338,
+      "step": 8708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0020056536494682e-05,
+      "loss": 0.6647,
+      "step": 8709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.001795474662366e-05,
+      "loss": 0.6037,
+      "step": 8710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.001585304788629e-05,
+      "loss": 0.6799,
+      "step": 8711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0013751440379212e-05,
+      "loss": 0.6358,
+      "step": 8712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0011649924199046e-05,
+      "loss": 0.6246,
+      "step": 8713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0009548499442428e-05,
+      "loss": 0.7009,
+      "step": 8714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0007447166205974e-05,
+      "loss": 0.6446,
+      "step": 8715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0005345924586312e-05,
+      "loss": 0.6888,
+      "step": 8716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0003244774680045e-05,
+      "loss": 0.6367,
+      "step": 8717
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0001143716583789e-05,
+      "loss": 0.6181,
+      "step": 8718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.99904275039415e-06,
+      "loss": 0.6026,
+      "step": 8719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.996941876207735e-06,
+      "loss": 0.7419,
+      "step": 8720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.994841094121127e-06,
+      "loss": 0.6717,
+      "step": 8721
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.99274040423093e-06,
+      "loss": 0.6699,
+      "step": 8722
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.99063980663373e-06,
+      "loss": 0.6548,
+      "step": 8723
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.988539301426115e-06,
+      "loss": 0.6433,
+      "step": 8724
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.98643888870466e-06,
+      "loss": 0.6034,
+      "step": 8725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.98433856856594e-06,
+      "loss": 0.6237,
+      "step": 8726
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.982238341106532e-06,
+      "loss": 0.6295,
+      "step": 8727
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.980138206423005e-06,
+      "loss": 0.6456,
+      "step": 8728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.978038164611916e-06,
+      "loss": 0.7327,
+      "step": 8729
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.975938215769827e-06,
+      "loss": 0.6447,
+      "step": 8730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.973838359993297e-06,
+      "loss": 0.672,
+      "step": 8731
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.97173859737887e-06,
+      "loss": 0.6554,
+      "step": 8732
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.969638928023094e-06,
+      "loss": 0.6149,
+      "step": 8733
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.967539352022513e-06,
+      "loss": 0.6796,
+      "step": 8734
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.965439869473663e-06,
+      "loss": 0.6013,
+      "step": 8735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.963340480473077e-06,
+      "loss": 0.6345,
+      "step": 8736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.961241185117286e-06,
+      "loss": 0.6277,
+      "step": 8737
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.959141983502813e-06,
+      "loss": 0.7265,
+      "step": 8738
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.957042875726183e-06,
+      "loss": 0.675,
+      "step": 8739
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.954943861883906e-06,
+      "loss": 0.6169,
+      "step": 8740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.952844942072495e-06,
+      "loss": 0.6095,
+      "step": 8741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.950746116388458e-06,
+      "loss": 0.6648,
+      "step": 8742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.948647384928302e-06,
+      "loss": 0.6311,
+      "step": 8743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.94654874778852e-06,
+      "loss": 0.6177,
+      "step": 8744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.944450205065609e-06,
+      "loss": 0.65,
+      "step": 8745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.942351756856064e-06,
+      "loss": 0.7307,
+      "step": 8746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.94025340325636e-06,
+      "loss": 0.6793,
+      "step": 8747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.938155144362986e-06,
+      "loss": 0.6496,
+      "step": 8748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.936056980272418e-06,
+      "loss": 0.6868,
+      "step": 8749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.93395891108113e-06,
+      "loss": 0.6471,
+      "step": 8750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.931860936885587e-06,
+      "loss": 0.7215,
+      "step": 8751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.929763057782255e-06,
+      "loss": 0.6387,
+      "step": 8752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.927665273867594e-06,
+      "loss": 0.6736,
+      "step": 8753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.925567585238062e-06,
+      "loss": 0.6025,
+      "step": 8754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.923469991990104e-06,
+      "loss": 0.6304,
+      "step": 8755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.921372494220168e-06,
+      "loss": 0.7384,
+      "step": 8756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.919275092024697e-06,
+      "loss": 0.6542,
+      "step": 8757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.917177785500133e-06,
+      "loss": 0.65,
+      "step": 8758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.915080574742901e-06,
+      "loss": 0.6868,
+      "step": 8759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.912983459849437e-06,
+      "loss": 0.6564,
+      "step": 8760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.91088644091616e-06,
+      "loss": 0.6437,
+      "step": 8761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.908789518039495e-06,
+      "loss": 0.6888,
+      "step": 8762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.906692691315855e-06,
+      "loss": 0.7131,
+      "step": 8763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.904595960841649e-06,
+      "loss": 0.6238,
+      "step": 8764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.902499326713294e-06,
+      "loss": 0.6459,
+      "step": 8765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.900402789027179e-06,
+      "loss": 0.6889,
+      "step": 8766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.898306347879707e-06,
+      "loss": 0.6741,
+      "step": 8767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.896210003367275e-06,
+      "loss": 0.6255,
+      "step": 8768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.89411375558627e-06,
+      "loss": 0.6746,
+      "step": 8769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.892017604633074e-06,
+      "loss": 0.677,
+      "step": 8770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.889921550604071e-06,
+      "loss": 0.6587,
+      "step": 8771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.887825593595635e-06,
+      "loss": 0.6935,
+      "step": 8772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.88572973370414e-06,
+      "loss": 0.6226,
+      "step": 8773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.883633971025947e-06,
+      "loss": 0.7361,
+      "step": 8774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.88153830565742e-06,
+      "loss": 0.6165,
+      "step": 8775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.879442737694921e-06,
+      "loss": 0.662,
+      "step": 8776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.877347267234802e-06,
+      "loss": 0.6681,
+      "step": 8777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.875251894373408e-06,
+      "loss": 0.6052,
+      "step": 8778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.873156619207085e-06,
+      "loss": 0.5815,
+      "step": 8779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.87106144183218e-06,
+      "loss": 0.6383,
+      "step": 8780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.868966362345018e-06,
+      "loss": 0.6735,
+      "step": 8781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.866871380841933e-06,
+      "loss": 0.6418,
+      "step": 8782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.864776497419255e-06,
+      "loss": 0.6199,
+      "step": 8783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.862681712173304e-06,
+      "loss": 0.6832,
+      "step": 8784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.860587025200397e-06,
+      "loss": 0.6467,
+      "step": 8785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.858492436596845e-06,
+      "loss": 0.6529,
+      "step": 8786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.856397946458958e-06,
+      "loss": 0.6472,
+      "step": 8787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.854303554883045e-06,
+      "loss": 0.646,
+      "step": 8788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.852209261965398e-06,
+      "loss": 0.6117,
+      "step": 8789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.850115067802311e-06,
+      "loss": 0.6652,
+      "step": 8790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.84802097249008e-06,
+      "loss": 0.6056,
+      "step": 8791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.845926976124988e-06,
+      "loss": 0.64,
+      "step": 8792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.843833078803315e-06,
+      "loss": 0.6685,
+      "step": 8793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.841739280621339e-06,
+      "loss": 0.6371,
+      "step": 8794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.839645581675336e-06,
+      "loss": 0.6283,
+      "step": 8795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.837551982061564e-06,
+      "loss": 0.601,
+      "step": 8796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.835458481876293e-06,
+      "loss": 0.6465,
+      "step": 8797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.833365081215778e-06,
+      "loss": 0.612,
+      "step": 8798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.831271780176278e-06,
+      "loss": 0.6571,
+      "step": 8799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.829178578854034e-06,
+      "loss": 0.6915,
+      "step": 8800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.827085477345296e-06,
+      "loss": 0.6716,
+      "step": 8801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.824992475746302e-06,
+      "loss": 0.659,
+      "step": 8802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.82289957415329e-06,
+      "loss": 0.593,
+      "step": 8803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.820806772662488e-06,
+      "loss": 0.564,
+      "step": 8804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.818714071370122e-06,
+      "loss": 0.6348,
+      "step": 8805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.816621470372414e-06,
+      "loss": 0.6158,
+      "step": 8806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.81452896976559e-06,
+      "loss": 0.6699,
+      "step": 8807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.812436569645845e-06,
+      "loss": 0.686,
+      "step": 8808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.810344270109397e-06,
+      "loss": 0.695,
+      "step": 8809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.808252071252448e-06,
+      "loss": 0.6545,
+      "step": 8810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.806159973171199e-06,
+      "loss": 0.6065,
+      "step": 8811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.804067975961838e-06,
+      "loss": 0.6054,
+      "step": 8812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.801976079720559e-06,
+      "loss": 0.5818,
+      "step": 8813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.799884284543548e-06,
+      "loss": 0.6517,
+      "step": 8814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.797792590526977e-06,
+      "loss": 0.6794,
+      "step": 8815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.795700997767028e-06,
+      "loss": 0.696,
+      "step": 8816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.793609506359867e-06,
+      "loss": 0.6619,
+      "step": 8817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.791518116401665e-06,
+      "loss": 0.6009,
+      "step": 8818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.78942682798858e-06,
+      "loss": 0.6747,
+      "step": 8819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.787335641216771e-06,
+      "loss": 0.6673,
+      "step": 8820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.785244556182387e-06,
+      "loss": 0.6305,
+      "step": 8821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.78315357298158e-06,
+      "loss": 0.7261,
+      "step": 8822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.781062691710486e-06,
+      "loss": 0.6761,
+      "step": 8823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.778971912465244e-06,
+      "loss": 0.6828,
+      "step": 8824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.776881235341991e-06,
+      "loss": 0.6769,
+      "step": 8825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.774790660436857e-06,
+      "loss": 0.6279,
+      "step": 8826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.772700187845958e-06,
+      "loss": 0.6005,
+      "step": 8827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.770609817665421e-06,
+      "loss": 0.6267,
+      "step": 8828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.768519549991359e-06,
+      "loss": 0.7332,
+      "step": 8829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.766429384919877e-06,
+      "loss": 0.6844,
+      "step": 8830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.76433932254708e-06,
+      "loss": 0.713,
+      "step": 8831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.762249362969075e-06,
+      "loss": 0.5843,
+      "step": 8832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.760159506281955e-06,
+      "loss": 0.741,
+      "step": 8833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.758069752581807e-06,
+      "loss": 0.6282,
+      "step": 8834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.755980101964716e-06,
+      "loss": 0.6605,
+      "step": 8835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.75389055452677e-06,
+      "loss": 0.7092,
+      "step": 8836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.751801110364042e-06,
+      "loss": 0.6565,
+      "step": 8837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.749711769572603e-06,
+      "loss": 0.6579,
+      "step": 8838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.74762253224852e-06,
+      "loss": 0.6764,
+      "step": 8839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.74553339848786e-06,
+      "loss": 0.553,
+      "step": 8840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.743444368386678e-06,
+      "loss": 0.5853,
+      "step": 8841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.741355442041021e-06,
+      "loss": 0.6567,
+      "step": 8842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.739266619546944e-06,
+      "loss": 0.6289,
+      "step": 8843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.737177901000485e-06,
+      "loss": 0.6831,
+      "step": 8844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.735089286497689e-06,
+      "loss": 0.6545,
+      "step": 8845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.733000776134585e-06,
+      "loss": 0.6355,
+      "step": 8846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.730912370007202e-06,
+      "loss": 0.6448,
+      "step": 8847
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.728824068211568e-06,
+      "loss": 0.6378,
+      "step": 8848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.726735870843696e-06,
+      "loss": 0.6761,
+      "step": 8849
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.724647777999603e-06,
+      "loss": 0.6183,
+      "step": 8850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.7225597897753e-06,
+      "loss": 0.7216,
+      "step": 8851
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.720471906266795e-06,
+      "loss": 0.7267,
+      "step": 8852
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.71838412757008e-06,
+      "loss": 0.6569,
+      "step": 8853
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.716296453781155e-06,
+      "loss": 0.7152,
+      "step": 8854
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.714208884996011e-06,
+      "loss": 0.6142,
+      "step": 8855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.712121421310636e-06,
+      "loss": 0.6474,
+      "step": 8856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.710034062821003e-06,
+      "loss": 0.7111,
+      "step": 8857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.707946809623093e-06,
+      "loss": 0.6319,
+      "step": 8858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.705859661812875e-06,
+      "loss": 0.5963,
+      "step": 8859
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.70377261948632e-06,
+      "loss": 0.5537,
+      "step": 8860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.701685682739384e-06,
+      "loss": 0.6501,
+      "step": 8861
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.699598851668024e-06,
+      "loss": 0.6029,
+      "step": 8862
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.697512126368198e-06,
+      "loss": 0.615,
+      "step": 8863
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.695425506935843e-06,
+      "loss": 0.6684,
+      "step": 8864
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.693338993466907e-06,
+      "loss": 0.6357,
+      "step": 8865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.691252586057324e-06,
+      "loss": 0.6035,
+      "step": 8866
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.689166284803033e-06,
+      "loss": 0.6437,
+      "step": 8867
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.687080089799954e-06,
+      "loss": 0.6421,
+      "step": 8868
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.68499400114401e-06,
+      "loss": 0.6659,
+      "step": 8869
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.682908018931119e-06,
+      "loss": 0.6892,
+      "step": 8870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.6808221432572e-06,
+      "loss": 0.6371,
+      "step": 8871
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.678736374218151e-06,
+      "loss": 0.6341,
+      "step": 8872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.67665071190988e-06,
+      "loss": 0.7001,
+      "step": 8873
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.674565156428286e-06,
+      "loss": 0.6596,
+      "step": 8874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.672479707869265e-06,
+      "loss": 0.7045,
+      "step": 8875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.670394366328695e-06,
+      "loss": 0.6746,
+      "step": 8876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.668309131902464e-06,
+      "loss": 0.6337,
+      "step": 8877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.666224004686452e-06,
+      "loss": 0.6805,
+      "step": 8878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.664138984776532e-06,
+      "loss": 0.5983,
+      "step": 8879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.662054072268571e-06,
+      "loss": 0.6839,
+      "step": 8880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.659969267258433e-06,
+      "loss": 0.6622,
+      "step": 8881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.65788456984198e-06,
+      "loss": 0.6036,
+      "step": 8882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.655799980115058e-06,
+      "loss": 0.6862,
+      "step": 8883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.65371549817352e-06,
+      "loss": 0.6188,
+      "step": 8884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.65163112411321e-06,
+      "loss": 0.6241,
+      "step": 8885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.649546858029968e-06,
+      "loss": 0.6292,
+      "step": 8886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.647462700019623e-06,
+      "loss": 0.6768,
+      "step": 8887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.645378650178007e-06,
+      "loss": 0.6327,
+      "step": 8888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.643294708600943e-06,
+      "loss": 0.6807,
+      "step": 8889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.641210875384254e-06,
+      "loss": 0.7251,
+      "step": 8890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.639127150623748e-06,
+      "loss": 0.6608,
+      "step": 8891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.637043534415233e-06,
+      "loss": 0.6184,
+      "step": 8892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.634960026854518e-06,
+      "loss": 0.6853,
+      "step": 8893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.6328766280374e-06,
+      "loss": 0.6632,
+      "step": 8894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.630793338059672e-06,
+      "loss": 0.6837,
+      "step": 8895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.628710157017122e-06,
+      "loss": 0.5533,
+      "step": 8896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.626627085005539e-06,
+      "loss": 0.6181,
+      "step": 8897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.624544122120693e-06,
+      "loss": 0.6734,
+      "step": 8898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.622461268458365e-06,
+      "loss": 0.6334,
+      "step": 8899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.62037852411432e-06,
+      "loss": 0.6367,
+      "step": 8900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.618295889184324e-06,
+      "loss": 0.7127,
+      "step": 8901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.616213363764135e-06,
+      "loss": 0.6761,
+      "step": 8902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.614130947949507e-06,
+      "loss": 0.6733,
+      "step": 8903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.612048641836185e-06,
+      "loss": 0.6415,
+      "step": 8904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.609966445519923e-06,
+      "loss": 0.6431,
+      "step": 8905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.607884359096448e-06,
+      "loss": 0.5542,
+      "step": 8906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.605802382661497e-06,
+      "loss": 0.6695,
+      "step": 8907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.603720516310799e-06,
+      "loss": 0.6064,
+      "step": 8908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.601638760140083e-06,
+      "loss": 0.6249,
+      "step": 8909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.599557114245057e-06,
+      "loss": 0.6642,
+      "step": 8910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.59747557872144e-06,
+      "loss": 0.6046,
+      "step": 8911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.59539415366494e-06,
+      "loss": 0.5875,
+      "step": 8912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.593312839171261e-06,
+      "loss": 0.6732,
+      "step": 8913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.591231635336099e-06,
+      "loss": 0.6541,
+      "step": 8914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.589150542255144e-06,
+      "loss": 0.6297,
+      "step": 8915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.587069560024095e-06,
+      "loss": 0.6241,
+      "step": 8916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.584988688738621e-06,
+      "loss": 0.7436,
+      "step": 8917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.582907928494407e-06,
+      "loss": 0.6788,
+      "step": 8918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.580827279387125e-06,
+      "loss": 0.6148,
+      "step": 8919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.578746741512442e-06,
+      "loss": 0.6048,
+      "step": 8920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.576666314966016e-06,
+      "loss": 0.654,
+      "step": 8921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.574585999843512e-06,
+      "loss": 0.6219,
+      "step": 8922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.572505796240577e-06,
+      "loss": 0.7027,
+      "step": 8923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.570425704252863e-06,
+      "loss": 0.6498,
+      "step": 8924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.568345723976003e-06,
+      "loss": 0.6039,
+      "step": 8925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.566265855505641e-06,
+      "loss": 0.6387,
+      "step": 8926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.564186098937405e-06,
+      "loss": 0.688,
+      "step": 8927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.562106454366924e-06,
+      "loss": 0.6051,
+      "step": 8928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.560026921889815e-06,
+      "loss": 0.6402,
+      "step": 8929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.557947501601699e-06,
+      "loss": 0.6423,
+      "step": 8930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.555868193598188e-06,
+      "loss": 0.628,
+      "step": 8931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.55378899797488e-06,
+      "loss": 0.5858,
+      "step": 8932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.551709914827379e-06,
+      "loss": 0.6172,
+      "step": 8933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.549630944251284e-06,
+      "loss": 0.7249,
+      "step": 8934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.547552086342183e-06,
+      "loss": 0.6415,
+      "step": 8935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.54547334119566e-06,
+      "loss": 0.6175,
+      "step": 8936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.543394708907294e-06,
+      "loss": 0.706,
+      "step": 8937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.541316189572664e-06,
+      "loss": 0.6219,
+      "step": 8938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.53923778328734e-06,
+      "loss": 0.625,
+      "step": 8939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.537159490146877e-06,
+      "loss": 0.5744,
+      "step": 8940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.535081310246843e-06,
+      "loss": 0.6447,
+      "step": 8941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.533003243682787e-06,
+      "loss": 0.646,
+      "step": 8942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.530925290550265e-06,
+      "loss": 0.6727,
+      "step": 8943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.528847450944811e-06,
+      "loss": 0.592,
+      "step": 8944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.52676972496197e-06,
+      "loss": 0.6553,
+      "step": 8945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.52469211269727e-06,
+      "loss": 0.6865,
+      "step": 8946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.522614614246243e-06,
+      "loss": 0.6235,
+      "step": 8947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.520537229704406e-06,
+      "loss": 0.6468,
+      "step": 8948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.518459959167284e-06,
+      "loss": 0.6377,
+      "step": 8949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.516382802730386e-06,
+      "loss": 0.6947,
+      "step": 8950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.514305760489216e-06,
+      "loss": 0.6273,
+      "step": 8951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.512228832539278e-06,
+      "loss": 0.6649,
+      "step": 8952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.510152018976067e-06,
+      "loss": 0.6456,
+      "step": 8953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.508075319895079e-06,
+      "loss": 0.6639,
+      "step": 8954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.50599873539179e-06,
+      "loss": 0.6125,
+      "step": 8955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.503922265561689e-06,
+      "loss": 0.7272,
+      "step": 8956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.501845910500248e-06,
+      "loss": 0.6694,
+      "step": 8957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.499769670302943e-06,
+      "loss": 0.6324,
+      "step": 8958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.497693545065227e-06,
+      "loss": 0.6988,
+      "step": 8959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.495617534882567e-06,
+      "loss": 0.6768,
+      "step": 8960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.493541639850417e-06,
+      "loss": 0.6534,
+      "step": 8961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.491465860064224e-06,
+      "loss": 0.6641,
+      "step": 8962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.489390195619432e-06,
+      "loss": 0.7026,
+      "step": 8963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.487314646611479e-06,
+      "loss": 0.6071,
+      "step": 8964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.485239213135802e-06,
+      "loss": 0.628,
+      "step": 8965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.483163895287821e-06,
+      "loss": 0.6554,
+      "step": 8966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.481088693162962e-06,
+      "loss": 0.629,
+      "step": 8967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.479013606856642e-06,
+      "loss": 0.6174,
+      "step": 8968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.476938636464276e-06,
+      "loss": 0.6401,
+      "step": 8969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.474863782081265e-06,
+      "loss": 0.6552,
+      "step": 8970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.47278904380301e-06,
+      "loss": 0.6887,
+      "step": 8971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.470714421724911e-06,
+      "loss": 0.6436,
+      "step": 8972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.46863991594236e-06,
+      "loss": 0.616,
+      "step": 8973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.466565526550733e-06,
+      "loss": 0.5954,
+      "step": 8974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.464491253645415e-06,
+      "loss": 0.6712,
+      "step": 8975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.46241709732178e-06,
+      "loss": 0.6462,
+      "step": 8976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.4603430576752e-06,
+      "loss": 0.6573,
+      "step": 8977
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.45826913480103e-06,
+      "loss": 0.6158,
+      "step": 8978
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.456195328794635e-06,
+      "loss": 0.702,
+      "step": 8979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.45412163975137e-06,
+      "loss": 0.7417,
+      "step": 8980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.452048067766574e-06,
+      "loss": 0.66,
+      "step": 8981
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.449974612935597e-06,
+      "loss": 0.6466,
+      "step": 8982
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.447901275353768e-06,
+      "loss": 0.6056,
+      "step": 8983
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.445828055116429e-06,
+      "loss": 0.6143,
+      "step": 8984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.443754952318894e-06,
+      "loss": 0.6479,
+      "step": 8985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.44168196705649e-06,
+      "loss": 0.6701,
+      "step": 8986
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.439609099424529e-06,
+      "loss": 0.6163,
+      "step": 8987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.437536349518326e-06,
+      "loss": 0.7088,
+      "step": 8988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.43546371743318e-06,
+      "loss": 0.6944,
+      "step": 8989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.43339120326439e-06,
+      "loss": 0.6591,
+      "step": 8990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.431318807107251e-06,
+      "loss": 0.6703,
+      "step": 8991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.429246529057055e-06,
+      "loss": 0.6749,
+      "step": 8992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.427174369209075e-06,
+      "loss": 0.6002,
+      "step": 8993
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.425102327658595e-06,
+      "loss": 0.6661,
+      "step": 8994
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.423030404500884e-06,
+      "loss": 0.6397,
+      "step": 8995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.420958599831212e-06,
+      "loss": 0.6047,
+      "step": 8996
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.418886913744834e-06,
+      "loss": 0.6381,
+      "step": 8997
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.416815346337009e-06,
+      "loss": 0.6639,
+      "step": 8998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.41474389770299e-06,
+      "loss": 0.6245,
+      "step": 8999
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.412672567938013e-06,
+      "loss": 0.6597,
+      "step": 9000
+    },
+    {
+      "epoch": 0.6,
+      "eval_text_loss": 0.6600448489189148,
+      "eval_text_runtime": 1.8654,
+      "eval_text_samples_per_second": 39.67,
+      "eval_text_steps_per_second": 1.072,
+      "step": 9000
+    },
+    {
+      "epoch": 0.6,
+      "eval_svit_loss": 0.7931904792785645,
+      "eval_svit_runtime": 6.6038,
+      "eval_svit_samples_per_second": 123.565,
+      "eval_svit_steps_per_second": 1.969,
+      "step": 9000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.410601357137321e-06,
+      "loss": 0.6399,
+      "step": 9001
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.40853026539615e-06,
+      "loss": 0.6276,
+      "step": 9002
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.406459292809725e-06,
+      "loss": 0.5941,
+      "step": 9003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.40438843947327e-06,
+      "loss": 0.6011,
+      "step": 9004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.402317705481998e-06,
+      "loss": 0.6457,
+      "step": 9005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.400247090931127e-06,
+      "loss": 0.6724,
+      "step": 9006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.398176595915863e-06,
+      "loss": 0.6475,
+      "step": 9007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.396106220531399e-06,
+      "loss": 0.6721,
+      "step": 9008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.394035964872934e-06,
+      "loss": 0.6157,
+      "step": 9009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.391965829035659e-06,
+      "loss": 0.6988,
+      "step": 9010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.389895813114757e-06,
+      "loss": 0.6532,
+      "step": 9011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.387825917205407e-06,
+      "loss": 0.6164,
+      "step": 9012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.38575614140278e-06,
+      "loss": 0.6094,
+      "step": 9013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.38368648580205e-06,
+      "loss": 0.6071,
+      "step": 9014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.38161695049837e-06,
+      "loss": 0.6132,
+      "step": 9015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.379547535586899e-06,
+      "loss": 0.6087,
+      "step": 9016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.37747824116279e-06,
+      "loss": 0.5924,
+      "step": 9017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.37540906732119e-06,
+      "loss": 0.6558,
+      "step": 9018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.373340014157236e-06,
+      "loss": 0.6632,
+      "step": 9019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.371271081766059e-06,
+      "loss": 0.6467,
+      "step": 9020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.369202270242793e-06,
+      "loss": 0.5913,
+      "step": 9021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.36713357968256e-06,
+      "loss": 0.7365,
+      "step": 9022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.365065010180476e-06,
+      "loss": 0.6576,
+      "step": 9023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.362996561831654e-06,
+      "loss": 0.6263,
+      "step": 9024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.360928234731199e-06,
+      "loss": 0.6456,
+      "step": 9025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.358860028974219e-06,
+      "loss": 0.6566,
+      "step": 9026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.356791944655799e-06,
+      "loss": 0.6022,
+      "step": 9027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.354723981871033e-06,
+      "loss": 0.6271,
+      "step": 9028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.352656140715006e-06,
+      "loss": 0.6329,
+      "step": 9029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.350588421282797e-06,
+      "loss": 0.6212,
+      "step": 9030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.348520823669476e-06,
+      "loss": 0.5936,
+      "step": 9031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.346453347970112e-06,
+      "loss": 0.6732,
+      "step": 9032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.34438599427977e-06,
+      "loss": 0.5763,
+      "step": 9033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.3423187626935e-06,
+      "loss": 0.6532,
+      "step": 9034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.340251653306355e-06,
+      "loss": 0.5794,
+      "step": 9035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.33818466621338e-06,
+      "loss": 0.65,
+      "step": 9036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.336117801509617e-06,
+      "loss": 0.5948,
+      "step": 9037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.334051059290094e-06,
+      "loss": 0.6233,
+      "step": 9038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.331984439649843e-06,
+      "loss": 0.6632,
+      "step": 9039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.329917942683885e-06,
+      "loss": 0.681,
+      "step": 9040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.32785156848724e-06,
+      "loss": 0.6841,
+      "step": 9041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.325785317154914e-06,
+      "loss": 0.6442,
+      "step": 9042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.323719188781915e-06,
+      "loss": 0.6878,
+      "step": 9043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.32165318346324e-06,
+      "loss": 0.7104,
+      "step": 9044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.31958730129389e-06,
+      "loss": 0.6493,
+      "step": 9045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.317521542368846e-06,
+      "loss": 0.6167,
+      "step": 9046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.315455906783093e-06,
+      "loss": 0.6122,
+      "step": 9047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.313390394631612e-06,
+      "loss": 0.6623,
+      "step": 9048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.311325006009369e-06,
+      "loss": 0.6504,
+      "step": 9049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.309259741011331e-06,
+      "loss": 0.6042,
+      "step": 9050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.307194599732459e-06,
+      "loss": 0.6732,
+      "step": 9051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.305129582267713e-06,
+      "loss": 0.5661,
+      "step": 9052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.30306468871203e-06,
+      "loss": 0.597,
+      "step": 9053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.30099991916036e-06,
+      "loss": 0.6966,
+      "step": 9054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.298935273707639e-06,
+      "loss": 0.6277,
+      "step": 9055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.296870752448802e-06,
+      "loss": 0.6276,
+      "step": 9056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.294806355478767e-06,
+      "loss": 0.6008,
+      "step": 9057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.292742082892462e-06,
+      "loss": 0.6423,
+      "step": 9058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.290677934784796e-06,
+      "loss": 0.6435,
+      "step": 9059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.288613911250684e-06,
+      "loss": 0.7081,
+      "step": 9060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.286550012385023e-06,
+      "loss": 0.5735,
+      "step": 9061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.28448623828271e-06,
+      "loss": 0.6559,
+      "step": 9062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.28242258903864e-06,
+      "loss": 0.6303,
+      "step": 9063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.280359064747701e-06,
+      "loss": 0.6285,
+      "step": 9064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.278295665504766e-06,
+      "loss": 0.6523,
+      "step": 9065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.276232391404715e-06,
+      "loss": 0.6813,
+      "step": 9066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.274169242542415e-06,
+      "loss": 0.6542,
+      "step": 9067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.272106219012727e-06,
+      "loss": 0.6108,
+      "step": 9068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.270043320910508e-06,
+      "loss": 0.6625,
+      "step": 9069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.267980548330609e-06,
+      "loss": 0.6743,
+      "step": 9070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.26591790136788e-06,
+      "loss": 0.6072,
+      "step": 9071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.263855380117155e-06,
+      "loss": 0.6886,
+      "step": 9072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.261792984673272e-06,
+      "loss": 0.6714,
+      "step": 9073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.259730715131057e-06,
+      "loss": 0.6235,
+      "step": 9074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.257668571585336e-06,
+      "loss": 0.6262,
+      "step": 9075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.255606554130919e-06,
+      "loss": 0.6354,
+      "step": 9076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.25354466286262e-06,
+      "loss": 0.753,
+      "step": 9077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.251482897875243e-06,
+      "loss": 0.6234,
+      "step": 9078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.249421259263592e-06,
+      "loss": 0.6147,
+      "step": 9079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.247359747122455e-06,
+      "loss": 0.6665,
+      "step": 9080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.24529836154662e-06,
+      "loss": 0.6758,
+      "step": 9081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.243237102630874e-06,
+      "loss": 0.6014,
+      "step": 9082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.241175970469985e-06,
+      "loss": 0.5686,
+      "step": 9083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.239114965158725e-06,
+      "loss": 0.6281,
+      "step": 9084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.237054086791862e-06,
+      "loss": 0.6704,
+      "step": 9085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.234993335464153e-06,
+      "loss": 0.6178,
+      "step": 9086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.232932711270349e-06,
+      "loss": 0.6981,
+      "step": 9087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.230872214305196e-06,
+      "loss": 0.6558,
+      "step": 9088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.228811844663438e-06,
+      "loss": 0.6313,
+      "step": 9089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.226751602439809e-06,
+      "loss": 0.6482,
+      "step": 9090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.224691487729037e-06,
+      "loss": 0.6422,
+      "step": 9091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.222631500625842e-06,
+      "loss": 0.6033,
+      "step": 9092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.220571641224948e-06,
+      "loss": 0.6585,
+      "step": 9093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.218511909621067e-06,
+      "loss": 0.598,
+      "step": 9094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.216452305908896e-06,
+      "loss": 0.6311,
+      "step": 9095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.214392830183142e-06,
+      "loss": 0.7206,
+      "step": 9096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.212333482538497e-06,
+      "loss": 0.6841,
+      "step": 9097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.210274263069647e-06,
+      "loss": 0.6458,
+      "step": 9098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.208215171871276e-06,
+      "loss": 0.6026,
+      "step": 9099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.20615620903806e-06,
+      "loss": 0.7368,
+      "step": 9100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.204097374664674e-06,
+      "loss": 0.597,
+      "step": 9101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.202038668845774e-06,
+      "loss": 0.6379,
+      "step": 9102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.199980091676021e-06,
+      "loss": 0.6415,
+      "step": 9103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.197921643250069e-06,
+      "loss": 0.5877,
+      "step": 9104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.195863323662568e-06,
+      "loss": 0.6946,
+      "step": 9105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.19380513300815e-06,
+      "loss": 0.5936,
+      "step": 9106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.191747071381458e-06,
+      "loss": 0.6969,
+      "step": 9107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.189689138877116e-06,
+      "loss": 0.6305,
+      "step": 9108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.187631335589752e-06,
+      "loss": 0.7223,
+      "step": 9109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.185573661613978e-06,
+      "loss": 0.6478,
+      "step": 9110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.183516117044407e-06,
+      "loss": 0.6072,
+      "step": 9111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.181458701975641e-06,
+      "loss": 0.5995,
+      "step": 9112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.179401416502287e-06,
+      "loss": 0.702,
+      "step": 9113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.177344260718928e-06,
+      "loss": 0.613,
+      "step": 9114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.175287234720158e-06,
+      "loss": 0.706,
+      "step": 9115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.173230338600561e-06,
+      "loss": 0.6744,
+      "step": 9116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.1711735724547e-06,
+      "loss": 0.6033,
+      "step": 9117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.169116936377158e-06,
+      "loss": 0.6179,
+      "step": 9118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.167060430462489e-06,
+      "loss": 0.6797,
+      "step": 9119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.165004054805256e-06,
+      "loss": 0.6065,
+      "step": 9120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.162947809500004e-06,
+      "loss": 0.6426,
+      "step": 9121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.160891694641286e-06,
+      "loss": 0.6988,
+      "step": 9122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.158835710323636e-06,
+      "loss": 0.6237,
+      "step": 9123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.156779856641593e-06,
+      "loss": 0.6278,
+      "step": 9124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.154724133689676e-06,
+      "loss": 0.6941,
+      "step": 9125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.152668541562413e-06,
+      "loss": 0.6911,
+      "step": 9126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.150613080354315e-06,
+      "loss": 0.668,
+      "step": 9127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.148557750159896e-06,
+      "loss": 0.6293,
+      "step": 9128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.146502551073655e-06,
+      "loss": 0.6222,
+      "step": 9129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.14444748319009e-06,
+      "loss": 0.6403,
+      "step": 9130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.142392546603695e-06,
+      "loss": 0.7027,
+      "step": 9131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.140337741408951e-06,
+      "loss": 0.6753,
+      "step": 9132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.13828306770034e-06,
+      "loss": 0.6433,
+      "step": 9133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.136228525572335e-06,
+      "loss": 0.6426,
+      "step": 9134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.134174115119406e-06,
+      "loss": 0.6353,
+      "step": 9135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.132119836436005e-06,
+      "loss": 0.5688,
+      "step": 9136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.130065689616592e-06,
+      "loss": 0.6875,
+      "step": 9137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.128011674755617e-06,
+      "loss": 0.6204,
+      "step": 9138
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.125957791947524e-06,
+      "loss": 0.7087,
+      "step": 9139
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.123904041286746e-06,
+      "loss": 0.6176,
+      "step": 9140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.121850422867716e-06,
+      "loss": 0.6577,
+      "step": 9141
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.119796936784856e-06,
+      "loss": 0.6701,
+      "step": 9142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.11774358313259e-06,
+      "loss": 0.7094,
+      "step": 9143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.115690362005325e-06,
+      "loss": 0.6849,
+      "step": 9144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.113637273497469e-06,
+      "loss": 0.6182,
+      "step": 9145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.111584317703421e-06,
+      "loss": 0.6701,
+      "step": 9146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.109531494717578e-06,
+      "loss": 0.6579,
+      "step": 9147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.107478804634324e-06,
+      "loss": 0.5434,
+      "step": 9148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.105426247548044e-06,
+      "loss": 0.6328,
+      "step": 9149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.103373823553117e-06,
+      "loss": 0.615,
+      "step": 9150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.101321532743903e-06,
+      "loss": 0.6428,
+      "step": 9151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.099269375214772e-06,
+      "loss": 0.651,
+      "step": 9152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.097217351060082e-06,
+      "loss": 0.6096,
+      "step": 9153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.095165460374182e-06,
+      "loss": 0.6545,
+      "step": 9154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.093113703251417e-06,
+      "loss": 0.7062,
+      "step": 9155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.091062079786129e-06,
+      "loss": 0.6897,
+      "step": 9156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.089010590072644e-06,
+      "loss": 0.6197,
+      "step": 9157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.086959234205302e-06,
+      "loss": 0.6173,
+      "step": 9158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.084908012278409e-06,
+      "loss": 0.6225,
+      "step": 9159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.082856924386283e-06,
+      "loss": 0.6771,
+      "step": 9160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.080805970623236e-06,
+      "loss": 0.64,
+      "step": 9161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.078755151083573e-06,
+      "loss": 0.6418,
+      "step": 9162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.076704465861578e-06,
+      "loss": 0.6774,
+      "step": 9163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.074653915051552e-06,
+      "loss": 0.6474,
+      "step": 9164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.072603498747777e-06,
+      "loss": 0.612,
+      "step": 9165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.070553217044524e-06,
+      "loss": 0.596,
+      "step": 9166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.068503070036068e-06,
+      "loss": 0.625,
+      "step": 9167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.066453057816675e-06,
+      "loss": 0.6891,
+      "step": 9168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.064403180480606e-06,
+      "loss": 0.6222,
+      "step": 9169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.062353438122105e-06,
+      "loss": 0.6978,
+      "step": 9170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.060303830835424e-06,
+      "loss": 0.6343,
+      "step": 9171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.058254358714804e-06,
+      "loss": 0.62,
+      "step": 9172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.056205021854478e-06,
+      "loss": 0.6866,
+      "step": 9173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.054155820348673e-06,
+      "loss": 0.7471,
+      "step": 9174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.052106754291609e-06,
+      "loss": 0.6105,
+      "step": 9175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.050057823777503e-06,
+      "loss": 0.6338,
+      "step": 9176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.04800902890057e-06,
+      "loss": 0.6363,
+      "step": 9177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.045960369755001e-06,
+      "loss": 0.6465,
+      "step": 9178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.043911846434998e-06,
+      "loss": 0.638,
+      "step": 9179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.041863459034752e-06,
+      "loss": 0.6386,
+      "step": 9180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.039815207648449e-06,
+      "loss": 0.6432,
+      "step": 9181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.037767092370262e-06,
+      "loss": 0.6114,
+      "step": 9182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.035719113294364e-06,
+      "loss": 0.5732,
+      "step": 9183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.033671270514927e-06,
+      "loss": 0.6663,
+      "step": 9184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.031623564126099e-06,
+      "loss": 0.6371,
+      "step": 9185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.02957599422204e-06,
+      "loss": 0.6362,
+      "step": 9186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.02752856089689e-06,
+      "loss": 0.7007,
+      "step": 9187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.0254812642448e-06,
+      "loss": 0.6409,
+      "step": 9188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.023434104359892e-06,
+      "loss": 0.6995,
+      "step": 9189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.021387081336302e-06,
+      "loss": 0.7113,
+      "step": 9190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.019340195268146e-06,
+      "loss": 0.6198,
+      "step": 9191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.017293446249547e-06,
+      "loss": 0.6683,
+      "step": 9192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.015246834374603e-06,
+      "loss": 0.6223,
+      "step": 9193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.01320035973742e-06,
+      "loss": 0.6325,
+      "step": 9194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.011154022432099e-06,
+      "loss": 0.6615,
+      "step": 9195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.009107822552725e-06,
+      "loss": 0.7106,
+      "step": 9196
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.007061760193382e-06,
+      "loss": 0.6406,
+      "step": 9197
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.005015835448147e-06,
+      "loss": 0.6575,
+      "step": 9198
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.002970048411097e-06,
+      "loss": 0.6277,
+      "step": 9199
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.000924399176287e-06,
+      "loss": 0.6533,
+      "step": 9200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.99887888783778e-06,
+      "loss": 0.6503,
+      "step": 9201
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.996833514489625e-06,
+      "loss": 0.6239,
+      "step": 9202
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.994788279225874e-06,
+      "loss": 0.6325,
+      "step": 9203
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.99274318214056e-06,
+      "loss": 0.6375,
+      "step": 9204
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.990698223327714e-06,
+      "loss": 0.6072,
+      "step": 9205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.98865340288137e-06,
+      "loss": 0.5441,
+      "step": 9206
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.986608720895542e-06,
+      "loss": 0.5588,
+      "step": 9207
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.984564177464243e-06,
+      "loss": 0.6151,
+      "step": 9208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.982519772681486e-06,
+      "loss": 0.676,
+      "step": 9209
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.980475506641266e-06,
+      "loss": 0.6046,
+      "step": 9210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.978431379437587e-06,
+      "loss": 0.5838,
+      "step": 9211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.976387391164423e-06,
+      "loss": 0.6482,
+      "step": 9212
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.974343541915766e-06,
+      "loss": 0.6431,
+      "step": 9213
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.972299831785588e-06,
+      "loss": 0.6211,
+      "step": 9214
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.970256260867858e-06,
+      "loss": 0.6196,
+      "step": 9215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.968212829256539e-06,
+      "loss": 0.6664,
+      "step": 9216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.966169537045588e-06,
+      "loss": 0.6453,
+      "step": 9217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.964126384328957e-06,
+      "loss": 0.5697,
+      "step": 9218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.962083371200583e-06,
+      "loss": 0.6571,
+      "step": 9219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.960040497754405e-06,
+      "loss": 0.6031,
+      "step": 9220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.957997764084357e-06,
+      "loss": 0.5867,
+      "step": 9221
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.95595517028436e-06,
+      "loss": 0.6613,
+      "step": 9222
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.953912716448333e-06,
+      "loss": 0.6126,
+      "step": 9223
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.951870402670186e-06,
+      "loss": 0.739,
+      "step": 9224
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.949828229043823e-06,
+      "loss": 0.6484,
+      "step": 9225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.94778619566315e-06,
+      "loss": 0.687,
+      "step": 9226
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.945744302622048e-06,
+      "loss": 0.6269,
+      "step": 9227
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.943702550014406e-06,
+      "loss": 0.6876,
+      "step": 9228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.941660937934104e-06,
+      "loss": 0.6792,
+      "step": 9229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.939619466475019e-06,
+      "loss": 0.6621,
+      "step": 9230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.93757813573101e-06,
+      "loss": 0.6658,
+      "step": 9231
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.935536945795938e-06,
+      "loss": 0.6967,
+      "step": 9232
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.933495896763662e-06,
+      "loss": 0.6539,
+      "step": 9233
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.931454988728022e-06,
+      "loss": 0.6927,
+      "step": 9234
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.929414221782858e-06,
+      "loss": 0.6046,
+      "step": 9235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.927373596022004e-06,
+      "loss": 0.6202,
+      "step": 9236
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.925333111539297e-06,
+      "loss": 0.6265,
+      "step": 9237
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.923292768428542e-06,
+      "loss": 0.6162,
+      "step": 9238
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.921252566783563e-06,
+      "loss": 0.6561,
+      "step": 9239
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.919212506698162e-06,
+      "loss": 0.6498,
+      "step": 9240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.91717258826615e-06,
+      "loss": 0.6324,
+      "step": 9241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.915132811581311e-06,
+      "loss": 0.6191,
+      "step": 9242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.913093176737438e-06,
+      "loss": 0.7131,
+      "step": 9243
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.911053683828311e-06,
+      "loss": 0.6419,
+      "step": 9244
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.909014332947712e-06,
+      "loss": 0.726,
+      "step": 9245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.906975124189397e-06,
+      "loss": 0.6012,
+      "step": 9246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.904936057647136e-06,
+      "loss": 0.664,
+      "step": 9247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.902897133414685e-06,
+      "loss": 0.6494,
+      "step": 9248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.900858351585788e-06,
+      "loss": 0.6996,
+      "step": 9249
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.898819712254192e-06,
+      "loss": 0.6851,
+      "step": 9250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.89678121551363e-06,
+      "loss": 0.6513,
+      "step": 9251
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.894742861457837e-06,
+      "loss": 0.5713,
+      "step": 9252
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.892704650180527e-06,
+      "loss": 0.6706,
+      "step": 9253
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.890666581775418e-06,
+      "loss": 0.625,
+      "step": 9254
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.888628656336225e-06,
+      "loss": 0.6494,
+      "step": 9255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.886590873956648e-06,
+      "loss": 0.612,
+      "step": 9256
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.884553234730383e-06,
+      "loss": 0.6596,
+      "step": 9257
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.882515738751118e-06,
+      "loss": 0.6325,
+      "step": 9258
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.88047838611254e-06,
+      "loss": 0.6893,
+      "step": 9259
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.878441176908327e-06,
+      "loss": 0.6976,
+      "step": 9260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.876404111232143e-06,
+      "loss": 0.6407,
+      "step": 9261
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.874367189177654e-06,
+      "loss": 0.727,
+      "step": 9262
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.872330410838515e-06,
+      "loss": 0.6458,
+      "step": 9263
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.870293776308384e-06,
+      "loss": 0.6763,
+      "step": 9264
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.868257285680898e-06,
+      "loss": 0.6341,
+      "step": 9265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.866220939049692e-06,
+      "loss": 0.6646,
+      "step": 9266
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.864184736508406e-06,
+      "loss": 0.6405,
+      "step": 9267
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.862148678150653e-06,
+      "loss": 0.6473,
+      "step": 9268
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.860112764070054e-06,
+      "loss": 0.6619,
+      "step": 9269
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.85807699436022e-06,
+      "loss": 0.6739,
+      "step": 9270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.85604136911476e-06,
+      "loss": 0.6411,
+      "step": 9271
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.854005888427262e-06,
+      "loss": 0.6029,
+      "step": 9272
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.85197055239132e-06,
+      "loss": 0.641,
+      "step": 9273
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.849935361100522e-06,
+      "loss": 0.61,
+      "step": 9274
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.847900314648444e-06,
+      "loss": 0.646,
+      "step": 9275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.84586541312865e-06,
+      "loss": 0.6123,
+      "step": 9276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.843830656634708e-06,
+      "loss": 0.6065,
+      "step": 9277
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.841796045260179e-06,
+      "loss": 0.6568,
+      "step": 9278
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.839761579098614e-06,
+      "loss": 0.6029,
+      "step": 9279
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.83772725824355e-06,
+      "loss": 0.5508,
+      "step": 9280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.835693082788526e-06,
+      "loss": 0.6487,
+      "step": 9281
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.833659052827078e-06,
+      "loss": 0.612,
+      "step": 9282
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.831625168452722e-06,
+      "loss": 0.6478,
+      "step": 9283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.82959142975898e-06,
+      "loss": 0.5816,
+      "step": 9284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.827557836839364e-06,
+      "loss": 0.6097,
+      "step": 9285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.82552438978738e-06,
+      "loss": 0.7151,
+      "step": 9286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.823491088696516e-06,
+      "loss": 0.6412,
+      "step": 9287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.821457933660266e-06,
+      "loss": 0.6215,
+      "step": 9288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.819424924772116e-06,
+      "loss": 0.6134,
+      "step": 9289
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.817392062125545e-06,
+      "loss": 0.5238,
+      "step": 9290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.815359345814016e-06,
+      "loss": 0.6622,
+      "step": 9291
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.813326775930998e-06,
+      "loss": 0.6237,
+      "step": 9292
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.811294352569946e-06,
+      "loss": 0.6373,
+      "step": 9293
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.809262075824314e-06,
+      "loss": 0.6487,
+      "step": 9294
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.80722994578754e-06,
+      "loss": 0.6045,
+      "step": 9295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.80519796255306e-06,
+      "loss": 0.5991,
+      "step": 9296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.803166126214304e-06,
+      "loss": 0.6267,
+      "step": 9297
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.8011344368647e-06,
+      "loss": 0.5847,
+      "step": 9298
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.79910289459766e-06,
+      "loss": 0.6282,
+      "step": 9299
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.797071499506596e-06,
+      "loss": 0.6184,
+      "step": 9300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.795040251684913e-06,
+      "loss": 0.6319,
+      "step": 9301
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.793009151225998e-06,
+      "loss": 0.6431,
+      "step": 9302
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.790978198223246e-06,
+      "loss": 0.6079,
+      "step": 9303
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.788947392770038e-06,
+      "loss": 0.6534,
+      "step": 9304
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.786916734959754e-06,
+      "loss": 0.6817,
+      "step": 9305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.784886224885758e-06,
+      "loss": 0.6309,
+      "step": 9306
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.782855862641414e-06,
+      "loss": 0.7101,
+      "step": 9307
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.780825648320076e-06,
+      "loss": 0.6692,
+      "step": 9308
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.778795582015096e-06,
+      "loss": 0.5682,
+      "step": 9309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.77676566381981e-06,
+      "loss": 0.6841,
+      "step": 9310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.774735893827555e-06,
+      "loss": 0.6703,
+      "step": 9311
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.77270627213166e-06,
+      "loss": 0.5993,
+      "step": 9312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.770676798825451e-06,
+      "loss": 0.618,
+      "step": 9313
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.768647474002232e-06,
+      "loss": 0.6663,
+      "step": 9314
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.766618297755316e-06,
+      "loss": 0.6704,
+      "step": 9315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.764589270178007e-06,
+      "loss": 0.6193,
+      "step": 9316
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.762560391363591e-06,
+      "loss": 0.6936,
+      "step": 9317
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.760531661405362e-06,
+      "loss": 0.6415,
+      "step": 9318
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.758503080396597e-06,
+      "loss": 0.6625,
+      "step": 9319
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.756474648430573e-06,
+      "loss": 0.6613,
+      "step": 9320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.754446365600548e-06,
+      "loss": 0.6468,
+      "step": 9321
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.75241823199979e-06,
+      "loss": 0.6208,
+      "step": 9322
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.75039024772155e-06,
+      "loss": 0.6545,
+      "step": 9323
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.748362412859071e-06,
+      "loss": 0.6021,
+      "step": 9324
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.746334727505595e-06,
+      "loss": 0.6503,
+      "step": 9325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.74430719175435e-06,
+      "loss": 0.6666,
+      "step": 9326
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.742279805698566e-06,
+      "loss": 0.6487,
+      "step": 9327
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.740252569431465e-06,
+      "loss": 0.6527,
+      "step": 9328
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.738225483046248e-06,
+      "loss": 0.6541,
+      "step": 9329
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.736198546636125e-06,
+      "loss": 0.5906,
+      "step": 9330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.734171760294294e-06,
+      "loss": 0.7379,
+      "step": 9331
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.732145124113948e-06,
+      "loss": 0.6005,
+      "step": 9332
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.730118638188264e-06,
+      "loss": 0.6882,
+      "step": 9333
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.728092302610425e-06,
+      "loss": 0.5969,
+      "step": 9334
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.726066117473603e-06,
+      "loss": 0.6647,
+      "step": 9335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.724040082870953e-06,
+      "loss": 0.6528,
+      "step": 9336
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.722014198895637e-06,
+      "loss": 0.7279,
+      "step": 9337
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.719988465640803e-06,
+      "loss": 0.6684,
+      "step": 9338
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.717962883199595e-06,
+      "loss": 0.6745,
+      "step": 9339
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.715937451665146e-06,
+      "loss": 0.6037,
+      "step": 9340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.713912171130587e-06,
+      "loss": 0.5691,
+      "step": 9341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.711887041689035e-06,
+      "loss": 0.682,
+      "step": 9342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.709862063433615e-06,
+      "loss": 0.6887,
+      "step": 9343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.707837236457422e-06,
+      "loss": 0.6309,
+      "step": 9344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.705812560853562e-06,
+      "loss": 0.634,
+      "step": 9345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.703788036715131e-06,
+      "loss": 0.6451,
+      "step": 9346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.701763664135218e-06,
+      "loss": 0.6461,
+      "step": 9347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.699739443206892e-06,
+      "loss": 0.6501,
+      "step": 9348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.697715374023233e-06,
+      "loss": 0.6383,
+      "step": 9349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.695691456677311e-06,
+      "loss": 0.6353,
+      "step": 9350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.693667691262178e-06,
+      "loss": 0.6006,
+      "step": 9351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.691644077870887e-06,
+      "loss": 0.6754,
+      "step": 9352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.689620616596486e-06,
+      "loss": 0.5679,
+      "step": 9353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.687597307532013e-06,
+      "loss": 0.6977,
+      "step": 9354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.685574150770495e-06,
+      "loss": 0.6466,
+      "step": 9355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.683551146404958e-06,
+      "loss": 0.6565,
+      "step": 9356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.681528294528419e-06,
+      "loss": 0.676,
+      "step": 9357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.679505595233888e-06,
+      "loss": 0.6359,
+      "step": 9358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.677483048614368e-06,
+      "loss": 0.6927,
+      "step": 9359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.675460654762856e-06,
+      "loss": 0.6205,
+      "step": 9360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.673438413772337e-06,
+      "loss": 0.6285,
+      "step": 9361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.671416325735802e-06,
+      "loss": 0.6549,
+      "step": 9362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.669394390746213e-06,
+      "loss": 0.6633,
+      "step": 9363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.667372608896545e-06,
+      "loss": 0.6388,
+      "step": 9364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.66535098027976e-06,
+      "loss": 0.6286,
+      "step": 9365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.663329504988808e-06,
+      "loss": 0.6743,
+      "step": 9366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.661308183116637e-06,
+      "loss": 0.6257,
+      "step": 9367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.659287014756186e-06,
+      "loss": 0.6052,
+      "step": 9368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.657266000000393e-06,
+      "loss": 0.6524,
+      "step": 9369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.655245138942175e-06,
+      "loss": 0.6533,
+      "step": 9370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.653224431674452e-06,
+      "loss": 0.6743,
+      "step": 9371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.65120387829014e-06,
+      "loss": 0.6606,
+      "step": 9372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.649183478882139e-06,
+      "loss": 0.6367,
+      "step": 9373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.64716323354335e-06,
+      "loss": 0.599,
+      "step": 9374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.645143142366659e-06,
+      "loss": 0.6746,
+      "step": 9375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.64312320544495e-06,
+      "loss": 0.5926,
+      "step": 9376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.641103422871105e-06,
+      "loss": 0.6268,
+      "step": 9377
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.639083794737982e-06,
+      "loss": 0.6387,
+      "step": 9378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.63706432113845e-06,
+      "loss": 0.6288,
+      "step": 9379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.635045002165361e-06,
+      "loss": 0.6418,
+      "step": 9380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.633025837911565e-06,
+      "loss": 0.6279,
+      "step": 9381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.631006828469899e-06,
+      "loss": 0.6592,
+      "step": 9382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.628987973933198e-06,
+      "loss": 0.6181,
+      "step": 9383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.626969274394293e-06,
+      "loss": 0.6869,
+      "step": 9384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.624950729945993e-06,
+      "loss": 0.7713,
+      "step": 9385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.622932340681114e-06,
+      "loss": 0.6121,
+      "step": 9386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.620914106692464e-06,
+      "loss": 0.6206,
+      "step": 9387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.61889602807284e-06,
+      "loss": 0.6392,
+      "step": 9388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.616878104915027e-06,
+      "loss": 0.625,
+      "step": 9389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.614860337311815e-06,
+      "loss": 0.6423,
+      "step": 9390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.612842725355975e-06,
+      "loss": 0.6828,
+      "step": 9391
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.61082526914028e-06,
+      "loss": 0.6031,
+      "step": 9392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.608807968757486e-06,
+      "loss": 0.6443,
+      "step": 9393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.606790824300355e-06,
+      "loss": 0.6814,
+      "step": 9394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.604773835861628e-06,
+      "loss": 0.6236,
+      "step": 9395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.602757003534056e-06,
+      "loss": 0.6245,
+      "step": 9396
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.600740327410359e-06,
+      "loss": 0.6484,
+      "step": 9397
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.598723807583268e-06,
+      "loss": 0.5535,
+      "step": 9398
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.596707444145505e-06,
+      "loss": 0.6469,
+      "step": 9399
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.594691237189776e-06,
+      "loss": 0.6498,
+      "step": 9400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.59267518680879e-06,
+      "loss": 0.575,
+      "step": 9401
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.590659293095241e-06,
+      "loss": 0.606,
+      "step": 9402
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.588643556141827e-06,
+      "loss": 0.5976,
+      "step": 9403
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.586627976041217e-06,
+      "loss": 0.6221,
+      "step": 9404
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.584612552886096e-06,
+      "loss": 0.6286,
+      "step": 9405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.58259728676913e-06,
+      "loss": 0.6753,
+      "step": 9406
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.580582177782983e-06,
+      "loss": 0.597,
+      "step": 9407
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.578567226020302e-06,
+      "loss": 0.5541,
+      "step": 9408
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.576552431573741e-06,
+      "loss": 0.6205,
+      "step": 9409
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.574537794535934e-06,
+      "loss": 0.6747,
+      "step": 9410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.572523314999521e-06,
+      "loss": 0.655,
+      "step": 9411
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.570508993057117e-06,
+      "loss": 0.5929,
+      "step": 9412
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.568494828801344e-06,
+      "loss": 0.6997,
+      "step": 9413
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.566480822324814e-06,
+      "loss": 0.6722,
+      "step": 9414
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.564466973720129e-06,
+      "loss": 0.5898,
+      "step": 9415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.562453283079886e-06,
+      "loss": 0.6314,
+      "step": 9416
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.56043975049667e-06,
+      "loss": 0.635,
+      "step": 9417
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.55842637606307e-06,
+      "loss": 0.659,
+      "step": 9418
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.55641315987165e-06,
+      "loss": 0.5481,
+      "step": 9419
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.554400102014984e-06,
+      "loss": 0.6552,
+      "step": 9420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.55238720258563e-06,
+      "loss": 0.6578,
+      "step": 9421
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.550374461676144e-06,
+      "loss": 0.6677,
+      "step": 9422
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.54836187937906e-06,
+      "loss": 0.6114,
+      "step": 9423
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.546349455786925e-06,
+      "loss": 0.666,
+      "step": 9424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.544337190992267e-06,
+      "loss": 0.6158,
+      "step": 9425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.542325085087612e-06,
+      "loss": 0.6034,
+      "step": 9426
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.54031313816547e-06,
+      "loss": 0.6092,
+      "step": 9427
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.538301350318355e-06,
+      "loss": 0.5774,
+      "step": 9428
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.536289721638764e-06,
+      "loss": 0.6942,
+      "step": 9429
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.534278252219197e-06,
+      "loss": 0.653,
+      "step": 9430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.532266942152132e-06,
+      "loss": 0.6376,
+      "step": 9431
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.530255791530053e-06,
+      "loss": 0.5989,
+      "step": 9432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.528244800445433e-06,
+      "loss": 0.6334,
+      "step": 9433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.526233968990733e-06,
+      "loss": 0.588,
+      "step": 9434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.524223297258413e-06,
+      "loss": 0.597,
+      "step": 9435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.522212785340923e-06,
+      "loss": 0.699,
+      "step": 9436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.520202433330706e-06,
+      "loss": 0.6503,
+      "step": 9437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.518192241320196e-06,
+      "loss": 0.6271,
+      "step": 9438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.516182209401817e-06,
+      "loss": 0.5739,
+      "step": 9439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.514172337667996e-06,
+      "loss": 0.5952,
+      "step": 9440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.512162626211142e-06,
+      "loss": 0.6526,
+      "step": 9441
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.510153075123662e-06,
+      "loss": 0.5701,
+      "step": 9442
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.508143684497957e-06,
+      "loss": 0.6526,
+      "step": 9443
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.506134454426411e-06,
+      "loss": 0.696,
+      "step": 9444
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.504125385001418e-06,
+      "loss": 0.6487,
+      "step": 9445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.502116476315344e-06,
+      "loss": 0.6305,
+      "step": 9446
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.500107728460562e-06,
+      "loss": 0.6604,
+      "step": 9447
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.498099141529432e-06,
+      "loss": 0.6642,
+      "step": 9448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.496090715614314e-06,
+      "loss": 0.6654,
+      "step": 9449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.494082450807546e-06,
+      "loss": 0.5842,
+      "step": 9450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.492074347201473e-06,
+      "loss": 0.6684,
+      "step": 9451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.490066404888429e-06,
+      "loss": 0.6554,
+      "step": 9452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.48805862396073e-06,
+      "loss": 0.6473,
+      "step": 9453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.486051004510695e-06,
+      "loss": 0.582,
+      "step": 9454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.48404354663064e-06,
+      "loss": 0.6137,
+      "step": 9455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.482036250412864e-06,
+      "loss": 0.6108,
+      "step": 9456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.480029115949657e-06,
+      "loss": 0.6726,
+      "step": 9457
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.47802214333331e-06,
+      "loss": 0.6788,
+      "step": 9458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.476015332656106e-06,
+      "loss": 0.6413,
+      "step": 9459
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.474008684010317e-06,
+      "loss": 0.5949,
+      "step": 9460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.4720021974882e-06,
+      "loss": 0.6374,
+      "step": 9461
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.469995873182018e-06,
+      "loss": 0.5636,
+      "step": 9462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.46798971118402e-06,
+      "loss": 0.6053,
+      "step": 9463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.465983711586455e-06,
+      "loss": 0.6383,
+      "step": 9464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.463977874481546e-06,
+      "loss": 0.5842,
+      "step": 9465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.461972199961527e-06,
+      "loss": 0.6032,
+      "step": 9466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.459966688118622e-06,
+      "loss": 0.6526,
+      "step": 9467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.457961339045035e-06,
+      "loss": 0.6667,
+      "step": 9468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.455956152832977e-06,
+      "loss": 0.6172,
+      "step": 9469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.453951129574643e-06,
+      "loss": 0.5751,
+      "step": 9470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.45194626936223e-06,
+      "loss": 0.5792,
+      "step": 9471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.44994157228791e-06,
+      "loss": 0.7075,
+      "step": 9472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.447937038443865e-06,
+      "loss": 0.6178,
+      "step": 9473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.44593266792226e-06,
+      "loss": 0.6325,
+      "step": 9474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.443928460815257e-06,
+      "loss": 0.5713,
+      "step": 9475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.441924417215006e-06,
+      "loss": 0.7002,
+      "step": 9476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.439920537213655e-06,
+      "loss": 0.6582,
+      "step": 9477
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.437916820903341e-06,
+      "loss": 0.6341,
+      "step": 9478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.435913268376196e-06,
+      "loss": 0.638,
+      "step": 9479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.433909879724338e-06,
+      "loss": 0.5945,
+      "step": 9480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.431906655039883e-06,
+      "loss": 0.574,
+      "step": 9481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.42990359441494e-06,
+      "loss": 0.6229,
+      "step": 9482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.42790069794161e-06,
+      "loss": 0.5714,
+      "step": 9483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.425897965711982e-06,
+      "loss": 0.6243,
+      "step": 9484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.423895397818145e-06,
+      "loss": 0.6425,
+      "step": 9485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.421892994352176e-06,
+      "loss": 0.646,
+      "step": 9486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.419890755406139e-06,
+      "loss": 0.6623,
+      "step": 9487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.4178886810721e-06,
+      "loss": 0.6907,
+      "step": 9488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.415886771442113e-06,
+      "loss": 0.6275,
+      "step": 9489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.413885026608229e-06,
+      "loss": 0.6653,
+      "step": 9490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.41188344666248e-06,
+      "loss": 0.6452,
+      "step": 9491
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.409882031696903e-06,
+      "loss": 0.6226,
+      "step": 9492
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.407880781803521e-06,
+      "loss": 0.6457,
+      "step": 9493
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.405879697074354e-06,
+      "loss": 0.6335,
+      "step": 9494
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.403878777601405e-06,
+      "loss": 0.639,
+      "step": 9495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.401878023476677e-06,
+      "loss": 0.632,
+      "step": 9496
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.399877434792165e-06,
+      "loss": 0.6614,
+      "step": 9497
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.39787701163986e-06,
+      "loss": 0.6948,
+      "step": 9498
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.395876754111734e-06,
+      "loss": 0.6808,
+      "step": 9499
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.393876662299758e-06,
+      "loss": 0.6161,
+      "step": 9500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.3918767362959e-06,
+      "loss": 0.623,
+      "step": 9501
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.38987697619211e-06,
+      "loss": 0.6479,
+      "step": 9502
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.387877382080344e-06,
+      "loss": 0.6352,
+      "step": 9503
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.385877954052532e-06,
+      "loss": 0.661,
+      "step": 9504
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.383878692200623e-06,
+      "loss": 0.6361,
+      "step": 9505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.381879596616524e-06,
+      "loss": 0.6237,
+      "step": 9506
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.379880667392161e-06,
+      "loss": 0.6501,
+      "step": 9507
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.377881904619446e-06,
+      "loss": 0.6629,
+      "step": 9508
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.37588330839028e-06,
+      "loss": 0.7077,
+      "step": 9509
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.373884878796555e-06,
+      "loss": 0.5844,
+      "step": 9510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.37188661593016e-06,
+      "loss": 0.6573,
+      "step": 9511
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.369888519882975e-06,
+      "loss": 0.6188,
+      "step": 9512
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.367890590746874e-06,
+      "loss": 0.663,
+      "step": 9513
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.365892828613713e-06,
+      "loss": 0.5629,
+      "step": 9514
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.363895233575355e-06,
+      "loss": 0.5965,
+      "step": 9515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.361897805723644e-06,
+      "loss": 0.6375,
+      "step": 9516
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.359900545150428e-06,
+      "loss": 0.6099,
+      "step": 9517
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.357903451947533e-06,
+      "loss": 0.5992,
+      "step": 9518
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.355906526206788e-06,
+      "loss": 0.6305,
+      "step": 9519
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.353909768020012e-06,
+      "loss": 0.6606,
+      "step": 9520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.351913177479013e-06,
+      "loss": 0.602,
+      "step": 9521
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.349916754675592e-06,
+      "loss": 0.7002,
+      "step": 9522
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.347920499701546e-06,
+      "loss": 0.6696,
+      "step": 9523
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.345924412648663e-06,
+      "loss": 0.6034,
+      "step": 9524
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.343928493608719e-06,
+      "loss": 0.6627,
+      "step": 9525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.341932742673487e-06,
+      "loss": 0.6756,
+      "step": 9526
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.339937159934732e-06,
+      "loss": 0.6798,
+      "step": 9527
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.337941745484214e-06,
+      "loss": 0.6753,
+      "step": 9528
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.335946499413673e-06,
+      "loss": 0.5935,
+      "step": 9529
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.333951421814852e-06,
+      "loss": 0.6725,
+      "step": 9530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.331956512779486e-06,
+      "loss": 0.6339,
+      "step": 9531
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.329961772399304e-06,
+      "loss": 0.7033,
+      "step": 9532
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.327967200766014e-06,
+      "loss": 0.6144,
+      "step": 9533
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.325972797971332e-06,
+      "loss": 0.5829,
+      "step": 9534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.323978564106958e-06,
+      "loss": 0.6556,
+      "step": 9535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.321984499264586e-06,
+      "loss": 0.659,
+      "step": 9536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.3199906035359e-06,
+      "loss": 0.7292,
+      "step": 9537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.317996877012584e-06,
+      "loss": 0.5713,
+      "step": 9538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.31600331978631e-06,
+      "loss": 0.652,
+      "step": 9539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.31400993194873e-06,
+      "loss": 0.6246,
+      "step": 9540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.312016713591509e-06,
+      "loss": 0.6546,
+      "step": 9541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.31002366480629e-06,
+      "loss": 0.5565,
+      "step": 9542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.308030785684715e-06,
+      "loss": 0.6378,
+      "step": 9543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.306038076318414e-06,
+      "loss": 0.6337,
+      "step": 9544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.304045536799013e-06,
+      "loss": 0.6174,
+      "step": 9545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.302053167218125e-06,
+      "loss": 0.6351,
+      "step": 9546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.300060967667365e-06,
+      "loss": 0.6216,
+      "step": 9547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.298068938238325e-06,
+      "loss": 0.5719,
+      "step": 9548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.296077079022603e-06,
+      "loss": 0.6802,
+      "step": 9549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.29408539011178e-06,
+      "loss": 0.634,
+      "step": 9550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.292093871597437e-06,
+      "loss": 0.6527,
+      "step": 9551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.290102523571142e-06,
+      "loss": 0.6875,
+      "step": 9552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.288111346124455e-06,
+      "loss": 0.6489,
+      "step": 9553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.286120339348936e-06,
+      "loss": 0.6024,
+      "step": 9554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.28412950333612e-06,
+      "loss": 0.6411,
+      "step": 9555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.282138838177552e-06,
+      "loss": 0.6432,
+      "step": 9556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.280148343964757e-06,
+      "loss": 0.5827,
+      "step": 9557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.278158020789267e-06,
+      "loss": 0.629,
+      "step": 9558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.276167868742585e-06,
+      "loss": 0.6255,
+      "step": 9559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.274177887916222e-06,
+      "loss": 0.6318,
+      "step": 9560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.272188078401675e-06,
+      "loss": 0.6363,
+      "step": 9561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.270198440290442e-06,
+      "loss": 0.7009,
+      "step": 9562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.268208973673997e-06,
+      "loss": 0.6596,
+      "step": 9563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.266219678643816e-06,
+      "loss": 0.6062,
+      "step": 9564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.26423055529137e-06,
+      "loss": 0.5741,
+      "step": 9565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.262241603708118e-06,
+      "loss": 0.6196,
+      "step": 9566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.260252823985504e-06,
+      "loss": 0.6223,
+      "step": 9567
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.258264216214977e-06,
+      "loss": 0.5916,
+      "step": 9568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.256275780487978e-06,
+      "loss": 0.6509,
+      "step": 9569
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.254287516895924e-06,
+      "loss": 0.6684,
+      "step": 9570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.252299425530238e-06,
+      "loss": 0.625,
+      "step": 9571
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.250311506482334e-06,
+      "loss": 0.5531,
+      "step": 9572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.248323759843616e-06,
+      "loss": 0.6712,
+      "step": 9573
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.246336185705475e-06,
+      "loss": 0.5816,
+      "step": 9574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.244348784159302e-06,
+      "loss": 0.6361,
+      "step": 9575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.242361555296478e-06,
+      "loss": 0.7108,
+      "step": 9576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.240374499208375e-06,
+      "loss": 0.649,
+      "step": 9577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.238387615986354e-06,
+      "loss": 0.6638,
+      "step": 9578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.236400905721773e-06,
+      "loss": 0.6141,
+      "step": 9579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.23441436850598e-06,
+      "loss": 0.6401,
+      "step": 9580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.23242800443032e-06,
+      "loss": 0.5826,
+      "step": 9581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.230441813586117e-06,
+      "loss": 0.6684,
+      "step": 9582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.228455796064696e-06,
+      "loss": 0.6814,
+      "step": 9583
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.226469951957382e-06,
+      "loss": 0.6724,
+      "step": 9584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.224484281355473e-06,
+      "loss": 0.6687,
+      "step": 9585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.222498784350276e-06,
+      "loss": 0.6949,
+      "step": 9586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.220513461033078e-06,
+      "loss": 0.6215,
+      "step": 9587
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.218528311495171e-06,
+      "loss": 0.6156,
+      "step": 9588
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.216543335827824e-06,
+      "loss": 0.5981,
+      "step": 9589
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.21455853412231e-06,
+      "loss": 0.5335,
+      "step": 9590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.212573906469886e-06,
+      "loss": 0.7261,
+      "step": 9591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.210589452961808e-06,
+      "loss": 0.6057,
+      "step": 9592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.208605173689317e-06,
+      "loss": 0.585,
+      "step": 9593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.20662106874365e-06,
+      "loss": 0.6096,
+      "step": 9594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.204637138216037e-06,
+      "loss": 0.6543,
+      "step": 9595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.202653382197702e-06,
+      "loss": 0.6601,
+      "step": 9596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.200669800779849e-06,
+      "loss": 0.6174,
+      "step": 9597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.198686394053686e-06,
+      "loss": 0.6296,
+      "step": 9598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.196703162110407e-06,
+      "loss": 0.6911,
+      "step": 9599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.194720105041207e-06,
+      "loss": 0.5973,
+      "step": 9600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.192737222937259e-06,
+      "loss": 0.6417,
+      "step": 9601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.190754515889739e-06,
+      "loss": 0.6797,
+      "step": 9602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.188771983989812e-06,
+      "loss": 0.6607,
+      "step": 9603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.186789627328631e-06,
+      "loss": 0.6785,
+      "step": 9604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.184807445997342e-06,
+      "loss": 0.6271,
+      "step": 9605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.18282544008709e-06,
+      "loss": 0.6435,
+      "step": 9606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.180843609689009e-06,
+      "loss": 0.6286,
+      "step": 9607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.178861954894213e-06,
+      "loss": 0.6636,
+      "step": 9608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.176880475793822e-06,
+      "loss": 0.6812,
+      "step": 9609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.174899172478948e-06,
+      "loss": 0.6552,
+      "step": 9610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.172918045040687e-06,
+      "loss": 0.6894,
+      "step": 9611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.17093709357013e-06,
+      "loss": 0.6648,
+      "step": 9612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.16895631815836e-06,
+      "loss": 0.6191,
+      "step": 9613
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.166975718896456e-06,
+      "loss": 0.656,
+      "step": 9614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.164995295875485e-06,
+      "loss": 0.6666,
+      "step": 9615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.163015049186501e-06,
+      "loss": 0.6532,
+      "step": 9616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.161034978920554e-06,
+      "loss": 0.6138,
+      "step": 9617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.159055085168695e-06,
+      "loss": 0.6639,
+      "step": 9618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.157075368021953e-06,
+      "loss": 0.6538,
+      "step": 9619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.155095827571356e-06,
+      "loss": 0.6115,
+      "step": 9620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.15311646390792e-06,
+      "loss": 0.6161,
+      "step": 9621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.151137277122663e-06,
+      "loss": 0.6167,
+      "step": 9622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.149158267306578e-06,
+      "loss": 0.6462,
+      "step": 9623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.147179434550663e-06,
+      "loss": 0.6494,
+      "step": 9624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.145200778945902e-06,
+      "loss": 0.5943,
+      "step": 9625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.14322230058328e-06,
+      "loss": 0.6179,
+      "step": 9626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.141243999553757e-06,
+      "loss": 0.6543,
+      "step": 9627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.139265875948298e-06,
+      "loss": 0.6285,
+      "step": 9628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.13728792985786e-06,
+      "loss": 0.6629,
+      "step": 9629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.135310161373389e-06,
+      "loss": 0.6744,
+      "step": 9630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.133332570585813e-06,
+      "loss": 0.6609,
+      "step": 9631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.131355157586067e-06,
+      "loss": 0.6221,
+      "step": 9632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.12937792246507e-06,
+      "loss": 0.6344,
+      "step": 9633
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.127400865313737e-06,
+      "loss": 0.7217,
+      "step": 9634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.125423986222971e-06,
+      "loss": 0.6502,
+      "step": 9635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.123447285283666e-06,
+      "loss": 0.6241,
+      "step": 9636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.121470762586715e-06,
+      "loss": 0.6451,
+      "step": 9637
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.119494418222991e-06,
+      "loss": 0.6638,
+      "step": 9638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.11751825228337e-06,
+      "loss": 0.6056,
+      "step": 9639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.115542264858714e-06,
+      "loss": 0.6503,
+      "step": 9640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.113566456039879e-06,
+      "loss": 0.7243,
+      "step": 9641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.11159082591771e-06,
+      "loss": 0.6151,
+      "step": 9642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.109615374583047e-06,
+      "loss": 0.6607,
+      "step": 9643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.107640102126721e-06,
+      "loss": 0.6376,
+      "step": 9644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.105665008639557e-06,
+      "loss": 0.6434,
+      "step": 9645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.103690094212362e-06,
+      "loss": 0.6719,
+      "step": 9646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.101715358935946e-06,
+      "loss": 0.6868,
+      "step": 9647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.099740802901103e-06,
+      "loss": 0.5859,
+      "step": 9648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.097766426198633e-06,
+      "loss": 0.6331,
+      "step": 9649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.095792228919303e-06,
+      "loss": 0.6918,
+      "step": 9650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.093818211153895e-06,
+      "loss": 0.6798,
+      "step": 9651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.09184437299317e-06,
+      "loss": 0.6185,
+      "step": 9652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.089870714527883e-06,
+      "loss": 0.6471,
+      "step": 9653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.087897235848785e-06,
+      "loss": 0.6275,
+      "step": 9654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.085923937046616e-06,
+      "loss": 0.7002,
+      "step": 9655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.083950818212108e-06,
+      "loss": 0.5693,
+      "step": 9656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.081977879435981e-06,
+      "loss": 0.5479,
+      "step": 9657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.08000512080895e-06,
+      "loss": 0.6351,
+      "step": 9658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.078032542421723e-06,
+      "loss": 0.7296,
+      "step": 9659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.076060144364998e-06,
+      "loss": 0.6457,
+      "step": 9660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.074087926729468e-06,
+      "loss": 0.6324,
+      "step": 9661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.07211588960581e-06,
+      "loss": 0.6014,
+      "step": 9662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.0701440330847e-06,
+      "loss": 0.6624,
+      "step": 9663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.068172357256807e-06,
+      "loss": 0.6121,
+      "step": 9664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.066200862212781e-06,
+      "loss": 0.6984,
+      "step": 9665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.064229548043272e-06,
+      "loss": 0.6657,
+      "step": 9666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.062258414838922e-06,
+      "loss": 0.5953,
+      "step": 9667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.060287462690364e-06,
+      "loss": 0.6357,
+      "step": 9668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.05831669168822e-06,
+      "loss": 0.6181,
+      "step": 9669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.056346101923106e-06,
+      "loss": 0.6579,
+      "step": 9670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.05437569348563e-06,
+      "loss": 0.5704,
+      "step": 9671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.052405466466385e-06,
+      "loss": 0.6316,
+      "step": 9672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.050435420955967e-06,
+      "loss": 0.6063,
+      "step": 9673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.048465557044955e-06,
+      "loss": 0.6141,
+      "step": 9674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.046495874823925e-06,
+      "loss": 0.6297,
+      "step": 9675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.04452637438344e-06,
+      "loss": 0.6129,
+      "step": 9676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.042557055814058e-06,
+      "loss": 0.6626,
+      "step": 9677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.040587919206326e-06,
+      "loss": 0.591,
+      "step": 9678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.03861896465079e-06,
+      "loss": 0.6541,
+      "step": 9679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.036650192237973e-06,
+      "loss": 0.6122,
+      "step": 9680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.034681602058402e-06,
+      "loss": 0.5947,
+      "step": 9681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.032713194202594e-06,
+      "loss": 0.6261,
+      "step": 9682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.030744968761058e-06,
+      "loss": 0.6249,
+      "step": 9683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.028776925824282e-06,
+      "loss": 0.6448,
+      "step": 9684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.026809065482764e-06,
+      "loss": 0.5912,
+      "step": 9685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.024841387826985e-06,
+      "loss": 0.6125,
+      "step": 9686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.022873892947414e-06,
+      "loss": 0.6227,
+      "step": 9687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.02090658093452e-06,
+      "loss": 0.6114,
+      "step": 9688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.018939451878756e-06,
+      "loss": 0.6876,
+      "step": 9689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.016972505870575e-06,
+      "loss": 0.5812,
+      "step": 9690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.01500574300041e-06,
+      "loss": 0.7376,
+      "step": 9691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.013039163358694e-06,
+      "loss": 0.6288,
+      "step": 9692
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.011072767035851e-06,
+      "loss": 0.6162,
+      "step": 9693
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.009106554122295e-06,
+      "loss": 0.6416,
+      "step": 9694
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.007140524708428e-06,
+      "loss": 0.6546,
+      "step": 9695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.005174678884652e-06,
+      "loss": 0.6279,
+      "step": 9696
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.003209016741353e-06,
+      "loss": 0.6026,
+      "step": 9697
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.001243538368917e-06,
+      "loss": 0.5808,
+      "step": 9698
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.999278243857709e-06,
+      "loss": 0.6245,
+      "step": 9699
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.997313133298093e-06,
+      "loss": 0.6095,
+      "step": 9700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.995348206780425e-06,
+      "loss": 0.5827,
+      "step": 9701
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.993383464395055e-06,
+      "loss": 0.662,
+      "step": 9702
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.991418906232316e-06,
+      "loss": 0.5955,
+      "step": 9703
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.989454532382541e-06,
+      "loss": 0.5414,
+      "step": 9704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.987490342936053e-06,
+      "loss": 0.6906,
+      "step": 9705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.985526337983159e-06,
+      "loss": 0.6089,
+      "step": 9706
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.983562517614165e-06,
+      "loss": 0.6272,
+      "step": 9707
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.981598881919367e-06,
+      "loss": 0.5961,
+      "step": 9708
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.979635430989057e-06,
+      "loss": 0.6968,
+      "step": 9709
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.977672164913507e-06,
+      "loss": 0.6605,
+      "step": 9710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.975709083782989e-06,
+      "loss": 0.6126,
+      "step": 9711
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.973746187687765e-06,
+      "loss": 0.5877,
+      "step": 9712
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.971783476718094e-06,
+      "loss": 0.6032,
+      "step": 9713
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.96982095096421e-06,
+      "loss": 0.6177,
+      "step": 9714
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.967858610516353e-06,
+      "loss": 0.6061,
+      "step": 9715
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.965896455464754e-06,
+      "loss": 0.6501,
+      "step": 9716
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.963934485899633e-06,
+      "loss": 0.656,
+      "step": 9717
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.961972701911195e-06,
+      "loss": 0.5852,
+      "step": 9718
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.960011103589644e-06,
+      "loss": 0.6434,
+      "step": 9719
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.958049691025175e-06,
+      "loss": 0.5741,
+      "step": 9720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.956088464307973e-06,
+      "loss": 0.6637,
+      "step": 9721
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.954127423528211e-06,
+      "loss": 0.6365,
+      "step": 9722
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.952166568776061e-06,
+      "loss": 0.5836,
+      "step": 9723
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.950205900141685e-06,
+      "loss": 0.6939,
+      "step": 9724
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.948245417715226e-06,
+      "loss": 0.6796,
+      "step": 9725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.946285121586827e-06,
+      "loss": 0.6544,
+      "step": 9726
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.944325011846628e-06,
+      "loss": 0.6292,
+      "step": 9727
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.942365088584752e-06,
+      "loss": 0.6947,
+      "step": 9728
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.94040535189131e-06,
+      "loss": 0.6369,
+      "step": 9729
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.938445801856414e-06,
+      "loss": 0.5217,
+      "step": 9730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.936486438570164e-06,
+      "loss": 0.613,
+      "step": 9731
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.934527262122654e-06,
+      "loss": 0.6443,
+      "step": 9732
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.93256827260396e-06,
+      "loss": 0.6082,
+      "step": 9733
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.930609470104154e-06,
+      "loss": 0.6643,
+      "step": 9734
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.928650854713309e-06,
+      "loss": 0.6619,
+      "step": 9735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.926692426521473e-06,
+      "loss": 0.6451,
+      "step": 9736
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.924734185618702e-06,
+      "loss": 0.6534,
+      "step": 9737
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.922776132095028e-06,
+      "loss": 0.6512,
+      "step": 9738
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.920818266040489e-06,
+      "loss": 0.6647,
+      "step": 9739
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.918860587545097e-06,
+      "loss": 0.6101,
+      "step": 9740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.916903096698872e-06,
+      "loss": 0.6144,
+      "step": 9741
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.914945793591817e-06,
+      "loss": 0.6405,
+      "step": 9742
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.912988678313929e-06,
+      "loss": 0.6982,
+      "step": 9743
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.911031750955193e-06,
+      "loss": 0.7144,
+      "step": 9744
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.909075011605592e-06,
+      "loss": 0.6274,
+      "step": 9745
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.90711846035509e-06,
+      "loss": 0.5763,
+      "step": 9746
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.905162097293658e-06,
+      "loss": 0.5424,
+      "step": 9747
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.90320592251124e-06,
+      "loss": 0.6372,
+      "step": 9748
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.90124993609778e-06,
+      "loss": 0.7058,
+      "step": 9749
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.899294138143216e-06,
+      "loss": 0.6729,
+      "step": 9750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.89733852873748e-06,
+      "loss": 0.6396,
+      "step": 9751
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.895383107970483e-06,
+      "loss": 0.6327,
+      "step": 9752
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.893427875932134e-06,
+      "loss": 0.6556,
+      "step": 9753
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.891472832712343e-06,
+      "loss": 0.6615,
+      "step": 9754
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.88951797840099e-06,
+      "loss": 0.6465,
+      "step": 9755
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.887563313087966e-06,
+      "loss": 0.617,
+      "step": 9756
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.885608836863142e-06,
+      "loss": 0.6662,
+      "step": 9757
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.88365454981639e-06,
+      "loss": 0.5898,
+      "step": 9758
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.881700452037561e-06,
+      "loss": 0.6165,
+      "step": 9759
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.879746543616505e-06,
+      "loss": 0.5679,
+      "step": 9760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.877792824643062e-06,
+      "loss": 0.573,
+      "step": 9761
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.875839295207066e-06,
+      "loss": 0.6691,
+      "step": 9762
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.873885955398338e-06,
+      "loss": 0.5999,
+      "step": 9763
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.871932805306689e-06,
+      "loss": 0.6137,
+      "step": 9764
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.869979845021925e-06,
+      "loss": 0.6557,
+      "step": 9765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.86802707463385e-06,
+      "loss": 0.653,
+      "step": 9766
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.86607449423224e-06,
+      "loss": 0.598,
+      "step": 9767
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.86412210390688e-06,
+      "loss": 0.6205,
+      "step": 9768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.86216990374754e-06,
+      "loss": 0.6417,
+      "step": 9769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.86021789384398e-06,
+      "loss": 0.6008,
+      "step": 9770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.858266074285954e-06,
+      "loss": 0.6499,
+      "step": 9771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.856314445163205e-06,
+      "loss": 0.642,
+      "step": 9772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.85436300656547e-06,
+      "loss": 0.5589,
+      "step": 9773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.852411758582472e-06,
+      "loss": 0.5826,
+      "step": 9774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.85046070130393e-06,
+      "loss": 0.5976,
+      "step": 9775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.848509834819553e-06,
+      "loss": 0.6274,
+      "step": 9776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.846559159219043e-06,
+      "loss": 0.6889,
+      "step": 9777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.844608674592087e-06,
+      "loss": 0.6336,
+      "step": 9778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.84265838102837e-06,
+      "loss": 0.6897,
+      "step": 9779
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.840708278617566e-06,
+      "loss": 0.6342,
+      "step": 9780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.838758367449344e-06,
+      "loss": 0.6306,
+      "step": 9781
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.83680864761335e-06,
+      "loss": 0.7063,
+      "step": 9782
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.834859119199238e-06,
+      "loss": 0.6454,
+      "step": 9783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.832909782296647e-06,
+      "loss": 0.614,
+      "step": 9784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.83096063699521e-06,
+      "loss": 0.6067,
+      "step": 9785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.829011683384536e-06,
+      "loss": 0.5785,
+      "step": 9786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.827062921554248e-06,
+      "loss": 0.6866,
+      "step": 9787
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.825114351593948e-06,
+      "loss": 0.6484,
+      "step": 9788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.823165973593225e-06,
+      "loss": 0.5658,
+      "step": 9789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.821217787641666e-06,
+      "loss": 0.6245,
+      "step": 9790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.819269793828853e-06,
+      "loss": 0.6217,
+      "step": 9791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.817321992244352e-06,
+      "loss": 0.6311,
+      "step": 9792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.815374382977716e-06,
+      "loss": 0.6681,
+      "step": 9793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.813426966118504e-06,
+      "loss": 0.6575,
+      "step": 9794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.81147974175625e-06,
+      "loss": 0.6256,
+      "step": 9795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.809532709980493e-06,
+      "loss": 0.6559,
+      "step": 9796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.807585870880754e-06,
+      "loss": 0.6513,
+      "step": 9797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.805639224546546e-06,
+      "loss": 0.6152,
+      "step": 9798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.803692771067378e-06,
+      "loss": 0.5974,
+      "step": 9799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.801746510532744e-06,
+      "loss": 0.6696,
+      "step": 9800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.799800443032136e-06,
+      "loss": 0.6318,
+      "step": 9801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.797854568655028e-06,
+      "loss": 0.6689,
+      "step": 9802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.795908887490901e-06,
+      "loss": 0.5946,
+      "step": 9803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.793963399629207e-06,
+      "loss": 0.6464,
+      "step": 9804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.7920181051594e-06,
+      "loss": 0.6504,
+      "step": 9805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.790073004170925e-06,
+      "loss": 0.6496,
+      "step": 9806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.788128096753222e-06,
+      "loss": 0.6804,
+      "step": 9807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.786183382995709e-06,
+      "loss": 0.6293,
+      "step": 9808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.784238862987807e-06,
+      "loss": 0.6194,
+      "step": 9809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.782294536818921e-06,
+      "loss": 0.6407,
+      "step": 9810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.780350404578461e-06,
+      "loss": 0.6108,
+      "step": 9811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.778406466355803e-06,
+      "loss": 0.6066,
+      "step": 9812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.776462722240337e-06,
+      "loss": 0.59,
+      "step": 9813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.774519172321433e-06,
+      "loss": 0.6187,
+      "step": 9814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.772575816688455e-06,
+      "loss": 0.633,
+      "step": 9815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.77063265543076e-06,
+      "loss": 0.6382,
+      "step": 9816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.76868968863769e-06,
+      "loss": 0.6412,
+      "step": 9817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.766746916398583e-06,
+      "loss": 0.616,
+      "step": 9818
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.764804338802774e-06,
+      "loss": 0.6064,
+      "step": 9819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.76286195593957e-06,
+      "loss": 0.5822,
+      "step": 9820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.760919767898287e-06,
+      "loss": 0.5348,
+      "step": 9821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.75897777476823e-06,
+      "loss": 0.5951,
+      "step": 9822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.757035976638681e-06,
+      "loss": 0.6988,
+      "step": 9823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.755094373598931e-06,
+      "loss": 0.6319,
+      "step": 9824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.753152965738251e-06,
+      "loss": 0.6384,
+      "step": 9825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.751211753145911e-06,
+      "loss": 0.6715,
+      "step": 9826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.74927073591116e-06,
+      "loss": 0.6139,
+      "step": 9827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.747329914123248e-06,
+      "loss": 0.6015,
+      "step": 9828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.745389287871412e-06,
+      "loss": 0.5936,
+      "step": 9829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.743448857244889e-06,
+      "loss": 0.5834,
+      "step": 9830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.741508622332888e-06,
+      "loss": 0.634,
+      "step": 9831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.739568583224626e-06,
+      "loss": 0.5869,
+      "step": 9832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.737628740009305e-06,
+      "loss": 0.562,
+      "step": 9833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.735689092776118e-06,
+      "loss": 0.6524,
+      "step": 9834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.733749641614248e-06,
+      "loss": 0.5847,
+      "step": 9835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.731810386612871e-06,
+      "loss": 0.6435,
+      "step": 9836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.729871327861159e-06,
+      "loss": 0.6832,
+      "step": 9837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.727932465448259e-06,
+      "loss": 0.5485,
+      "step": 9838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.725993799463322e-06,
+      "loss": 0.5745,
+      "step": 9839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.724055329995491e-06,
+      "loss": 0.5363,
+      "step": 9840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.722117057133895e-06,
+      "loss": 0.6132,
+      "step": 9841
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.720178980967653e-06,
+      "loss": 0.5928,
+      "step": 9842
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.718241101585875e-06,
+      "loss": 0.6071,
+      "step": 9843
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.71630341907767e-06,
+      "loss": 0.6451,
+      "step": 9844
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.71436593353213e-06,
+      "loss": 0.6583,
+      "step": 9845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.712428645038333e-06,
+      "loss": 0.6673,
+      "step": 9846
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.710491553685362e-06,
+      "loss": 0.6171,
+      "step": 9847
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.708554659562283e-06,
+      "loss": 0.5894,
+      "step": 9848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.706617962758151e-06,
+      "loss": 0.5847,
+      "step": 9849
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.704681463362015e-06,
+      "loss": 0.6539,
+      "step": 9850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.702745161462918e-06,
+      "loss": 0.6334,
+      "step": 9851
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.700809057149888e-06,
+      "loss": 0.6653,
+      "step": 9852
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.698873150511947e-06,
+      "loss": 0.6299,
+      "step": 9853
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.696937441638105e-06,
+      "loss": 0.6654,
+      "step": 9854
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.695001930617365e-06,
+      "loss": 0.6364,
+      "step": 9855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.693066617538728e-06,
+      "loss": 0.6194,
+      "step": 9856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.691131502491169e-06,
+      "loss": 0.6657,
+      "step": 9857
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.689196585563668e-06,
+      "loss": 0.6029,
+      "step": 9858
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.687261866845192e-06,
+      "loss": 0.6661,
+      "step": 9859
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.685327346424703e-06,
+      "loss": 0.6646,
+      "step": 9860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.68339302439114e-06,
+      "loss": 0.5749,
+      "step": 9861
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.681458900833447e-06,
+      "loss": 0.5861,
+      "step": 9862
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.679524975840557e-06,
+      "loss": 0.6625,
+      "step": 9863
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.677591249501391e-06,
+      "loss": 0.6575,
+      "step": 9864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.675657721904855e-06,
+      "loss": 0.6505,
+      "step": 9865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.673724393139856e-06,
+      "loss": 0.5978,
+      "step": 9866
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.671791263295283e-06,
+      "loss": 0.6629,
+      "step": 9867
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.669858332460033e-06,
+      "loss": 0.6427,
+      "step": 9868
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.66792560072297e-06,
+      "loss": 0.6722,
+      "step": 9869
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.66599306817296e-06,
+      "loss": 0.6498,
+      "step": 9870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.66406073489887e-06,
+      "loss": 0.6304,
+      "step": 9871
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.662128600989538e-06,
+      "loss": 0.6442,
+      "step": 9872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.660196666533806e-06,
+      "loss": 0.6546,
+      "step": 9873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.658264931620502e-06,
+      "loss": 0.654,
+      "step": 9874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.656333396338454e-06,
+      "loss": 0.5997,
+      "step": 9875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.654402060776463e-06,
+      "loss": 0.6384,
+      "step": 9876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.652470925023334e-06,
+      "loss": 0.6704,
+      "step": 9877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.650539989167863e-06,
+      "loss": 0.6088,
+      "step": 9878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.648609253298834e-06,
+      "loss": 0.6041,
+      "step": 9879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.646678717505017e-06,
+      "loss": 0.5747,
+      "step": 9880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.64474838187518e-06,
+      "loss": 0.6228,
+      "step": 9881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.642818246498077e-06,
+      "loss": 0.5737,
+      "step": 9882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.640888311462458e-06,
+      "loss": 0.6026,
+      "step": 9883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.638958576857058e-06,
+      "loss": 0.6092,
+      "step": 9884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.637029042770608e-06,
+      "loss": 0.6428,
+      "step": 9885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.635099709291829e-06,
+      "loss": 0.6,
+      "step": 9886
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.633170576509424e-06,
+      "loss": 0.6213,
+      "step": 9887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.631241644512097e-06,
+      "loss": 0.6225,
+      "step": 9888
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.6293129133885405e-06,
+      "loss": 0.6408,
+      "step": 9889
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.62738438322744e-06,
+      "loss": 0.6217,
+      "step": 9890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.625456054117462e-06,
+      "loss": 0.6435,
+      "step": 9891
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.623527926147273e-06,
+      "loss": 0.652,
+      "step": 9892
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.621599999405527e-06,
+      "loss": 0.6196,
+      "step": 9893
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.619672273980873e-06,
+      "loss": 0.6693,
+      "step": 9894
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.6177447499619435e-06,
+      "loss": 0.6245,
+      "step": 9895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.615817427437363e-06,
+      "loss": 0.6605,
+      "step": 9896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.613890306495754e-06,
+      "loss": 0.5845,
+      "step": 9897
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.611963387225721e-06,
+      "loss": 0.5929,
+      "step": 9898
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.610036669715866e-06,
+      "loss": 0.6405,
+      "step": 9899
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.608110154054778e-06,
+      "loss": 0.6034,
+      "step": 9900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.606183840331035e-06,
+      "loss": 0.635,
+      "step": 9901
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.604257728633215e-06,
+      "loss": 0.6731,
+      "step": 9902
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.60233181904987e-06,
+      "loss": 0.7529,
+      "step": 9903
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.600406111669558e-06,
+      "loss": 0.6663,
+      "step": 9904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.598480606580824e-06,
+      "loss": 0.6235,
+      "step": 9905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5965553038722e-06,
+      "loss": 0.6698,
+      "step": 9906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.594630203632206e-06,
+      "loss": 0.7003,
+      "step": 9907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.592705305949364e-06,
+      "loss": 0.6488,
+      "step": 9908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.590780610912182e-06,
+      "loss": 0.699,
+      "step": 9909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.588856118609145e-06,
+      "loss": 0.7025,
+      "step": 9910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.586931829128749e-06,
+      "loss": 0.6148,
+      "step": 9911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.585007742559472e-06,
+      "loss": 0.621,
+      "step": 9912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.583083858989785e-06,
+      "loss": 0.5403,
+      "step": 9913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.58116017850814e-06,
+      "loss": 0.6851,
+      "step": 9914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.57923670120299e-06,
+      "loss": 0.6972,
+      "step": 9915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.577313427162778e-06,
+      "loss": 0.7087,
+      "step": 9916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.575390356475933e-06,
+      "loss": 0.7018,
+      "step": 9917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.57346748923088e-06,
+      "loss": 0.6339,
+      "step": 9918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.571544825516029e-06,
+      "loss": 0.6572,
+      "step": 9919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.569622365419786e-06,
+      "loss": 0.5732,
+      "step": 9920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.567700109030542e-06,
+      "loss": 0.6279,
+      "step": 9921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5657780564366816e-06,
+      "loss": 0.645,
+      "step": 9922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.56385620772658e-06,
+      "loss": 0.6127,
+      "step": 9923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.56193456298861e-06,
+      "loss": 0.5862,
+      "step": 9924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5600131223111165e-06,
+      "loss": 0.6426,
+      "step": 9925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5580918857824516e-06,
+      "loss": 0.6941,
+      "step": 9926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.556170853490954e-06,
+      "loss": 0.6765,
+      "step": 9927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.554250025524956e-06,
+      "loss": 0.6458,
+      "step": 9928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.55232940197277e-06,
+      "loss": 0.5964,
+      "step": 9929
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5504089829227055e-06,
+      "loss": 0.5926,
+      "step": 9930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.548488768463065e-06,
+      "loss": 0.5892,
+      "step": 9931
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.546568758682138e-06,
+      "loss": 0.6127,
+      "step": 9932
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.544648953668207e-06,
+      "loss": 0.5401,
+      "step": 9933
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.542729353509545e-06,
+      "loss": 0.5642,
+      "step": 9934
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.540809958294413e-06,
+      "loss": 0.6194,
+      "step": 9935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.538890768111067e-06,
+      "loss": 0.602,
+      "step": 9936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5369717830477445e-06,
+      "loss": 0.6346,
+      "step": 9937
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.535053003192682e-06,
+      "loss": 0.6166,
+      "step": 9938
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.53313442863411e-06,
+      "loss": 0.7051,
+      "step": 9939
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.531216059460234e-06,
+      "loss": 0.6267,
+      "step": 9940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.529297895759264e-06,
+      "loss": 0.7081,
+      "step": 9941
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5273799376193984e-06,
+      "loss": 0.6336,
+      "step": 9942
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.525462185128826e-06,
+      "loss": 0.7042,
+      "step": 9943
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.523544638375717e-06,
+      "loss": 0.6077,
+      "step": 9944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.521627297448243e-06,
+      "loss": 0.6511,
+      "step": 9945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.519710162434566e-06,
+      "loss": 0.6158,
+      "step": 9946
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.517793233422834e-06,
+      "loss": 0.6477,
+      "step": 9947
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.515876510501182e-06,
+      "loss": 0.5871,
+      "step": 9948
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.513959993757742e-06,
+      "loss": 0.6233,
+      "step": 9949
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.512043683280636e-06,
+      "loss": 0.6726,
+      "step": 9950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.510127579157975e-06,
+      "loss": 0.6688,
+      "step": 9951
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.508211681477861e-06,
+      "loss": 0.6332,
+      "step": 9952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.506295990328385e-06,
+      "loss": 0.6454,
+      "step": 9953
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.504380505797636e-06,
+      "loss": 0.6219,
+      "step": 9954
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.502465227973677e-06,
+      "loss": 0.6468,
+      "step": 9955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.500550156944575e-06,
+      "loss": 0.5881,
+      "step": 9956
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.498635292798389e-06,
+      "loss": 0.639,
+      "step": 9957
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4967206356231614e-06,
+      "loss": 0.616,
+      "step": 9958
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4948061855069235e-06,
+      "loss": 0.666,
+      "step": 9959
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.492891942537704e-06,
+      "loss": 0.6939,
+      "step": 9960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.490977906803519e-06,
+      "loss": 0.6133,
+      "step": 9961
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.489064078392378e-06,
+      "loss": 0.6731,
+      "step": 9962
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.487150457392272e-06,
+      "loss": 0.6447,
+      "step": 9963
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.485237043891189e-06,
+      "loss": 0.7101,
+      "step": 9964
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.483323837977113e-06,
+      "loss": 0.5875,
+      "step": 9965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.481410839738006e-06,
+      "loss": 0.6429,
+      "step": 9966
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.479498049261832e-06,
+      "loss": 0.5881,
+      "step": 9967
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.477585466636537e-06,
+      "loss": 0.58,
+      "step": 9968
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.475673091950063e-06,
+      "loss": 0.5648,
+      "step": 9969
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.473760925290343e-06,
+      "loss": 0.6095,
+      "step": 9970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.471848966745289e-06,
+      "loss": 0.6043,
+      "step": 9971
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.469937216402819e-06,
+      "loss": 0.5784,
+      "step": 9972
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.468025674350834e-06,
+      "loss": 0.5965,
+      "step": 9973
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.466114340677222e-06,
+      "loss": 0.5844,
+      "step": 9974
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.464203215469868e-06,
+      "loss": 0.603,
+      "step": 9975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4622922988166434e-06,
+      "loss": 0.6432,
+      "step": 9976
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.460381590805417e-06,
+      "loss": 0.6509,
+      "step": 9977
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.458471091524033e-06,
+      "loss": 0.5742,
+      "step": 9978
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.456560801060342e-06,
+      "loss": 0.6042,
+      "step": 9979
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4546507195021775e-06,
+      "loss": 0.6968,
+      "step": 9980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.452740846937365e-06,
+      "loss": 0.7065,
+      "step": 9981
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.450831183453716e-06,
+      "loss": 0.6461,
+      "step": 9982
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.448921729139039e-06,
+      "loss": 0.6153,
+      "step": 9983
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.447012484081126e-06,
+      "loss": 0.6836,
+      "step": 9984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.445103448367771e-06,
+      "loss": 0.6651,
+      "step": 9985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.443194622086743e-06,
+      "loss": 0.6261,
+      "step": 9986
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.441286005325812e-06,
+      "loss": 0.6023,
+      "step": 9987
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.439377598172742e-06,
+      "loss": 0.6514,
+      "step": 9988
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.43746940071527e-06,
+      "loss": 0.6039,
+      "step": 9989
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.435561413041136e-06,
+      "loss": 0.6208,
+      "step": 9990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.433653635238076e-06,
+      "loss": 0.6427,
+      "step": 9991
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.431746067393805e-06,
+      "loss": 0.6958,
+      "step": 9992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4298387095960276e-06,
+      "loss": 0.7229,
+      "step": 9993
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.427931561932447e-06,
+      "loss": 0.707,
+      "step": 9994
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.426024624490755e-06,
+      "loss": 0.5878,
+      "step": 9995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.424117897358631e-06,
+      "loss": 0.6353,
+      "step": 9996
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.422211380623743e-06,
+      "loss": 0.677,
+      "step": 9997
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.420305074373753e-06,
+      "loss": 0.5691,
+      "step": 9998
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.418398978696312e-06,
+      "loss": 0.635,
+      "step": 9999
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.416493093679062e-06,
+      "loss": 0.6275,
+      "step": 10000
+    },
+    {
+      "epoch": 0.66,
+      "eval_text_loss": 0.6529978513717651,
+      "eval_text_runtime": 1.8507,
+      "eval_text_samples_per_second": 39.984,
+      "eval_text_steps_per_second": 1.081,
+      "step": 10000
+    },
+    {
+      "epoch": 0.66,
+      "eval_svit_loss": 0.7801365852355957,
+      "eval_svit_runtime": 6.6261,
+      "eval_svit_samples_per_second": 123.149,
+      "eval_svit_steps_per_second": 1.962,
+      "step": 10000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.414587419409636e-06,
+      "loss": 0.6101,
+      "step": 10001
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.412681955975654e-06,
+      "loss": 0.6044,
+      "step": 10002
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.41077670346473e-06,
+      "loss": 0.6483,
+      "step": 10003
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4088716619644714e-06,
+      "loss": 0.61,
+      "step": 10004
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.40696683156246e-06,
+      "loss": 0.607,
+      "step": 10005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4050622123462884e-06,
+      "loss": 0.6517,
+      "step": 10006
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4031578044035286e-06,
+      "loss": 0.5399,
+      "step": 10007
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4012536078217386e-06,
+      "loss": 0.5889,
+      "step": 10008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.39934962268848e-06,
+      "loss": 0.6409,
+      "step": 10009
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.397445849091291e-06,
+      "loss": 0.6665,
+      "step": 10010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.395542287117717e-06,
+      "loss": 0.5957,
+      "step": 10011
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.393638936855269e-06,
+      "loss": 0.6618,
+      "step": 10012
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.39173579839147e-06,
+      "loss": 0.6435,
+      "step": 10013
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.389832871813825e-06,
+      "loss": 0.6552,
+      "step": 10014
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.387930157209835e-06,
+      "loss": 0.5648,
+      "step": 10015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.386027654666975e-06,
+      "loss": 0.6355,
+      "step": 10016
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.384125364272725e-06,
+      "loss": 0.6032,
+      "step": 10017
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.382223286114555e-06,
+      "loss": 0.6763,
+      "step": 10018
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.380321420279919e-06,
+      "loss": 0.6699,
+      "step": 10019
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.378419766856267e-06,
+      "loss": 0.5963,
+      "step": 10020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.376518325931033e-06,
+      "loss": 0.6459,
+      "step": 10021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.37461709759165e-06,
+      "loss": 0.6052,
+      "step": 10022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.372716081925527e-06,
+      "loss": 0.5571,
+      "step": 10023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.370815279020078e-06,
+      "loss": 0.611,
+      "step": 10024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.368914688962699e-06,
+      "loss": 0.6228,
+      "step": 10025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.367014311840784e-06,
+      "loss": 0.6359,
+      "step": 10026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.365114147741701e-06,
+      "loss": 0.6463,
+      "step": 10027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.363214196752825e-06,
+      "loss": 0.5896,
+      "step": 10028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.361314458961514e-06,
+      "loss": 0.6608,
+      "step": 10029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.359414934455122e-06,
+      "loss": 0.568,
+      "step": 10030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.357515623320981e-06,
+      "loss": 0.6063,
+      "step": 10031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3556165256464226e-06,
+      "loss": 0.7004,
+      "step": 10032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.35371764151877e-06,
+      "loss": 0.5983,
+      "step": 10033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.351818971025327e-06,
+      "loss": 0.6824,
+      "step": 10034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3499205142533994e-06,
+      "loss": 0.7256,
+      "step": 10035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3480222712902754e-06,
+      "loss": 0.5865,
+      "step": 10036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.34612424222324e-06,
+      "loss": 0.578,
+      "step": 10037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.344226427139553e-06,
+      "loss": 0.6531,
+      "step": 10038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.342328826126484e-06,
+      "loss": 0.6847,
+      "step": 10039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.340431439271281e-06,
+      "loss": 0.5746,
+      "step": 10040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.33853426666119e-06,
+      "loss": 0.6028,
+      "step": 10041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.336637308383432e-06,
+      "loss": 0.6301,
+      "step": 10042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.334740564525235e-06,
+      "loss": 0.6207,
+      "step": 10043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.332844035173811e-06,
+      "loss": 0.6782,
+      "step": 10044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.330947720416363e-06,
+      "loss": 0.5921,
+      "step": 10045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.329051620340078e-06,
+      "loss": 0.65,
+      "step": 10046
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.327155735032139e-06,
+      "loss": 0.6077,
+      "step": 10047
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.325260064579718e-06,
+      "loss": 0.6597,
+      "step": 10048
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.323364609069984e-06,
+      "loss": 0.6523,
+      "step": 10049
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.3214693685900795e-06,
+      "loss": 0.6071,
+      "step": 10050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.31957434322715e-06,
+      "loss": 0.6258,
+      "step": 10051
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.317679533068329e-06,
+      "loss": 0.6037,
+      "step": 10052
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.315784938200741e-06,
+      "loss": 0.6801,
+      "step": 10053
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.313890558711494e-06,
+      "loss": 0.6809,
+      "step": 10054
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.311996394687697e-06,
+      "loss": 0.7064,
+      "step": 10055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.31010244621644e-06,
+      "loss": 0.6481,
+      "step": 10056
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.3082087133848066e-06,
+      "loss": 0.6634,
+      "step": 10057
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.306315196279864e-06,
+      "loss": 0.7274,
+      "step": 10058
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.304421894988683e-06,
+      "loss": 0.5845,
+      "step": 10059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.3025288095983176e-06,
+      "loss": 0.677,
+      "step": 10060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.300635940195802e-06,
+      "loss": 0.6078,
+      "step": 10061
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.298743286868178e-06,
+      "loss": 0.5693,
+      "step": 10062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.296850849702464e-06,
+      "loss": 0.7088,
+      "step": 10063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.29495862878568e-06,
+      "loss": 0.7024,
+      "step": 10064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.293066624204822e-06,
+      "loss": 0.6281,
+      "step": 10065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.291174836046885e-06,
+      "loss": 0.6161,
+      "step": 10066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.289283264398858e-06,
+      "loss": 0.5039,
+      "step": 10067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2873919093477105e-06,
+      "loss": 0.579,
+      "step": 10068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.285500770980405e-06,
+      "loss": 0.5967,
+      "step": 10069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.283609849383898e-06,
+      "loss": 0.6074,
+      "step": 10070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.281719144645137e-06,
+      "loss": 0.6323,
+      "step": 10071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.279828656851048e-06,
+      "loss": 0.6457,
+      "step": 10072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2779383860885585e-06,
+      "loss": 0.6268,
+      "step": 10073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.276048332444581e-06,
+      "loss": 0.7102,
+      "step": 10074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.274158496006025e-06,
+      "loss": 0.6258,
+      "step": 10075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2722688768597765e-06,
+      "loss": 0.5523,
+      "step": 10076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.270379475092723e-06,
+      "loss": 0.6424,
+      "step": 10077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2684902907917396e-06,
+      "loss": 0.623,
+      "step": 10078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.266601324043693e-06,
+      "loss": 0.6575,
+      "step": 10079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.264712574935427e-06,
+      "loss": 0.6582,
+      "step": 10080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.262824043553794e-06,
+      "loss": 0.6742,
+      "step": 10081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.260935729985626e-06,
+      "loss": 0.5659,
+      "step": 10082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.259047634317746e-06,
+      "loss": 0.6171,
+      "step": 10083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.257159756636966e-06,
+      "loss": 0.659,
+      "step": 10084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.255272097030096e-06,
+      "loss": 0.6135,
+      "step": 10085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.253384655583925e-06,
+      "loss": 0.7414,
+      "step": 10086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.251497432385242e-06,
+      "loss": 0.5654,
+      "step": 10087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.249610427520813e-06,
+      "loss": 0.6453,
+      "step": 10088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.247723641077406e-06,
+      "loss": 0.6712,
+      "step": 10089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.245837073141777e-06,
+      "loss": 0.6272,
+      "step": 10090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2439507238006635e-06,
+      "loss": 0.6199,
+      "step": 10091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2420645931408026e-06,
+      "loss": 0.6111,
+      "step": 10092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.240178681248918e-06,
+      "loss": 0.6102,
+      "step": 10093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.238292988211726e-06,
+      "loss": 0.7208,
+      "step": 10094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.236407514115924e-06,
+      "loss": 0.6114,
+      "step": 10095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.234522259048207e-06,
+      "loss": 0.5698,
+      "step": 10096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.23263722309526e-06,
+      "loss": 0.646,
+      "step": 10097
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2307524063437605e-06,
+      "loss": 0.6035,
+      "step": 10098
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.228867808880362e-06,
+      "loss": 0.6409,
+      "step": 10099
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2269834307917215e-06,
+      "loss": 0.6017,
+      "step": 10100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.225099272164485e-06,
+      "loss": 0.6339,
+      "step": 10101
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.223215333085281e-06,
+      "loss": 0.6679,
+      "step": 10102
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.221331613640736e-06,
+      "loss": 0.6202,
+      "step": 10103
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.219448113917461e-06,
+      "loss": 0.6238,
+      "step": 10104
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.217564834002062e-06,
+      "loss": 0.6518,
+      "step": 10105
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.215681773981127e-06,
+      "loss": 0.6167,
+      "step": 10106
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.213798933941236e-06,
+      "loss": 0.6351,
+      "step": 10107
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.211916313968966e-06,
+      "loss": 0.5669,
+      "step": 10108
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.210033914150883e-06,
+      "loss": 0.6526,
+      "step": 10109
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.208151734573527e-06,
+      "loss": 0.6409,
+      "step": 10110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.206269775323449e-06,
+      "loss": 0.6319,
+      "step": 10111
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.204388036487178e-06,
+      "loss": 0.6068,
+      "step": 10112
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2025065181512405e-06,
+      "loss": 0.6528,
+      "step": 10113
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.200625220402139e-06,
+      "loss": 0.5947,
+      "step": 10114
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.19874414332638e-06,
+      "loss": 0.627,
+      "step": 10115
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.196863287010454e-06,
+      "loss": 0.567,
+      "step": 10116
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.19498265154084e-06,
+      "loss": 0.6333,
+      "step": 10117
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1931022370040126e-06,
+      "loss": 0.5571,
+      "step": 10118
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.191222043486433e-06,
+      "loss": 0.5159,
+      "step": 10119
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1893420710745454e-06,
+      "loss": 0.6193,
+      "step": 10120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.187462319854801e-06,
+      "loss": 0.5682,
+      "step": 10121
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1855827899136195e-06,
+      "loss": 0.6352,
+      "step": 10122
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.183703481337425e-06,
+      "loss": 0.6304,
+      "step": 10123
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.181824394212629e-06,
+      "loss": 0.5412,
+      "step": 10124
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1799455286256275e-06,
+      "loss": 0.679,
+      "step": 10125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.178066884662811e-06,
+      "loss": 0.5924,
+      "step": 10126
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.176188462410559e-06,
+      "loss": 0.6629,
+      "step": 10127
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.174310261955246e-06,
+      "loss": 0.6851,
+      "step": 10128
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.172432283383223e-06,
+      "loss": 0.6521,
+      "step": 10129
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1705545267808416e-06,
+      "loss": 0.6278,
+      "step": 10130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.168676992234441e-06,
+      "loss": 0.6638,
+      "step": 10131
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1667996798303506e-06,
+      "loss": 0.6382,
+      "step": 10132
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.164922589654885e-06,
+      "loss": 0.6218,
+      "step": 10133
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.163045721794353e-06,
+      "loss": 0.5514,
+      "step": 10134
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.161169076335055e-06,
+      "loss": 0.5805,
+      "step": 10135
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.159292653363274e-06,
+      "loss": 0.6647,
+      "step": 10136
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1574164529652925e-06,
+      "loss": 0.6407,
+      "step": 10137
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.155540475227373e-06,
+      "loss": 0.6365,
+      "step": 10138
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.15366472023578e-06,
+      "loss": 0.6013,
+      "step": 10139
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.151789188076747e-06,
+      "loss": 0.5842,
+      "step": 10140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.14991387883652e-06,
+      "loss": 0.6673,
+      "step": 10141
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1480387926013215e-06,
+      "loss": 0.6691,
+      "step": 10142
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.146163929457371e-06,
+      "loss": 0.6901,
+      "step": 10143
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.144289289490867e-06,
+      "loss": 0.6369,
+      "step": 10144
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1424148727880085e-06,
+      "loss": 0.6247,
+      "step": 10145
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1405406794349814e-06,
+      "loss": 0.6637,
+      "step": 10146
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.138666709517963e-06,
+      "loss": 0.632,
+      "step": 10147
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1367929631231095e-06,
+      "loss": 0.6186,
+      "step": 10148
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1349194403365815e-06,
+      "loss": 0.6625,
+      "step": 10149
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1330461412445195e-06,
+      "loss": 0.6641,
+      "step": 10150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.131173065933058e-06,
+      "loss": 0.6445,
+      "step": 10151
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.129300214488322e-06,
+      "loss": 0.609,
+      "step": 10152
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.127427586996422e-06,
+      "loss": 0.5912,
+      "step": 10153
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.125555183543463e-06,
+      "loss": 0.6158,
+      "step": 10154
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.12368300421554e-06,
+      "loss": 0.5945,
+      "step": 10155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1218110490987275e-06,
+      "loss": 0.6233,
+      "step": 10156
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.119939318279101e-06,
+      "loss": 0.5923,
+      "step": 10157
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1180678118427265e-06,
+      "loss": 0.5986,
+      "step": 10158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.116196529875645e-06,
+      "loss": 0.6876,
+      "step": 10159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1143254724639046e-06,
+      "loss": 0.5983,
+      "step": 10160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.112454639693533e-06,
+      "loss": 0.677,
+      "step": 10161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1105840316505555e-06,
+      "loss": 0.5962,
+      "step": 10162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.108713648420976e-06,
+      "loss": 0.5395,
+      "step": 10163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.106843490090794e-06,
+      "loss": 0.5957,
+      "step": 10164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.104973556746e-06,
+      "loss": 0.63,
+      "step": 10165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.103103848472579e-06,
+      "loss": 0.6629,
+      "step": 10166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.101234365356488e-06,
+      "loss": 0.6454,
+      "step": 10167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.099365107483691e-06,
+      "loss": 0.5626,
+      "step": 10168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.097496074940137e-06,
+      "loss": 0.6543,
+      "step": 10169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.095627267811761e-06,
+      "loss": 0.6258,
+      "step": 10170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0937586861844906e-06,
+      "loss": 0.6395,
+      "step": 10171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.091890330144244e-06,
+      "loss": 0.6621,
+      "step": 10172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.090022199776929e-06,
+      "loss": 0.5943,
+      "step": 10173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.088154295168436e-06,
+      "loss": 0.5984,
+      "step": 10174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.086286616404651e-06,
+      "loss": 0.6456,
+      "step": 10175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.084419163571454e-06,
+      "loss": 0.6797,
+      "step": 10176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.082551936754709e-06,
+      "loss": 0.5888,
+      "step": 10177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0806849360402665e-06,
+      "loss": 0.6758,
+      "step": 10178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.078818161513972e-06,
+      "loss": 0.617,
+      "step": 10179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0769516132616585e-06,
+      "loss": 0.6757,
+      "step": 10180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.075085291369154e-06,
+      "loss": 0.6581,
+      "step": 10181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.073219195922264e-06,
+      "loss": 0.639,
+      "step": 10182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0713533270067944e-06,
+      "loss": 0.674,
+      "step": 10183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.069487684708534e-06,
+      "loss": 0.6284,
+      "step": 10184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0676222691132704e-06,
+      "loss": 0.6527,
+      "step": 10185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.06575708030677e-06,
+      "loss": 0.6164,
+      "step": 10186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.063892118374796e-06,
+      "loss": 0.5628,
+      "step": 10187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.062027383403101e-06,
+      "loss": 0.6622,
+      "step": 10188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0601628754774164e-06,
+      "loss": 0.6067,
+      "step": 10189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.058298594683476e-06,
+      "loss": 0.6016,
+      "step": 10190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.056434541107002e-06,
+      "loss": 0.5802,
+      "step": 10191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.054570714833701e-06,
+      "loss": 0.5995,
+      "step": 10192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.052707115949268e-06,
+      "loss": 0.6445,
+      "step": 10193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.05084374453939e-06,
+      "loss": 0.6102,
+      "step": 10194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.048980600689749e-06,
+      "loss": 0.6502,
+      "step": 10195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.047117684486011e-06,
+      "loss": 0.6407,
+      "step": 10196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.045254996013828e-06,
+      "loss": 0.6575,
+      "step": 10197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.043392535358847e-06,
+      "loss": 0.7135,
+      "step": 10198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.041530302606705e-06,
+      "loss": 0.6275,
+      "step": 10199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.039668297843029e-06,
+      "loss": 0.667,
+      "step": 10200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.0378065211534265e-06,
+      "loss": 0.624,
+      "step": 10201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.035944972623505e-06,
+      "loss": 0.6019,
+      "step": 10202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.034083652338856e-06,
+      "loss": 0.5921,
+      "step": 10203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.032222560385065e-06,
+      "loss": 0.6118,
+      "step": 10204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.030361696847705e-06,
+      "loss": 0.5773,
+      "step": 10205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.028501061812334e-06,
+      "loss": 0.6009,
+      "step": 10206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.02664065536451e-06,
+      "loss": 0.5541,
+      "step": 10207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.024780477589764e-06,
+      "loss": 0.684,
+      "step": 10208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.022920528573631e-06,
+      "loss": 0.5969,
+      "step": 10209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.021060808401631e-06,
+      "loss": 0.6101,
+      "step": 10210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.019201317159276e-06,
+      "loss": 0.6703,
+      "step": 10211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.01734205493206e-06,
+      "loss": 0.6745,
+      "step": 10212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.015483021805473e-06,
+      "loss": 0.6771,
+      "step": 10213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.013624217864991e-06,
+      "loss": 0.678,
+      "step": 10214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.011765643196087e-06,
+      "loss": 0.6133,
+      "step": 10215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.009907297884209e-06,
+      "loss": 0.6615,
+      "step": 10216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.008049182014807e-06,
+      "loss": 0.6051,
+      "step": 10217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.006191295673318e-06,
+      "loss": 0.5705,
+      "step": 10218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.004333638945162e-06,
+      "loss": 0.6103,
+      "step": 10219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.002476211915758e-06,
+      "loss": 0.6947,
+      "step": 10220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.00061901467051e-06,
+      "loss": 0.643,
+      "step": 10221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.998762047294812e-06,
+      "loss": 0.5461,
+      "step": 10222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.99690530987404e-06,
+      "loss": 0.623,
+      "step": 10223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.995048802493571e-06,
+      "loss": 0.5962,
+      "step": 10224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.993192525238767e-06,
+      "loss": 0.6302,
+      "step": 10225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.99133647819498e-06,
+      "loss": 0.5439,
+      "step": 10226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.989480661447544e-06,
+      "loss": 0.6817,
+      "step": 10227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.987625075081794e-06,
+      "loss": 0.618,
+      "step": 10228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.985769719183045e-06,
+      "loss": 0.6761,
+      "step": 10229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9839145938366145e-06,
+      "loss": 0.682,
+      "step": 10230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.982059699127789e-06,
+      "loss": 0.6277,
+      "step": 10231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.980205035141862e-06,
+      "loss": 0.6415,
+      "step": 10232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.978350601964109e-06,
+      "loss": 0.5311,
+      "step": 10233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.976496399679802e-06,
+      "loss": 0.6335,
+      "step": 10234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.974642428374183e-06,
+      "loss": 0.5922,
+      "step": 10235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9727886881325065e-06,
+      "loss": 0.6208,
+      "step": 10236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.970935179040006e-06,
+      "loss": 0.6199,
+      "step": 10237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.969081901181904e-06,
+      "loss": 0.653,
+      "step": 10238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.967228854643413e-06,
+      "loss": 0.5826,
+      "step": 10239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.965376039509735e-06,
+      "loss": 0.5897,
+      "step": 10240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.963523455866068e-06,
+      "loss": 0.5776,
+      "step": 10241
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.961671103797581e-06,
+      "loss": 0.6837,
+      "step": 10242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9598189833894545e-06,
+      "loss": 0.5913,
+      "step": 10243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.957967094726844e-06,
+      "loss": 0.6195,
+      "step": 10244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.956115437894902e-06,
+      "loss": 0.6303,
+      "step": 10245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.954264012978759e-06,
+      "loss": 0.6222,
+      "step": 10246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.952412820063551e-06,
+      "loss": 0.6131,
+      "step": 10247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.950561859234392e-06,
+      "loss": 0.6357,
+      "step": 10248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.948711130576392e-06,
+      "loss": 0.6994,
+      "step": 10249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.94686063417464e-06,
+      "loss": 0.6335,
+      "step": 10250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.945010370114222e-06,
+      "loss": 0.6383,
+      "step": 10251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9431603384802175e-06,
+      "loss": 0.6473,
+      "step": 10252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.941310539357687e-06,
+      "loss": 0.6612,
+      "step": 10253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.939460972831684e-06,
+      "loss": 0.5782,
+      "step": 10254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.937611638987252e-06,
+      "loss": 0.5821,
+      "step": 10255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.935762537909423e-06,
+      "loss": 0.5958,
+      "step": 10256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.933913669683215e-06,
+      "loss": 0.6521,
+      "step": 10257
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.932065034393641e-06,
+      "loss": 0.6592,
+      "step": 10258
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.930216632125697e-06,
+      "loss": 0.6018,
+      "step": 10259
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.928368462964377e-06,
+      "loss": 0.6914,
+      "step": 10260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.926520526994655e-06,
+      "loss": 0.6098,
+      "step": 10261
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9246728243014985e-06,
+      "loss": 0.6154,
+      "step": 10262
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.922825354969866e-06,
+      "loss": 0.6288,
+      "step": 10263
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.920978119084704e-06,
+      "loss": 0.5316,
+      "step": 10264
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.919131116730945e-06,
+      "loss": 0.6384,
+      "step": 10265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.917284347993514e-06,
+      "loss": 0.6146,
+      "step": 10266
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.915437812957326e-06,
+      "loss": 0.6481,
+      "step": 10267
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.913591511707282e-06,
+      "loss": 0.6136,
+      "step": 10268
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.911745444328275e-06,
+      "loss": 0.6105,
+      "step": 10269
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.909899610905188e-06,
+      "loss": 0.65,
+      "step": 10270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9080540115228905e-06,
+      "loss": 0.658,
+      "step": 10271
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.906208646266247e-06,
+      "loss": 0.5469,
+      "step": 10272
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.904363515220099e-06,
+      "loss": 0.5887,
+      "step": 10273
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.902518618469287e-06,
+      "loss": 0.6427,
+      "step": 10274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.900673956098644e-06,
+      "loss": 0.6472,
+      "step": 10275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.898829528192978e-06,
+      "loss": 0.6589,
+      "step": 10276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8969853348371e-06,
+      "loss": 0.6303,
+      "step": 10277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.895141376115805e-06,
+      "loss": 0.6362,
+      "step": 10278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.893297652113882e-06,
+      "loss": 0.5984,
+      "step": 10279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.891454162916095e-06,
+      "loss": 0.5889,
+      "step": 10280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.889610908607213e-06,
+      "loss": 0.6199,
+      "step": 10281
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.887767889271987e-06,
+      "loss": 0.6185,
+      "step": 10282
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.885925104995161e-06,
+      "loss": 0.6458,
+      "step": 10283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8840825558614615e-06,
+      "loss": 0.6542,
+      "step": 10284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.882240241955608e-06,
+      "loss": 0.7011,
+      "step": 10285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.88039816336231e-06,
+      "loss": 0.6276,
+      "step": 10286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.878556320166269e-06,
+      "loss": 0.6309,
+      "step": 10287
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.876714712452168e-06,
+      "loss": 0.6326,
+      "step": 10288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.874873340304686e-06,
+      "loss": 0.6842,
+      "step": 10289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.873032203808491e-06,
+      "loss": 0.612,
+      "step": 10290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.871191303048231e-06,
+      "loss": 0.6968,
+      "step": 10291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.869350638108551e-06,
+      "loss": 0.5923,
+      "step": 10292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.867510209074088e-06,
+      "loss": 0.6093,
+      "step": 10293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.865670016029467e-06,
+      "loss": 0.6136,
+      "step": 10294
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.86383005905929e-06,
+      "loss": 0.5985,
+      "step": 10295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.861990338248163e-06,
+      "loss": 0.6615,
+      "step": 10296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.860150853680674e-06,
+      "loss": 0.639,
+      "step": 10297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.858311605441407e-06,
+      "loss": 0.675,
+      "step": 10298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.856472593614924e-06,
+      "loss": 0.6009,
+      "step": 10299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.854633818285781e-06,
+      "loss": 0.635,
+      "step": 10300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.852795279538528e-06,
+      "loss": 0.5984,
+      "step": 10301
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.850956977457699e-06,
+      "loss": 0.5939,
+      "step": 10302
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.849118912127817e-06,
+      "loss": 0.6134,
+      "step": 10303
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.847281083633398e-06,
+      "loss": 0.6289,
+      "step": 10304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.845443492058944e-06,
+      "loss": 0.6143,
+      "step": 10305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8436061374889495e-06,
+      "loss": 0.6247,
+      "step": 10306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8417690200078886e-06,
+      "loss": 0.6249,
+      "step": 10307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8399321397002355e-06,
+      "loss": 0.5862,
+      "step": 10308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.838095496650451e-06,
+      "loss": 0.6799,
+      "step": 10309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.836259090942977e-06,
+      "loss": 0.6136,
+      "step": 10310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.834422922662256e-06,
+      "loss": 0.5787,
+      "step": 10311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.832586991892711e-06,
+      "loss": 0.6429,
+      "step": 10312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.830751298718764e-06,
+      "loss": 0.61,
+      "step": 10313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.828915843224812e-06,
+      "loss": 0.6379,
+      "step": 10314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.827080625495249e-06,
+      "loss": 0.5778,
+      "step": 10315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.82524564561446e-06,
+      "loss": 0.6335,
+      "step": 10316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.823410903666821e-06,
+      "loss": 0.5747,
+      "step": 10317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.821576399736683e-06,
+      "loss": 0.6486,
+      "step": 10318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.819742133908401e-06,
+      "loss": 0.5851,
+      "step": 10319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.817908106266314e-06,
+      "loss": 0.6518,
+      "step": 10320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.816074316894749e-06,
+      "loss": 0.6212,
+      "step": 10321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.814240765878025e-06,
+      "loss": 0.7122,
+      "step": 10322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.812407453300444e-06,
+      "loss": 0.6035,
+      "step": 10323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8105743792463095e-06,
+      "loss": 0.6202,
+      "step": 10324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.808741543799894e-06,
+      "loss": 0.6558,
+      "step": 10325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.806908947045476e-06,
+      "loss": 0.6263,
+      "step": 10326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.805076589067316e-06,
+      "loss": 0.6163,
+      "step": 10327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.803244469949673e-06,
+      "loss": 0.6133,
+      "step": 10328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8014125897767746e-06,
+      "loss": 0.7316,
+      "step": 10329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.799580948632857e-06,
+      "loss": 0.563,
+      "step": 10330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.797749546602136e-06,
+      "loss": 0.6552,
+      "step": 10331
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7959183837688254e-06,
+      "loss": 0.6315,
+      "step": 10332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.794087460217111e-06,
+      "loss": 0.5944,
+      "step": 10333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.792256776031184e-06,
+      "loss": 0.6277,
+      "step": 10334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.790426331295216e-06,
+      "loss": 0.5739,
+      "step": 10335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.788596126093372e-06,
+      "loss": 0.6149,
+      "step": 10336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.786766160509804e-06,
+      "loss": 0.6497,
+      "step": 10337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.784936434628652e-06,
+      "loss": 0.5902,
+      "step": 10338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7831069485340514e-06,
+      "loss": 0.6071,
+      "step": 10339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7812777023101136e-06,
+      "loss": 0.669,
+      "step": 10340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.779448696040947e-06,
+      "loss": 0.6345,
+      "step": 10341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.777619929810653e-06,
+      "loss": 0.679,
+      "step": 10342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.775791403703318e-06,
+      "loss": 0.6456,
+      "step": 10343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.773963117803013e-06,
+      "loss": 0.6419,
+      "step": 10344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.772135072193804e-06,
+      "loss": 0.6645,
+      "step": 10345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.770307266959741e-06,
+      "loss": 0.6283,
+      "step": 10346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7684797021848725e-06,
+      "loss": 0.579,
+      "step": 10347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.766652377953221e-06,
+      "loss": 0.6518,
+      "step": 10348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.764825294348811e-06,
+      "loss": 0.6296,
+      "step": 10349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.762998451455648e-06,
+      "loss": 0.6254,
+      "step": 10350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.761171849357737e-06,
+      "loss": 0.6068,
+      "step": 10351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7593454881390545e-06,
+      "loss": 0.6882,
+      "step": 10352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.757519367883581e-06,
+      "loss": 0.6249,
+      "step": 10353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.755693488675279e-06,
+      "loss": 0.6595,
+      "step": 10354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.753867850598103e-06,
+      "loss": 0.6253,
+      "step": 10355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.752042453735996e-06,
+      "loss": 0.638,
+      "step": 10356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.750217298172887e-06,
+      "loss": 0.6179,
+      "step": 10357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.748392383992701e-06,
+      "loss": 0.6662,
+      "step": 10358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.746567711279338e-06,
+      "loss": 0.6345,
+      "step": 10359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.744743280116702e-06,
+      "loss": 0.5864,
+      "step": 10360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.742919090588676e-06,
+      "loss": 0.6585,
+      "step": 10361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.741095142779142e-06,
+      "loss": 0.6071,
+      "step": 10362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.739271436771957e-06,
+      "loss": 0.6135,
+      "step": 10363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.737447972650975e-06,
+      "loss": 0.6194,
+      "step": 10364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.735624750500042e-06,
+      "loss": 0.5552,
+      "step": 10365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.733801770402992e-06,
+      "loss": 0.6092,
+      "step": 10366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.731979032443637e-06,
+      "loss": 0.6505,
+      "step": 10367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.730156536705786e-06,
+      "loss": 0.6054,
+      "step": 10368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.728334283273243e-06,
+      "loss": 0.682,
+      "step": 10369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.726512272229788e-06,
+      "loss": 0.6591,
+      "step": 10370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.724690503659202e-06,
+      "loss": 0.7037,
+      "step": 10371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.722868977645245e-06,
+      "loss": 0.6541,
+      "step": 10372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.721047694271676e-06,
+      "loss": 0.6626,
+      "step": 10373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7192266536222295e-06,
+      "loss": 0.6156,
+      "step": 10374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.717405855780639e-06,
+      "loss": 0.5043,
+      "step": 10375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7155853008306235e-06,
+      "loss": 0.6514,
+      "step": 10376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7137649888558985e-06,
+      "loss": 0.5757,
+      "step": 10377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.711944919940149e-06,
+      "loss": 0.6478,
+      "step": 10378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.710125094167067e-06,
+      "loss": 0.6733,
+      "step": 10379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.708305511620328e-06,
+      "loss": 0.6476,
+      "step": 10380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7064861723835995e-06,
+      "loss": 0.6143,
+      "step": 10381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.704667076540525e-06,
+      "loss": 0.6108,
+      "step": 10382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.702848224174751e-06,
+      "loss": 0.7191,
+      "step": 10383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.701029615369907e-06,
+      "loss": 0.6169,
+      "step": 10384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6992112502096145e-06,
+      "loss": 0.6115,
+      "step": 10385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.697393128777477e-06,
+      "loss": 0.6202,
+      "step": 10386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.695575251157091e-06,
+      "loss": 0.6096,
+      "step": 10387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.693757617432044e-06,
+      "loss": 0.6258,
+      "step": 10388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6919402276859075e-06,
+      "loss": 0.7207,
+      "step": 10389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.690123082002249e-06,
+      "loss": 0.6527,
+      "step": 10390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.688306180464616e-06,
+      "loss": 0.628,
+      "step": 10391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.686489523156555e-06,
+      "loss": 0.5853,
+      "step": 10392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6846731101615866e-06,
+      "loss": 0.5764,
+      "step": 10393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.682856941563233e-06,
+      "loss": 0.6052,
+      "step": 10394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.681041017445001e-06,
+      "loss": 0.6762,
+      "step": 10395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.679225337890388e-06,
+      "loss": 0.6676,
+      "step": 10396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.677409902982873e-06,
+      "loss": 0.6352,
+      "step": 10397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.675594712805933e-06,
+      "loss": 0.637,
+      "step": 10398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.673779767443026e-06,
+      "loss": 0.6626,
+      "step": 10399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.67196506697761e-06,
+      "loss": 0.5883,
+      "step": 10400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.670150611493116e-06,
+      "loss": 0.6053,
+      "step": 10401
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.668336401072974e-06,
+      "loss": 0.6391,
+      "step": 10402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.666522435800604e-06,
+      "loss": 0.6514,
+      "step": 10403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.664708715759408e-06,
+      "loss": 0.6292,
+      "step": 10404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.662895241032781e-06,
+      "loss": 0.6067,
+      "step": 10405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.661082011704107e-06,
+      "loss": 0.631,
+      "step": 10406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6592690278567614e-06,
+      "loss": 0.5957,
+      "step": 10407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.657456289574095e-06,
+      "loss": 0.6235,
+      "step": 10408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6556437969394615e-06,
+      "loss": 0.5804,
+      "step": 10409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6538315500362e-06,
+      "loss": 0.6551,
+      "step": 10410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.652019548947639e-06,
+      "loss": 0.5665,
+      "step": 10411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.650207793757086e-06,
+      "loss": 0.6104,
+      "step": 10412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6483962845478484e-06,
+      "loss": 0.5997,
+      "step": 10413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6465850214032205e-06,
+      "loss": 0.6489,
+      "step": 10414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.644774004406484e-06,
+      "loss": 0.584,
+      "step": 10415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6429632336409055e-06,
+      "loss": 0.6395,
+      "step": 10416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.641152709189745e-06,
+      "loss": 0.6717,
+      "step": 10417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.639342431136249e-06,
+      "loss": 0.645,
+      "step": 10418
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.637532399563657e-06,
+      "loss": 0.7043,
+      "step": 10419
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6357226145551885e-06,
+      "loss": 0.6226,
+      "step": 10420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.633913076194059e-06,
+      "loss": 0.6963,
+      "step": 10421
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.632103784563468e-06,
+      "loss": 0.6301,
+      "step": 10422
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.630294739746611e-06,
+      "loss": 0.6014,
+      "step": 10423
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6284859418266625e-06,
+      "loss": 0.5765,
+      "step": 10424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.626677390886792e-06,
+      "loss": 0.6578,
+      "step": 10425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.62486908701016e-06,
+      "loss": 0.6389,
+      "step": 10426
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.623061030279904e-06,
+      "loss": 0.6061,
+      "step": 10427
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.621253220779163e-06,
+      "loss": 0.5887,
+      "step": 10428
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.619445658591057e-06,
+      "loss": 0.6041,
+      "step": 10429
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.617638343798701e-06,
+      "loss": 0.6238,
+      "step": 10430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.615831276485188e-06,
+      "loss": 0.6372,
+      "step": 10431
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.61402445673361e-06,
+      "loss": 0.6146,
+      "step": 10432
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.612217884627042e-06,
+      "loss": 0.6086,
+      "step": 10433
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.610411560248554e-06,
+      "loss": 0.6781,
+      "step": 10434
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.608605483681195e-06,
+      "loss": 0.5861,
+      "step": 10435
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.60679965500801e-06,
+      "loss": 0.6699,
+      "step": 10436
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.604994074312029e-06,
+      "loss": 0.6419,
+      "step": 10437
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.603188741676274e-06,
+      "loss": 0.6353,
+      "step": 10438
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.601383657183751e-06,
+      "loss": 0.6093,
+      "step": 10439
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.59957882091746e-06,
+      "loss": 0.6499,
+      "step": 10440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.597774232960387e-06,
+      "loss": 0.6277,
+      "step": 10441
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.595969893395502e-06,
+      "loss": 0.6412,
+      "step": 10442
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5941658023057685e-06,
+      "loss": 0.6446,
+      "step": 10443
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.59236195977414e-06,
+      "loss": 0.6694,
+      "step": 10444
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.590558365883562e-06,
+      "loss": 0.692,
+      "step": 10445
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.588755020716951e-06,
+      "loss": 0.6187,
+      "step": 10446
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5869519243572305e-06,
+      "loss": 0.6361,
+      "step": 10447
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.585149076887306e-06,
+      "loss": 0.6172,
+      "step": 10448
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.583346478390074e-06,
+      "loss": 0.6303,
+      "step": 10449
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.581544128948413e-06,
+      "loss": 0.6205,
+      "step": 10450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.579742028645195e-06,
+      "loss": 0.6566,
+      "step": 10451
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.577940177563281e-06,
+      "loss": 0.6066,
+      "step": 10452
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.576138575785518e-06,
+      "loss": 0.6313,
+      "step": 10453
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.574337223394745e-06,
+      "loss": 0.6861,
+      "step": 10454
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.572536120473788e-06,
+      "loss": 0.6085,
+      "step": 10455
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.570735267105459e-06,
+      "loss": 0.638,
+      "step": 10456
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.568934663372564e-06,
+      "loss": 0.6404,
+      "step": 10457
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.567134309357889e-06,
+      "loss": 0.5708,
+      "step": 10458
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.565334205144213e-06,
+      "loss": 0.6782,
+      "step": 10459
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.563534350814313e-06,
+      "loss": 0.5794,
+      "step": 10460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.561734746450936e-06,
+      "loss": 0.6956,
+      "step": 10461
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.559935392136829e-06,
+      "loss": 0.6519,
+      "step": 10462
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.558136287954728e-06,
+      "loss": 0.6085,
+      "step": 10463
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.556337433987359e-06,
+      "loss": 0.6032,
+      "step": 10464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5545388303174215e-06,
+      "loss": 0.5988,
+      "step": 10465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.552740477027623e-06,
+      "loss": 0.662,
+      "step": 10466
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5509423742006486e-06,
+      "loss": 0.6836,
+      "step": 10467
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.549144521919178e-06,
+      "loss": 0.6137,
+      "step": 10468
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.547346920265869e-06,
+      "loss": 0.5792,
+      "step": 10469
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.545549569323378e-06,
+      "loss": 0.6253,
+      "step": 10470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.543752469174347e-06,
+      "loss": 0.5666,
+      "step": 10471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.541955619901404e-06,
+      "loss": 0.6827,
+      "step": 10472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.54015902158717e-06,
+      "loss": 0.6141,
+      "step": 10473
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.53836267431425e-06,
+      "loss": 0.5737,
+      "step": 10474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.536566578165245e-06,
+      "loss": 0.5681,
+      "step": 10475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.534770733222729e-06,
+      "loss": 0.7319,
+      "step": 10476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.532975139569281e-06,
+      "loss": 0.6147,
+      "step": 10477
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.531179797287458e-06,
+      "loss": 0.6364,
+      "step": 10478
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.529384706459815e-06,
+      "loss": 0.5932,
+      "step": 10479
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.527589867168882e-06,
+      "loss": 0.5632,
+      "step": 10480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.525795279497189e-06,
+      "loss": 0.6018,
+      "step": 10481
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5240009435272495e-06,
+      "loss": 0.6131,
+      "step": 10482
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.522206859341571e-06,
+      "loss": 0.6413,
+      "step": 10483
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.520413027022635e-06,
+      "loss": 0.6148,
+      "step": 10484
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.518619446652929e-06,
+      "loss": 0.6198,
+      "step": 10485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.516826118314917e-06,
+      "loss": 0.5923,
+      "step": 10486
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.515033042091059e-06,
+      "loss": 0.5948,
+      "step": 10487
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.513240218063797e-06,
+      "loss": 0.6317,
+      "step": 10488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.511447646315568e-06,
+      "loss": 0.6547,
+      "step": 10489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.509655326928792e-06,
+      "loss": 0.6764,
+      "step": 10490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.507863259985877e-06,
+      "loss": 0.598,
+      "step": 10491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.506071445569223e-06,
+      "loss": 0.6669,
+      "step": 10492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.504279883761216e-06,
+      "loss": 0.6609,
+      "step": 10493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.502488574644234e-06,
+      "loss": 0.6556,
+      "step": 10494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.500697518300638e-06,
+      "loss": 0.6355,
+      "step": 10495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.498906714812781e-06,
+      "loss": 0.6602,
+      "step": 10496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.497116164263002e-06,
+      "loss": 0.5899,
+      "step": 10497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.495325866733636e-06,
+      "loss": 0.736,
+      "step": 10498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.493535822306992e-06,
+      "loss": 0.6237,
+      "step": 10499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.491746031065379e-06,
+      "loss": 0.5523,
+      "step": 10500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.489956493091089e-06,
+      "loss": 0.5926,
+      "step": 10501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.488167208466412e-06,
+      "loss": 0.5834,
+      "step": 10502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.486378177273607e-06,
+      "loss": 0.6123,
+      "step": 10503
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.484589399594939e-06,
+      "loss": 0.6667,
+      "step": 10504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.482800875512656e-06,
+      "loss": 0.6475,
+      "step": 10505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.481012605108991e-06,
+      "loss": 0.665,
+      "step": 10506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.479224588466172e-06,
+      "loss": 0.6289,
+      "step": 10507
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4774368256664075e-06,
+      "loss": 0.5318,
+      "step": 10508
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.475649316791901e-06,
+      "loss": 0.5895,
+      "step": 10509
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.473862061924839e-06,
+      "loss": 0.6242,
+      "step": 10510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4720750611473984e-06,
+      "loss": 0.5815,
+      "step": 10511
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.470288314541745e-06,
+      "loss": 0.6841,
+      "step": 10512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.468501822190039e-06,
+      "loss": 0.6307,
+      "step": 10513
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.466715584174412e-06,
+      "loss": 0.6038,
+      "step": 10514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.464929600577001e-06,
+      "loss": 0.5901,
+      "step": 10515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.463143871479924e-06,
+      "loss": 0.6642,
+      "step": 10516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.461358396965291e-06,
+      "loss": 0.5911,
+      "step": 10517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.45957317711519e-06,
+      "loss": 0.6715,
+      "step": 10518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.457788212011708e-06,
+      "loss": 0.6269,
+      "step": 10519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.456003501736919e-06,
+      "loss": 0.6401,
+      "step": 10520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.454219046372882e-06,
+      "loss": 0.6367,
+      "step": 10521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4524348460016465e-06,
+      "loss": 0.6259,
+      "step": 10522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.450650900705248e-06,
+      "loss": 0.6163,
+      "step": 10523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.448867210565714e-06,
+      "loss": 0.6205,
+      "step": 10524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4470837756650545e-06,
+      "loss": 0.5614,
+      "step": 10525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.445300596085272e-06,
+      "loss": 0.5781,
+      "step": 10526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.443517671908355e-06,
+      "loss": 0.706,
+      "step": 10527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4417350032162896e-06,
+      "loss": 0.6175,
+      "step": 10528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.439952590091032e-06,
+      "loss": 0.5915,
+      "step": 10529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.438170432614541e-06,
+      "loss": 0.6642,
+      "step": 10530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.436388530868759e-06,
+      "loss": 0.6731,
+      "step": 10531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.434606884935621e-06,
+      "loss": 0.5556,
+      "step": 10532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.43282549489704e-06,
+      "loss": 0.6451,
+      "step": 10533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4310443608349245e-06,
+      "loss": 0.6553,
+      "step": 10534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4292634828311755e-06,
+      "loss": 0.6786,
+      "step": 10535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4274828609676775e-06,
+      "loss": 0.5892,
+      "step": 10536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4257024953262945e-06,
+      "loss": 0.6535,
+      "step": 10537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4239223859888904e-06,
+      "loss": 0.6619,
+      "step": 10538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.422142533037318e-06,
+      "loss": 0.64,
+      "step": 10539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.42036293655341e-06,
+      "loss": 0.5936,
+      "step": 10540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.418583596618993e-06,
+      "loss": 0.6156,
+      "step": 10541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.416804513315881e-06,
+      "loss": 0.696,
+      "step": 10542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.415025686725879e-06,
+      "loss": 0.5908,
+      "step": 10543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.413247116930767e-06,
+      "loss": 0.6586,
+      "step": 10544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4114688040123305e-06,
+      "loss": 0.5376,
+      "step": 10545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4096907480523345e-06,
+      "loss": 0.5914,
+      "step": 10546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.407912949132535e-06,
+      "loss": 0.5865,
+      "step": 10547
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.406135407334668e-06,
+      "loss": 0.6617,
+      "step": 10548
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.404358122740469e-06,
+      "loss": 0.6416,
+      "step": 10549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4025810954316565e-06,
+      "loss": 0.6633,
+      "step": 10550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.400804325489941e-06,
+      "loss": 0.6738,
+      "step": 10551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.399027812997008e-06,
+      "loss": 0.592,
+      "step": 10552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.397251558034548e-06,
+      "loss": 0.5889,
+      "step": 10553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.39547556068423e-06,
+      "loss": 0.574,
+      "step": 10554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.393699821027716e-06,
+      "loss": 0.6307,
+      "step": 10555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.391924339146652e-06,
+      "loss": 0.7109,
+      "step": 10556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3901491151226745e-06,
+      "loss": 0.6368,
+      "step": 10557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.388374149037411e-06,
+      "loss": 0.5983,
+      "step": 10558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.386599440972467e-06,
+      "loss": 0.588,
+      "step": 10559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.384824991009447e-06,
+      "loss": 0.607,
+      "step": 10560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.383050799229936e-06,
+      "loss": 0.5967,
+      "step": 10561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.38127686571552e-06,
+      "loss": 0.6525,
+      "step": 10562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.379503190547751e-06,
+      "loss": 0.5939,
+      "step": 10563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.377729773808188e-06,
+      "loss": 0.6583,
+      "step": 10564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.375956615578375e-06,
+      "loss": 0.707,
+      "step": 10565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.374183715939839e-06,
+      "loss": 0.6558,
+      "step": 10566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.372411074974093e-06,
+      "loss": 0.5988,
+      "step": 10567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.370638692762647e-06,
+      "loss": 0.6426,
+      "step": 10568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.368866569386992e-06,
+      "loss": 0.6563,
+      "step": 10569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.367094704928614e-06,
+      "loss": 0.6068,
+      "step": 10570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.365323099468976e-06,
+      "loss": 0.5656,
+      "step": 10571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.363551753089538e-06,
+      "loss": 0.6851,
+      "step": 10572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.361780665871747e-06,
+      "loss": 0.5548,
+      "step": 10573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.360009837897036e-06,
+      "loss": 0.5615,
+      "step": 10574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.358239269246828e-06,
+      "loss": 0.6632,
+      "step": 10575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.356468960002533e-06,
+      "loss": 0.6694,
+      "step": 10576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.354698910245552e-06,
+      "loss": 0.6573,
+      "step": 10577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.352929120057265e-06,
+      "loss": 0.5991,
+      "step": 10578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.351159589519048e-06,
+      "loss": 0.5912,
+      "step": 10579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.349390318712265e-06,
+      "loss": 0.6597,
+      "step": 10580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.34762130771827e-06,
+      "loss": 0.53,
+      "step": 10581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3458525566183924e-06,
+      "loss": 0.6254,
+      "step": 10582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.344084065493965e-06,
+      "loss": 0.6829,
+      "step": 10583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3423158344263e-06,
+      "loss": 0.6021,
+      "step": 10584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.340547863496703e-06,
+      "loss": 0.6457,
+      "step": 10585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.338780152786461e-06,
+      "loss": 0.6404,
+      "step": 10586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.337012702376852e-06,
+      "loss": 0.637,
+      "step": 10587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.335245512349147e-06,
+      "loss": 0.6956,
+      "step": 10588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3334785827845965e-06,
+      "loss": 0.6532,
+      "step": 10589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.331711913764447e-06,
+      "loss": 0.561,
+      "step": 10590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.329945505369924e-06,
+      "loss": 0.5506,
+      "step": 10591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3281793576822565e-06,
+      "loss": 0.6275,
+      "step": 10592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.326413470782638e-06,
+      "loss": 0.6298,
+      "step": 10593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.324647844752273e-06,
+      "loss": 0.5785,
+      "step": 10594
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.322882479672338e-06,
+      "loss": 0.5978,
+      "step": 10595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.32111737562401e-06,
+      "loss": 0.6056,
+      "step": 10596
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.319352532688444e-06,
+      "loss": 0.6991,
+      "step": 10597
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.317587950946785e-06,
+      "loss": 0.5921,
+      "step": 10598
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3158236304801695e-06,
+      "loss": 0.6007,
+      "step": 10599
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.314059571369726e-06,
+      "loss": 0.5922,
+      "step": 10600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.312295773696555e-06,
+      "loss": 0.6187,
+      "step": 10601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.310532237541761e-06,
+      "loss": 0.6694,
+      "step": 10602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.30876896298643e-06,
+      "loss": 0.6828,
+      "step": 10603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.30700595011164e-06,
+      "loss": 0.6322,
+      "step": 10604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.305243198998447e-06,
+      "loss": 0.648,
+      "step": 10605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.303480709727904e-06,
+      "loss": 0.6114,
+      "step": 10606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3017184823810506e-06,
+      "loss": 0.5795,
+      "step": 10607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.299956517038913e-06,
+      "loss": 0.6232,
+      "step": 10608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.298194813782505e-06,
+      "loss": 0.5867,
+      "step": 10609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.296433372692831e-06,
+      "loss": 0.6494,
+      "step": 10610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.294672193850884e-06,
+      "loss": 0.6236,
+      "step": 10611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.292911277337634e-06,
+      "loss": 0.6766,
+      "step": 10612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.291150623234052e-06,
+      "loss": 0.6193,
+      "step": 10613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.28939023162109e-06,
+      "loss": 0.5563,
+      "step": 10614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.287630102579699e-06,
+      "loss": 0.6043,
+      "step": 10615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.285870236190795e-06,
+      "loss": 0.6148,
+      "step": 10616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.284110632535305e-06,
+      "loss": 0.6475,
+      "step": 10617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.282351291694132e-06,
+      "loss": 0.6087,
+      "step": 10618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.280592213748173e-06,
+      "loss": 0.6397,
+      "step": 10619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.278833398778306e-06,
+      "loss": 0.5952,
+      "step": 10620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.277074846865401e-06,
+      "loss": 0.5878,
+      "step": 10621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.275316558090317e-06,
+      "loss": 0.6439,
+      "step": 10622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.273558532533897e-06,
+      "loss": 0.6975,
+      "step": 10623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.271800770276976e-06,
+      "loss": 0.613,
+      "step": 10624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.270043271400378e-06,
+      "loss": 0.6548,
+      "step": 10625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.268286035984911e-06,
+      "loss": 0.6596,
+      "step": 10626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2665290641113675e-06,
+      "loss": 0.6368,
+      "step": 10627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.264772355860535e-06,
+      "loss": 0.6084,
+      "step": 10628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.263015911313187e-06,
+      "loss": 0.6261,
+      "step": 10629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.261259730550088e-06,
+      "loss": 0.6424,
+      "step": 10630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.259503813651977e-06,
+      "loss": 0.6814,
+      "step": 10631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.257748160699596e-06,
+      "loss": 0.5781,
+      "step": 10632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.255992771773669e-06,
+      "loss": 0.5992,
+      "step": 10633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.254237646954912e-06,
+      "loss": 0.6357,
+      "step": 10634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.252482786324016e-06,
+      "loss": 0.587,
+      "step": 10635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.250728189961674e-06,
+      "loss": 0.6434,
+      "step": 10636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.248973857948561e-06,
+      "loss": 0.6188,
+      "step": 10637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.247219790365341e-06,
+      "loss": 0.6407,
+      "step": 10638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.245465987292664e-06,
+      "loss": 0.6712,
+      "step": 10639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2437124488111716e-06,
+      "loss": 0.5907,
+      "step": 10640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.241959175001493e-06,
+      "loss": 0.6312,
+      "step": 10641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.240206165944235e-06,
+      "loss": 0.6424,
+      "step": 10642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.238453421720006e-06,
+      "loss": 0.6273,
+      "step": 10643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.236700942409393e-06,
+      "loss": 0.6208,
+      "step": 10644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.234948728092982e-06,
+      "loss": 0.5937,
+      "step": 10645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.233196778851329e-06,
+      "loss": 0.6972,
+      "step": 10646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.231445094764993e-06,
+      "loss": 0.6103,
+      "step": 10647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.229693675914516e-06,
+      "loss": 0.6526,
+      "step": 10648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.22794252238043e-06,
+      "loss": 0.6018,
+      "step": 10649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.226191634243245e-06,
+      "loss": 0.5557,
+      "step": 10650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.224441011583468e-06,
+      "loss": 0.5823,
+      "step": 10651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.2226906544815974e-06,
+      "loss": 0.7028,
+      "step": 10652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.220940563018114e-06,
+      "loss": 0.6756,
+      "step": 10653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.219190737273478e-06,
+      "loss": 0.6815,
+      "step": 10654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.217441177328151e-06,
+      "loss": 0.5944,
+      "step": 10655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.2156918832625785e-06,
+      "loss": 0.5975,
+      "step": 10656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.213942855157189e-06,
+      "loss": 0.6326,
+      "step": 10657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.212194093092404e-06,
+      "loss": 0.6396,
+      "step": 10658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.210445597148631e-06,
+      "loss": 0.6184,
+      "step": 10659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.208697367406269e-06,
+      "loss": 0.5714,
+      "step": 10660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.2069494039456915e-06,
+      "loss": 0.6836,
+      "step": 10661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.2052017068472756e-06,
+      "loss": 0.5631,
+      "step": 10662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.2034542761913775e-06,
+      "loss": 0.6894,
+      "step": 10663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.20170711205835e-06,
+      "loss": 0.6224,
+      "step": 10664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.199960214528516e-06,
+      "loss": 0.646,
+      "step": 10665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.198213583682203e-06,
+      "loss": 0.6047,
+      "step": 10666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.196467219599719e-06,
+      "loss": 0.5655,
+      "step": 10667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.194721122361366e-06,
+      "loss": 0.5624,
+      "step": 10668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.192975292047422e-06,
+      "loss": 0.6391,
+      "step": 10669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1912297287381625e-06,
+      "loss": 0.5197,
+      "step": 10670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.189484432513846e-06,
+      "loss": 0.6327,
+      "step": 10671
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.187739403454725e-06,
+      "loss": 0.6653,
+      "step": 10672
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.18599464164103e-06,
+      "loss": 0.6553,
+      "step": 10673
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1842501471529885e-06,
+      "loss": 0.6549,
+      "step": 10674
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.182505920070814e-06,
+      "loss": 0.621,
+      "step": 10675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.180761960474696e-06,
+      "loss": 0.6342,
+      "step": 10676
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.179018268444828e-06,
+      "loss": 0.6219,
+      "step": 10677
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.177274844061383e-06,
+      "loss": 0.6426,
+      "step": 10678
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.175531687404527e-06,
+      "loss": 0.6079,
+      "step": 10679
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1737887985543995e-06,
+      "loss": 0.6162,
+      "step": 10680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.172046177591146e-06,
+      "loss": 0.5736,
+      "step": 10681
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.170303824594889e-06,
+      "loss": 0.648,
+      "step": 10682
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.168561739645745e-06,
+      "loss": 0.6378,
+      "step": 10683
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.166819922823807e-06,
+      "loss": 0.5872,
+      "step": 10684
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.165078374209166e-06,
+      "loss": 0.671,
+      "step": 10685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1633370938819e-06,
+      "loss": 0.598,
+      "step": 10686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.161596081922075e-06,
+      "loss": 0.6645,
+      "step": 10687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.159855338409732e-06,
+      "loss": 0.549,
+      "step": 10688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.158114863424918e-06,
+      "loss": 0.6945,
+      "step": 10689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1563746570476565e-06,
+      "loss": 0.5865,
+      "step": 10690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1546347193579635e-06,
+      "loss": 0.6235,
+      "step": 10691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.152895050435836e-06,
+      "loss": 0.6298,
+      "step": 10692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.15115565036127e-06,
+      "loss": 0.6192,
+      "step": 10693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.14941651921424e-06,
+      "loss": 0.6915,
+      "step": 10694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.147677657074707e-06,
+      "loss": 0.5962,
+      "step": 10695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.145939064022624e-06,
+      "loss": 0.5849,
+      "step": 10696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.144200740137934e-06,
+      "loss": 0.6043,
+      "step": 10697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.142462685500568e-06,
+      "loss": 0.6146,
+      "step": 10698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.14072490019043e-06,
+      "loss": 0.648,
+      "step": 10699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.13898738428743e-06,
+      "loss": 0.5562,
+      "step": 10700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1372501378714575e-06,
+      "loss": 0.6219,
+      "step": 10701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.135513161022391e-06,
+      "loss": 0.6586,
+      "step": 10702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1337764538200936e-06,
+      "loss": 0.5826,
+      "step": 10703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.132040016344418e-06,
+      "loss": 0.6241,
+      "step": 10704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.130303848675208e-06,
+      "loss": 0.5861,
+      "step": 10705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.128567950892291e-06,
+      "loss": 0.6478,
+      "step": 10706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.126832323075481e-06,
+      "loss": 0.649,
+      "step": 10707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.125096965304582e-06,
+      "loss": 0.649,
+      "step": 10708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.123361877659392e-06,
+      "loss": 0.6344,
+      "step": 10709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.121627060219677e-06,
+      "loss": 0.6264,
+      "step": 10710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.119892513065212e-06,
+      "loss": 0.6001,
+      "step": 10711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.118158236275748e-06,
+      "loss": 0.6283,
+      "step": 10712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.116424229931029e-06,
+      "loss": 0.632,
+      "step": 10713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1146904941107814e-06,
+      "loss": 0.6828,
+      "step": 10714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.112957028894719e-06,
+      "loss": 0.6328,
+      "step": 10715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.11122383436255e-06,
+      "loss": 0.732,
+      "step": 10716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.10949091059397e-06,
+      "loss": 0.641,
+      "step": 10717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1077582576686485e-06,
+      "loss": 0.6144,
+      "step": 10718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.106025875666256e-06,
+      "loss": 0.6278,
+      "step": 10719
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.104293764666448e-06,
+      "loss": 0.6387,
+      "step": 10720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1025619247488705e-06,
+      "loss": 0.5651,
+      "step": 10721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.100830355993143e-06,
+      "loss": 0.6155,
+      "step": 10722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.099099058478888e-06,
+      "loss": 0.6161,
+      "step": 10723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.097368032285709e-06,
+      "loss": 0.6291,
+      "step": 10724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.095637277493199e-06,
+      "loss": 0.6496,
+      "step": 10725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0939067941809365e-06,
+      "loss": 0.6395,
+      "step": 10726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0921765824284876e-06,
+      "loss": 0.6943,
+      "step": 10727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.090446642315412e-06,
+      "loss": 0.5504,
+      "step": 10728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.088716973921243e-06,
+      "loss": 0.6483,
+      "step": 10729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.086987577325514e-06,
+      "loss": 0.5663,
+      "step": 10730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.085258452607745e-06,
+      "loss": 0.573,
+      "step": 10731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.08352959984744e-06,
+      "loss": 0.6051,
+      "step": 10732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.081801019124084e-06,
+      "loss": 0.5822,
+      "step": 10733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0800727105171635e-06,
+      "loss": 0.6,
+      "step": 10734
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.078344674106141e-06,
+      "loss": 0.5989,
+      "step": 10735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.076616909970479e-06,
+      "loss": 0.605,
+      "step": 10736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.074889418189609e-06,
+      "loss": 0.6577,
+      "step": 10737
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.073162198842964e-06,
+      "loss": 0.5973,
+      "step": 10738
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.071435252009962e-06,
+      "loss": 0.5516,
+      "step": 10739
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.069708577770009e-06,
+      "loss": 0.6061,
+      "step": 10740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.067982176202493e-06,
+      "loss": 0.6337,
+      "step": 10741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.066256047386797e-06,
+      "loss": 0.6506,
+      "step": 10742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0645301914022865e-06,
+      "loss": 0.6567,
+      "step": 10743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.062804608328314e-06,
+      "loss": 0.6078,
+      "step": 10744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.061079298244222e-06,
+      "loss": 0.6611,
+      "step": 10745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0593542612293375e-06,
+      "loss": 0.632,
+      "step": 10746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.057629497362985e-06,
+      "loss": 0.6376,
+      "step": 10747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0559050067244585e-06,
+      "loss": 0.6344,
+      "step": 10748
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.054180789393054e-06,
+      "loss": 0.6412,
+      "step": 10749
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.05245684544805e-06,
+      "loss": 0.5503,
+      "step": 10750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.050733174968716e-06,
+      "loss": 0.6806,
+      "step": 10751
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.049009778034298e-06,
+      "loss": 0.5819,
+      "step": 10752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.047286654724043e-06,
+      "loss": 0.652,
+      "step": 10753
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.045563805117177e-06,
+      "loss": 0.7152,
+      "step": 10754
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.04384122929292e-06,
+      "loss": 0.6604,
+      "step": 10755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.04211892733047e-06,
+      "loss": 0.6831,
+      "step": 10756
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.04039689930902e-06,
+      "loss": 0.5875,
+      "step": 10757
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.038675145307747e-06,
+      "loss": 0.6289,
+      "step": 10758
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.03695366540582e-06,
+      "loss": 0.5618,
+      "step": 10759
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.035232459682387e-06,
+      "loss": 0.5749,
+      "step": 10760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0335115282165925e-06,
+      "loss": 0.6008,
+      "step": 10761
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.031790871087568e-06,
+      "loss": 0.6127,
+      "step": 10762
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.030070488374416e-06,
+      "loss": 0.6542,
+      "step": 10763
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.02835038015625e-06,
+      "loss": 0.6447,
+      "step": 10764
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.026630546512154e-06,
+      "loss": 0.6643,
+      "step": 10765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.024910987521213e-06,
+      "loss": 0.6149,
+      "step": 10766
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.023191703262481e-06,
+      "loss": 0.6009,
+      "step": 10767
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0214726938150174e-06,
+      "loss": 0.6254,
+      "step": 10768
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.01975395925786e-06,
+      "loss": 0.6121,
+      "step": 10769
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.018035499670036e-06,
+      "loss": 0.5856,
+      "step": 10770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.016317315130556e-06,
+      "loss": 0.5892,
+      "step": 10771
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.014599405718423e-06,
+      "loss": 0.6992,
+      "step": 10772
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.012881771512627e-06,
+      "loss": 0.6473,
+      "step": 10773
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.011164412592145e-06,
+      "loss": 0.6454,
+      "step": 10774
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.009447329035939e-06,
+      "loss": 0.6273,
+      "step": 10775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.007730520922959e-06,
+      "loss": 0.5729,
+      "step": 10776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.00601398833215e-06,
+      "loss": 0.623,
+      "step": 10777
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.004297731342427e-06,
+      "loss": 0.6096,
+      "step": 10778
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0025817500327065e-06,
+      "loss": 0.6362,
+      "step": 10779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.000866044481892e-06,
+      "loss": 0.6299,
+      "step": 10780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9991506147688695e-06,
+      "loss": 0.5838,
+      "step": 10781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.997435460972511e-06,
+      "loss": 0.616,
+      "step": 10782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.99572058317168e-06,
+      "loss": 0.5811,
+      "step": 10783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.994005981445229e-06,
+      "loss": 0.6352,
+      "step": 10784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.992291655871993e-06,
+      "loss": 0.6452,
+      "step": 10785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.990577606530792e-06,
+      "loss": 0.6299,
+      "step": 10786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9888638335004406e-06,
+      "loss": 0.6192,
+      "step": 10787
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.987150336859737e-06,
+      "loss": 0.6151,
+      "step": 10788
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.985437116687472e-06,
+      "loss": 0.5973,
+      "step": 10789
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.98372417306241e-06,
+      "loss": 0.6951,
+      "step": 10790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.982011506063315e-06,
+      "loss": 0.5971,
+      "step": 10791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.980299115768936e-06,
+      "loss": 0.6559,
+      "step": 10792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.978587002258007e-06,
+      "loss": 0.6203,
+      "step": 10793
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.976875165609251e-06,
+      "loss": 0.6549,
+      "step": 10794
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.975163605901377e-06,
+      "loss": 0.5764,
+      "step": 10795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.973452323213087e-06,
+      "loss": 0.6122,
+      "step": 10796
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.971741317623055e-06,
+      "loss": 0.5985,
+      "step": 10797
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.970030589209958e-06,
+      "loss": 0.6328,
+      "step": 10798
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9683201380524535e-06,
+      "loss": 0.6155,
+      "step": 10799
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.966609964229193e-06,
+      "loss": 0.5907,
+      "step": 10800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9649000678188e-06,
+      "loss": 0.5467,
+      "step": 10801
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9631904488999e-06,
+      "loss": 0.5884,
+      "step": 10802
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.961481107551099e-06,
+      "loss": 0.5973,
+      "step": 10803
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9597720438509975e-06,
+      "loss": 0.6734,
+      "step": 10804
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.95806325787817e-06,
+      "loss": 0.6063,
+      "step": 10805
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.956354749711188e-06,
+      "loss": 0.5596,
+      "step": 10806
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.954646519428609e-06,
+      "loss": 0.5821,
+      "step": 10807
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.952938567108976e-06,
+      "loss": 0.6523,
+      "step": 10808
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.951230892830821e-06,
+      "loss": 0.6296,
+      "step": 10809
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.949523496672662e-06,
+      "loss": 0.6252,
+      "step": 10810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.947816378713006e-06,
+      "loss": 0.5678,
+      "step": 10811
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.946109539030341e-06,
+      "loss": 0.6554,
+      "step": 10812
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.94440297770315e-06,
+      "loss": 0.5518,
+      "step": 10813
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.942696694809901e-06,
+      "loss": 0.6003,
+      "step": 10814
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.940990690429048e-06,
+      "loss": 0.5608,
+      "step": 10815
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.939284964639029e-06,
+      "loss": 0.6527,
+      "step": 10816
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.937579517518274e-06,
+      "loss": 0.5911,
+      "step": 10817
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9358743491451995e-06,
+      "loss": 0.6337,
+      "step": 10818
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.934169459598212e-06,
+      "loss": 0.6804,
+      "step": 10819
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9324648489556945e-06,
+      "loss": 0.6059,
+      "step": 10820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.930760517296028e-06,
+      "loss": 0.6203,
+      "step": 10821
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9290564646975775e-06,
+      "loss": 0.6238,
+      "step": 10822
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.927352691238698e-06,
+      "loss": 0.5538,
+      "step": 10823
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.925649196997719e-06,
+      "loss": 0.6391,
+      "step": 10824
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9239459820529695e-06,
+      "loss": 0.6379,
+      "step": 10825
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.922243046482773e-06,
+      "loss": 0.5797,
+      "step": 10826
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9205403903654165e-06,
+      "loss": 0.5958,
+      "step": 10827
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.918838013779194e-06,
+      "loss": 0.6048,
+      "step": 10828
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.917135916802377e-06,
+      "loss": 0.5944,
+      "step": 10829
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.915434099513236e-06,
+      "loss": 0.5903,
+      "step": 10830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9137325619900075e-06,
+      "loss": 0.7357,
+      "step": 10831
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.912031304310932e-06,
+      "loss": 0.7293,
+      "step": 10832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.910330326554236e-06,
+      "loss": 0.6423,
+      "step": 10833
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9086296287981295e-06,
+      "loss": 0.6634,
+      "step": 10834
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.906929211120804e-06,
+      "loss": 0.5996,
+      "step": 10835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9052290736004495e-06,
+      "loss": 0.6105,
+      "step": 10836
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.903529216315235e-06,
+      "loss": 0.6223,
+      "step": 10837
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.901829639343322e-06,
+      "loss": 0.6232,
+      "step": 10838
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.900130342762855e-06,
+      "loss": 0.6422,
+      "step": 10839
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.898431326651965e-06,
+      "loss": 0.5869,
+      "step": 10840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.896732591088772e-06,
+      "loss": 0.6284,
+      "step": 10841
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.895034136151388e-06,
+      "loss": 0.5924,
+      "step": 10842
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.893335961917902e-06,
+      "loss": 0.6409,
+      "step": 10843
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8916380684663984e-06,
+      "loss": 0.6398,
+      "step": 10844
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.889940455874947e-06,
+      "loss": 0.585,
+      "step": 10845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8882431242216e-06,
+      "loss": 0.6892,
+      "step": 10846
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.886546073584399e-06,
+      "loss": 0.7123,
+      "step": 10847
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.884849304041378e-06,
+      "loss": 0.6077,
+      "step": 10848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.883152815670555e-06,
+      "loss": 0.6487,
+      "step": 10849
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.881456608549926e-06,
+      "loss": 0.6406,
+      "step": 10850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.879760682757487e-06,
+      "loss": 0.6128,
+      "step": 10851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.878065038371214e-06,
+      "loss": 0.6335,
+      "step": 10852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.876369675469079e-06,
+      "loss": 0.6943,
+      "step": 10853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.874674594129025e-06,
+      "loss": 0.5935,
+      "step": 10854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.872979794428995e-06,
+      "loss": 0.6289,
+      "step": 10855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.871285276446914e-06,
+      "loss": 0.5666,
+      "step": 10856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.869591040260697e-06,
+      "loss": 0.6569,
+      "step": 10857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.867897085948244e-06,
+      "loss": 0.6342,
+      "step": 10858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.866203413587441e-06,
+      "loss": 0.594,
+      "step": 10859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.864510023256169e-06,
+      "loss": 0.6587,
+      "step": 10860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8628169150322775e-06,
+      "loss": 0.5838,
+      "step": 10861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.861124088993623e-06,
+      "loss": 0.6108,
+      "step": 10862
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.859431545218037e-06,
+      "loss": 0.6595,
+      "step": 10863
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.857739283783351e-06,
+      "loss": 0.6293,
+      "step": 10864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.856047304767362e-06,
+      "loss": 0.591,
+      "step": 10865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.854355608247871e-06,
+      "loss": 0.6249,
+      "step": 10866
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8526641943026654e-06,
+      "loss": 0.6303,
+      "step": 10867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.850973063009515e-06,
+      "loss": 0.622,
+      "step": 10868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.849282214446173e-06,
+      "loss": 0.6624,
+      "step": 10869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.847591648690386e-06,
+      "loss": 0.6147,
+      "step": 10870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.845901365819885e-06,
+      "loss": 0.6278,
+      "step": 10871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.844211365912393e-06,
+      "loss": 0.6368,
+      "step": 10872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.842521649045609e-06,
+      "loss": 0.6342,
+      "step": 10873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8408322152972295e-06,
+      "loss": 0.5969,
+      "step": 10874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8391430647449325e-06,
+      "loss": 0.6193,
+      "step": 10875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.837454197466384e-06,
+      "loss": 0.6232,
+      "step": 10876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.835765613539239e-06,
+      "loss": 0.6093,
+      "step": 10877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.834077313041137e-06,
+      "loss": 0.6382,
+      "step": 10878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.832389296049709e-06,
+      "loss": 0.5928,
+      "step": 10879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.830701562642563e-06,
+      "loss": 0.6567,
+      "step": 10880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.829014112897303e-06,
+      "loss": 0.5955,
+      "step": 10881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.827326946891517e-06,
+      "loss": 0.5869,
+      "step": 10882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.825640064702785e-06,
+      "loss": 0.645,
+      "step": 10883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.823953466408662e-06,
+      "loss": 0.5748,
+      "step": 10884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8222671520866985e-06,
+      "loss": 0.5757,
+      "step": 10885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8205811218144334e-06,
+      "loss": 0.5992,
+      "step": 10886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8188953756693925e-06,
+      "loss": 0.7003,
+      "step": 10887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8172099137290775e-06,
+      "loss": 0.6047,
+      "step": 10888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8155247360709886e-06,
+      "loss": 0.6553,
+      "step": 10889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.813839842772612e-06,
+      "loss": 0.6128,
+      "step": 10890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8121552339114165e-06,
+      "loss": 0.5832,
+      "step": 10891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.810470909564859e-06,
+      "loss": 0.6028,
+      "step": 10892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.808786869810387e-06,
+      "loss": 0.6012,
+      "step": 10893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.807103114725432e-06,
+      "loss": 0.6369,
+      "step": 10894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.805419644387408e-06,
+      "loss": 0.6675,
+      "step": 10895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.803736458873723e-06,
+      "loss": 0.631,
+      "step": 10896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.802053558261769e-06,
+      "loss": 0.6034,
+      "step": 10897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.800370942628928e-06,
+      "loss": 0.6828,
+      "step": 10898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.79868861205256e-06,
+      "loss": 0.6171,
+      "step": 10899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7970065666100205e-06,
+      "loss": 0.623,
+      "step": 10900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.795324806378651e-06,
+      "loss": 0.5629,
+      "step": 10901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.79364333143578e-06,
+      "loss": 0.5901,
+      "step": 10902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.791962141858715e-06,
+      "loss": 0.5851,
+      "step": 10903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.790281237724759e-06,
+      "loss": 0.6533,
+      "step": 10904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.788600619111199e-06,
+      "loss": 0.6267,
+      "step": 10905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.786920286095315e-06,
+      "loss": 0.6314,
+      "step": 10906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.785240238754359e-06,
+      "loss": 0.5169,
+      "step": 10907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.783560477165582e-06,
+      "loss": 0.6534,
+      "step": 10908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.78188100140622e-06,
+      "loss": 0.6252,
+      "step": 10909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7802018115534954e-06,
+      "loss": 0.5525,
+      "step": 10910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7785229076846135e-06,
+      "loss": 0.7047,
+      "step": 10911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.776844289876774e-06,
+      "loss": 0.6072,
+      "step": 10912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7751659582071594e-06,
+      "loss": 0.635,
+      "step": 10913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.773487912752932e-06,
+      "loss": 0.5839,
+      "step": 10914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.771810153591253e-06,
+      "loss": 0.6034,
+      "step": 10915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.770132680799264e-06,
+      "loss": 0.6049,
+      "step": 10916
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7684554944540965e-06,
+      "loss": 0.6062,
+      "step": 10917
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.766778594632862e-06,
+      "loss": 0.5847,
+      "step": 10918
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7651019814126655e-06,
+      "loss": 0.6433,
+      "step": 10919
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.763425654870599e-06,
+      "loss": 0.6346,
+      "step": 10920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.761749615083742e-06,
+      "loss": 0.605,
+      "step": 10921
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.76007386212915e-06,
+      "loss": 0.6692,
+      "step": 10922
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.758398396083879e-06,
+      "loss": 0.6093,
+      "step": 10923
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.756723217024964e-06,
+      "loss": 0.6198,
+      "step": 10924
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.75504832502943e-06,
+      "loss": 0.5587,
+      "step": 10925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.753373720174289e-06,
+      "loss": 0.6398,
+      "step": 10926
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.751699402536538e-06,
+      "loss": 0.6072,
+      "step": 10927
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.750025372193164e-06,
+      "loss": 0.6254,
+      "step": 10928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.748351629221132e-06,
+      "loss": 0.5646,
+      "step": 10929
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7466781736974045e-06,
+      "loss": 0.6077,
+      "step": 10930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7450050056989245e-06,
+      "loss": 0.6881,
+      "step": 10931
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.743332125302628e-06,
+      "loss": 0.5947,
+      "step": 10932
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.741659532585426e-06,
+      "loss": 0.6429,
+      "step": 10933
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.739987227624228e-06,
+      "loss": 0.6662,
+      "step": 10934
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.738315210495924e-06,
+      "loss": 0.6364,
+      "step": 10935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.736643481277398e-06,
+      "loss": 0.6111,
+      "step": 10936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.734972040045509e-06,
+      "loss": 0.5345,
+      "step": 10937
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.733300886877111e-06,
+      "loss": 0.6407,
+      "step": 10938
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.731630021849044e-06,
+      "loss": 0.6432,
+      "step": 10939
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.729959445038136e-06,
+      "loss": 0.6525,
+      "step": 10940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7282891565211945e-06,
+      "loss": 0.6401,
+      "step": 10941
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7266191563750196e-06,
+      "loss": 0.5558,
+      "step": 10942
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7249494446764e-06,
+      "loss": 0.6227,
+      "step": 10943
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.723280021502107e-06,
+      "loss": 0.6021,
+      "step": 10944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.721610886928899e-06,
+      "loss": 0.6201,
+      "step": 10945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.719942041033524e-06,
+      "loss": 0.6004,
+      "step": 10946
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.718273483892716e-06,
+      "loss": 0.6387,
+      "step": 10947
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.716605215583191e-06,
+      "loss": 0.6584,
+      "step": 10948
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.714937236181656e-06,
+      "loss": 0.6087,
+      "step": 10949
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7132695457648055e-06,
+      "loss": 0.5692,
+      "step": 10950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.711602144409322e-06,
+      "loss": 0.6404,
+      "step": 10951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.709935032191865e-06,
+      "loss": 0.6438,
+      "step": 10952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.708268209189092e-06,
+      "loss": 0.5904,
+      "step": 10953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.706601675477641e-06,
+      "loss": 0.5663,
+      "step": 10954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.704935431134144e-06,
+      "loss": 0.6481,
+      "step": 10955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.7032694762352066e-06,
+      "loss": 0.6201,
+      "step": 10956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.701603810857432e-06,
+      "loss": 0.607,
+      "step": 10957
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.699938435077407e-06,
+      "loss": 0.6033,
+      "step": 10958
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.698273348971705e-06,
+      "loss": 0.613,
+      "step": 10959
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.696608552616884e-06,
+      "loss": 0.6515,
+      "step": 10960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.694944046089494e-06,
+      "loss": 0.6002,
+      "step": 10961
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.693279829466072e-06,
+      "loss": 0.6169,
+      "step": 10962
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.691615902823127e-06,
+      "loss": 0.5527,
+      "step": 10963
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6899522662371735e-06,
+      "loss": 0.6555,
+      "step": 10964
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.688288919784702e-06,
+      "loss": 0.6442,
+      "step": 10965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.686625863542197e-06,
+      "loss": 0.663,
+      "step": 10966
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.684963097586118e-06,
+      "loss": 0.6246,
+      "step": 10967
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.683300621992922e-06,
+      "loss": 0.6127,
+      "step": 10968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.681638436839049e-06,
+      "loss": 0.6856,
+      "step": 10969
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6799765422009286e-06,
+      "loss": 0.6365,
+      "step": 10970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.678314938154967e-06,
+      "loss": 0.6077,
+      "step": 10971
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.676653624777569e-06,
+      "loss": 0.7133,
+      "step": 10972
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.674992602145121e-06,
+      "loss": 0.6009,
+      "step": 10973
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.673331870333998e-06,
+      "loss": 0.6171,
+      "step": 10974
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.671671429420553e-06,
+      "loss": 0.6412,
+      "step": 10975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.670011279481138e-06,
+      "loss": 0.5985,
+      "step": 10976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.668351420592083e-06,
+      "loss": 0.5859,
+      "step": 10977
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.66669185282971e-06,
+      "loss": 0.644,
+      "step": 10978
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.665032576270326e-06,
+      "loss": 0.665,
+      "step": 10979
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6633735909902205e-06,
+      "loss": 0.5917,
+      "step": 10980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.661714897065678e-06,
+      "loss": 0.6652,
+      "step": 10981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.660056494572959e-06,
+      "loss": 0.6031,
+      "step": 10982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.658398383588319e-06,
+      "loss": 0.6037,
+      "step": 10983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6567405641879945e-06,
+      "loss": 0.6613,
+      "step": 10984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.655083036448219e-06,
+      "loss": 0.5624,
+      "step": 10985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6534258004451975e-06,
+      "loss": 0.6339,
+      "step": 10986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.651768856255128e-06,
+      "loss": 0.6387,
+      "step": 10987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.650112203954202e-06,
+      "loss": 0.7051,
+      "step": 10988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6484558436185935e-06,
+      "loss": 0.6028,
+      "step": 10989
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.646799775324451e-06,
+      "loss": 0.5691,
+      "step": 10990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.645143999147927e-06,
+      "loss": 0.5637,
+      "step": 10991
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.643488515165152e-06,
+      "loss": 0.7023,
+      "step": 10992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.641833323452245e-06,
+      "loss": 0.6719,
+      "step": 10993
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.640178424085311e-06,
+      "loss": 0.6184,
+      "step": 10994
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.638523817140441e-06,
+      "loss": 0.5494,
+      "step": 10995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.636869502693717e-06,
+      "loss": 0.6477,
+      "step": 10996
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.635215480821197e-06,
+      "loss": 0.6148,
+      "step": 10997
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.633561751598936e-06,
+      "loss": 0.5946,
+      "step": 10998
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6319083151029696e-06,
+      "loss": 0.6173,
+      "step": 10999
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.630255171409329e-06,
+      "loss": 0.5768,
+      "step": 11000
+    },
+    {
+      "epoch": 0.73,
+      "eval_text_loss": 0.6551623940467834,
+      "eval_text_runtime": 1.8681,
+      "eval_text_samples_per_second": 39.612,
+      "eval_text_steps_per_second": 1.071,
+      "step": 11000
+    },
+    {
+      "epoch": 0.73,
+      "eval_svit_loss": 0.7722283005714417,
+      "eval_svit_runtime": 6.6708,
+      "eval_svit_samples_per_second": 122.324,
+      "eval_svit_steps_per_second": 1.949,
+      "step": 11000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.628602320594016e-06,
+      "loss": 0.5884,
+      "step": 11001
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.626949762733033e-06,
+      "loss": 0.6239,
+      "step": 11002
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.625297497902362e-06,
+      "loss": 0.6829,
+      "step": 11003
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.623645526177978e-06,
+      "loss": 0.6201,
+      "step": 11004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.62199384763583e-06,
+      "loss": 0.5965,
+      "step": 11005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.620342462351868e-06,
+      "loss": 0.6468,
+      "step": 11006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.618691370402019e-06,
+      "loss": 0.6396,
+      "step": 11007
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.617040571862204e-06,
+      "loss": 0.728,
+      "step": 11008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.615390066808319e-06,
+      "loss": 0.5662,
+      "step": 11009
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.613739855316257e-06,
+      "loss": 0.5386,
+      "step": 11010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.612089937461895e-06,
+      "loss": 0.5991,
+      "step": 11011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.610440313321095e-06,
+      "loss": 0.639,
+      "step": 11012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6087909829697054e-06,
+      "loss": 0.5897,
+      "step": 11013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.607141946483562e-06,
+      "loss": 0.5564,
+      "step": 11014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.605493203938491e-06,
+      "loss": 0.6115,
+      "step": 11015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.603844755410294e-06,
+      "loss": 0.5769,
+      "step": 11016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.602196600974768e-06,
+      "loss": 0.7164,
+      "step": 11017
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.600548740707695e-06,
+      "loss": 0.6048,
+      "step": 11018
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.598901174684849e-06,
+      "loss": 0.6554,
+      "step": 11019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.597253902981974e-06,
+      "loss": 0.6462,
+      "step": 11020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.595606925674816e-06,
+      "loss": 0.6366,
+      "step": 11021
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.593960242839103e-06,
+      "loss": 0.5772,
+      "step": 11022
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.59231385455055e-06,
+      "loss": 0.6002,
+      "step": 11023
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.590667760884853e-06,
+      "loss": 0.6214,
+      "step": 11024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.589021961917701e-06,
+      "loss": 0.6433,
+      "step": 11025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.587376457724766e-06,
+      "loss": 0.6541,
+      "step": 11026
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.58573124838171e-06,
+      "loss": 0.6313,
+      "step": 11027
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.584086333964176e-06,
+      "loss": 0.6889,
+      "step": 11028
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5824417145477995e-06,
+      "loss": 0.6441,
+      "step": 11029
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5807973902082025e-06,
+      "loss": 0.595,
+      "step": 11030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5791533610209814e-06,
+      "loss": 0.6177,
+      "step": 11031
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.577509627061735e-06,
+      "loss": 0.6209,
+      "step": 11032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.575866188406037e-06,
+      "loss": 0.635,
+      "step": 11033
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.574223045129459e-06,
+      "loss": 0.6028,
+      "step": 11034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.572580197307543e-06,
+      "loss": 0.6184,
+      "step": 11035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.570937645015832e-06,
+      "loss": 0.6013,
+      "step": 11036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.569295388329848e-06,
+      "loss": 0.6495,
+      "step": 11037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.567653427325107e-06,
+      "loss": 0.6753,
+      "step": 11038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.566011762077097e-06,
+      "loss": 0.6427,
+      "step": 11039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.564370392661305e-06,
+      "loss": 0.7079,
+      "step": 11040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5627293191532e-06,
+      "loss": 0.6185,
+      "step": 11041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.561088541628238e-06,
+      "loss": 0.6059,
+      "step": 11042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.559448060161863e-06,
+      "loss": 0.625,
+      "step": 11043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.557807874829501e-06,
+      "loss": 0.6692,
+      "step": 11044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.556167985706572e-06,
+      "loss": 0.6299,
+      "step": 11045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.554528392868471e-06,
+      "loss": 0.6183,
+      "step": 11046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.55288909639059e-06,
+      "loss": 0.5811,
+      "step": 11047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5512500963483e-06,
+      "loss": 0.5517,
+      "step": 11048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.549611392816968e-06,
+      "loss": 0.6205,
+      "step": 11049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.547972985871932e-06,
+      "loss": 0.7028,
+      "step": 11050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5463348755885305e-06,
+      "loss": 0.6542,
+      "step": 11051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.544697062042083e-06,
+      "loss": 0.6307,
+      "step": 11052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.543059545307898e-06,
+      "loss": 0.624,
+      "step": 11053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.54142232546126e-06,
+      "loss": 0.6509,
+      "step": 11054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.539785402577453e-06,
+      "loss": 0.6485,
+      "step": 11055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.538148776731744e-06,
+      "loss": 0.5833,
+      "step": 11056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.536512447999383e-06,
+      "loss": 0.5552,
+      "step": 11057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.534876416455603e-06,
+      "loss": 0.6187,
+      "step": 11058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.533240682175632e-06,
+      "loss": 0.5934,
+      "step": 11059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5316052452346815e-06,
+      "loss": 0.5706,
+      "step": 11060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.529970105707946e-06,
+      "loss": 0.6053,
+      "step": 11061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.52833526367061e-06,
+      "loss": 0.5594,
+      "step": 11062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.526700719197842e-06,
+      "loss": 0.5657,
+      "step": 11063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.525066472364802e-06,
+      "loss": 0.6968,
+      "step": 11064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.523432523246624e-06,
+      "loss": 0.5763,
+      "step": 11065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.52179887191844e-06,
+      "loss": 0.6484,
+      "step": 11066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.520165518455366e-06,
+      "loss": 0.6043,
+      "step": 11067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.518532462932505e-06,
+      "loss": 0.5169,
+      "step": 11068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.516899705424939e-06,
+      "loss": 0.634,
+      "step": 11069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.515267246007743e-06,
+      "loss": 0.6353,
+      "step": 11070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5136350847559776e-06,
+      "loss": 0.6096,
+      "step": 11071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5120032217446915e-06,
+      "loss": 0.6721,
+      "step": 11072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.510371657048913e-06,
+      "loss": 0.5834,
+      "step": 11073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.508740390743661e-06,
+      "loss": 0.552,
+      "step": 11074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.507109422903942e-06,
+      "loss": 0.6152,
+      "step": 11075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.505478753604747e-06,
+      "loss": 0.6548,
+      "step": 11076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.503848382921054e-06,
+      "loss": 0.6616,
+      "step": 11077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.502218310927827e-06,
+      "loss": 0.6653,
+      "step": 11078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.500588537700018e-06,
+      "loss": 0.6136,
+      "step": 11079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.498959063312558e-06,
+      "loss": 0.6751,
+      "step": 11080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.497329887840372e-06,
+      "loss": 0.6351,
+      "step": 11081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.49570101135837e-06,
+      "loss": 0.6628,
+      "step": 11082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.494072433941449e-06,
+      "loss": 0.5777,
+      "step": 11083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.492444155664483e-06,
+      "loss": 0.6179,
+      "step": 11084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.4908161766023466e-06,
+      "loss": 0.5976,
+      "step": 11085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.48918849682989e-06,
+      "loss": 0.5912,
+      "step": 11086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.487561116421958e-06,
+      "loss": 0.5861,
+      "step": 11087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.485934035453371e-06,
+      "loss": 0.5908,
+      "step": 11088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.484307253998943e-06,
+      "loss": 0.6422,
+      "step": 11089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.482680772133474e-06,
+      "loss": 0.6389,
+      "step": 11090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.481054589931751e-06,
+      "loss": 0.5525,
+      "step": 11091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.479428707468542e-06,
+      "loss": 0.5468,
+      "step": 11092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.477803124818604e-06,
+      "loss": 0.5953,
+      "step": 11093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.476177842056682e-06,
+      "loss": 0.5879,
+      "step": 11094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.474552859257506e-06,
+      "loss": 0.6748,
+      "step": 11095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.472928176495792e-06,
+      "loss": 0.5763,
+      "step": 11096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.471303793846242e-06,
+      "loss": 0.5715,
+      "step": 11097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.469679711383548e-06,
+      "loss": 0.6403,
+      "step": 11098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.468055929182378e-06,
+      "loss": 0.5685,
+      "step": 11099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.466432447317396e-06,
+      "loss": 0.6038,
+      "step": 11100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.464809265863251e-06,
+      "loss": 0.658,
+      "step": 11101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.4631863848945754e-06,
+      "loss": 0.5911,
+      "step": 11102
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.461563804485985e-06,
+      "loss": 0.6304,
+      "step": 11103
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.459941524712089e-06,
+      "loss": 0.5755,
+      "step": 11104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.458319545647477e-06,
+      "loss": 0.6294,
+      "step": 11105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.456697867366733e-06,
+      "loss": 0.5502,
+      "step": 11106
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.455076489944412e-06,
+      "loss": 0.5509,
+      "step": 11107
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.453455413455069e-06,
+      "loss": 0.6146,
+      "step": 11108
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.45183463797324e-06,
+      "loss": 0.6561,
+      "step": 11109
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.450214163573447e-06,
+      "loss": 0.606,
+      "step": 11110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.448593990330202e-06,
+      "loss": 0.5547,
+      "step": 11111
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.446974118317995e-06,
+      "loss": 0.6093,
+      "step": 11112
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.445354547611314e-06,
+      "loss": 0.6031,
+      "step": 11113
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.443735278284618e-06,
+      "loss": 0.6859,
+      "step": 11114
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.442116310412363e-06,
+      "loss": 0.5794,
+      "step": 11115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.440497644068991e-06,
+      "loss": 0.5778,
+      "step": 11116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.438879279328929e-06,
+      "loss": 0.5887,
+      "step": 11117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.437261216266582e-06,
+      "loss": 0.5848,
+      "step": 11118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.435643454956352e-06,
+      "loss": 0.6013,
+      "step": 11119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.434025995472624e-06,
+      "loss": 0.6717,
+      "step": 11120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.43240883788977e-06,
+      "loss": 0.6499,
+      "step": 11121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.430791982282139e-06,
+      "loss": 0.6086,
+      "step": 11122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4291754287240775e-06,
+      "loss": 0.616,
+      "step": 11123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.427559177289915e-06,
+      "loss": 0.5788,
+      "step": 11124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.425943228053968e-06,
+      "loss": 0.5971,
+      "step": 11125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.424327581090532e-06,
+      "loss": 0.6122,
+      "step": 11126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.422712236473895e-06,
+      "loss": 0.6332,
+      "step": 11127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.421097194278332e-06,
+      "loss": 0.6318,
+      "step": 11128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.419482454578101e-06,
+      "loss": 0.6651,
+      "step": 11129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4178680174474474e-06,
+      "loss": 0.612,
+      "step": 11130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.416253882960602e-06,
+      "loss": 0.6363,
+      "step": 11131
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4146400511917855e-06,
+      "loss": 0.5509,
+      "step": 11132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.413026522215195e-06,
+      "loss": 0.6388,
+      "step": 11133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.411413296105023e-06,
+      "loss": 0.6143,
+      "step": 11134
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.409800372935444e-06,
+      "loss": 0.6034,
+      "step": 11135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4081877527806245e-06,
+      "loss": 0.6404,
+      "step": 11136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.406575435714705e-06,
+      "loss": 0.6889,
+      "step": 11137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.404963421811822e-06,
+      "loss": 0.5579,
+      "step": 11138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.403351711146095e-06,
+      "loss": 0.5804,
+      "step": 11139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.401740303791634e-06,
+      "loss": 0.5724,
+      "step": 11140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.400129199822525e-06,
+      "loss": 0.6061,
+      "step": 11141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.398518399312846e-06,
+      "loss": 0.5383,
+      "step": 11142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.396907902336665e-06,
+      "loss": 0.6937,
+      "step": 11143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.39529770896803e-06,
+      "loss": 0.6488,
+      "step": 11144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3936878192809774e-06,
+      "loss": 0.6545,
+      "step": 11145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.392078233349529e-06,
+      "loss": 0.6113,
+      "step": 11146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3904689512476945e-06,
+      "loss": 0.6211,
+      "step": 11147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.388859973049466e-06,
+      "loss": 0.6159,
+      "step": 11148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.387251298828822e-06,
+      "loss": 0.5479,
+      "step": 11149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3856429286597315e-06,
+      "loss": 0.6,
+      "step": 11150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3840348626161505e-06,
+      "loss": 0.6655,
+      "step": 11151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.382427100772009e-06,
+      "loss": 0.6521,
+      "step": 11152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.380819643201234e-06,
+      "loss": 0.5729,
+      "step": 11153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.379212489977738e-06,
+      "loss": 0.6439,
+      "step": 11154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.377605641175419e-06,
+      "loss": 0.6152,
+      "step": 11155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3759990968681545e-06,
+      "loss": 0.549,
+      "step": 11156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.374392857129813e-06,
+      "loss": 0.566,
+      "step": 11157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.372786922034251e-06,
+      "loss": 0.5696,
+      "step": 11158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3711812916553125e-06,
+      "loss": 0.645,
+      "step": 11159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3695759660668145e-06,
+      "loss": 0.5924,
+      "step": 11160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.367970945342576e-06,
+      "loss": 0.6359,
+      "step": 11161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3663662295563916e-06,
+      "loss": 0.6017,
+      "step": 11162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.364761818782049e-06,
+      "loss": 0.5981,
+      "step": 11163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.363157713093315e-06,
+      "loss": 0.6021,
+      "step": 11164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.361553912563949e-06,
+      "loss": 0.5962,
+      "step": 11165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.359950417267693e-06,
+      "loss": 0.6804,
+      "step": 11166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.358347227278272e-06,
+      "loss": 0.5652,
+      "step": 11167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.356744342669401e-06,
+      "loss": 0.6097,
+      "step": 11168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.355141763514781e-06,
+      "loss": 0.5976,
+      "step": 11169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3535394898881006e-06,
+      "loss": 0.5613,
+      "step": 11170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3519375218630245e-06,
+      "loss": 0.638,
+      "step": 11171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.350335859513216e-06,
+      "loss": 0.6382,
+      "step": 11172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.348734502912317e-06,
+      "loss": 0.6272,
+      "step": 11173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3471334521339615e-06,
+      "loss": 0.6419,
+      "step": 11174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.345532707251759e-06,
+      "loss": 0.568,
+      "step": 11175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.343932268339312e-06,
+      "loss": 0.638,
+      "step": 11176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.34233213547021e-06,
+      "loss": 0.5619,
+      "step": 11177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.340732308718026e-06,
+      "loss": 0.596,
+      "step": 11178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.339132788156322e-06,
+      "loss": 0.6231,
+      "step": 11179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.337533573858639e-06,
+      "loss": 0.6208,
+      "step": 11180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.335934665898513e-06,
+      "loss": 0.5794,
+      "step": 11181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3343360643494564e-06,
+      "loss": 0.6372,
+      "step": 11182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.332737769284972e-06,
+      "loss": 0.5679,
+      "step": 11183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.331139780778552e-06,
+      "loss": 0.5936,
+      "step": 11184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.329542098903674e-06,
+      "loss": 0.5829,
+      "step": 11185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.327944723733792e-06,
+      "loss": 0.6227,
+      "step": 11186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.326347655342356e-06,
+      "loss": 0.6513,
+      "step": 11187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3247508938027965e-06,
+      "loss": 0.6345,
+      "step": 11188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.323154439188538e-06,
+      "loss": 0.5697,
+      "step": 11189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3215582915729776e-06,
+      "loss": 0.6352,
+      "step": 11190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.319962451029507e-06,
+      "loss": 0.6049,
+      "step": 11191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3183669176315045e-06,
+      "loss": 0.6727,
+      "step": 11192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.316771691452335e-06,
+      "loss": 0.7134,
+      "step": 11193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.31517677256534e-06,
+      "loss": 0.6134,
+      "step": 11194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3135821610438545e-06,
+      "loss": 0.5921,
+      "step": 11195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3119878569612e-06,
+      "loss": 0.6359,
+      "step": 11196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.310393860390682e-06,
+      "loss": 0.5711,
+      "step": 11197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.308800171405591e-06,
+      "loss": 0.5131,
+      "step": 11198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.307206790079204e-06,
+      "loss": 0.6104,
+      "step": 11199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.30561371648479e-06,
+      "loss": 0.6177,
+      "step": 11200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.304020950695587e-06,
+      "loss": 0.6196,
+      "step": 11201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.302428492784836e-06,
+      "loss": 0.5468,
+      "step": 11202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.300836342825757e-06,
+      "loss": 0.6096,
+      "step": 11203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.29924450089156e-06,
+      "loss": 0.6075,
+      "step": 11204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.297652967055429e-06,
+      "loss": 0.6348,
+      "step": 11205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2960617413905474e-06,
+      "loss": 0.6403,
+      "step": 11206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.294470823970078e-06,
+      "loss": 0.62,
+      "step": 11207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.292880214867173e-06,
+      "loss": 0.5895,
+      "step": 11208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.291289914154963e-06,
+      "loss": 0.5893,
+      "step": 11209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.289699921906574e-06,
+      "loss": 0.5756,
+      "step": 11210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.28811023819511e-06,
+      "loss": 0.667,
+      "step": 11211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.286520863093665e-06,
+      "loss": 0.6326,
+      "step": 11212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.28493179667532e-06,
+      "loss": 0.6348,
+      "step": 11213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2833430390131365e-06,
+      "loss": 0.623,
+      "step": 11214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.281754590180169e-06,
+      "loss": 0.6185,
+      "step": 11215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2801664502494494e-06,
+      "loss": 0.6252,
+      "step": 11216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.278578619294e-06,
+      "loss": 0.6634,
+      "step": 11217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.276991097386831e-06,
+      "loss": 0.6165,
+      "step": 11218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.275403884600938e-06,
+      "loss": 0.6097,
+      "step": 11219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.273816981009296e-06,
+      "loss": 0.5781,
+      "step": 11220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.272230386684869e-06,
+      "loss": 0.6331,
+      "step": 11221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.270644101700613e-06,
+      "loss": 0.6234,
+      "step": 11222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.269058126129464e-06,
+      "loss": 0.5965,
+      "step": 11223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.267472460044341e-06,
+      "loss": 0.59,
+      "step": 11224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.265887103518152e-06,
+      "loss": 0.5932,
+      "step": 11225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.264302056623795e-06,
+      "loss": 0.6323,
+      "step": 11226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.262717319434149e-06,
+      "loss": 0.6003,
+      "step": 11227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.261132892022078e-06,
+      "loss": 0.68,
+      "step": 11228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.259548774460434e-06,
+      "loss": 0.6052,
+      "step": 11229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.257964966822057e-06,
+      "loss": 0.5491,
+      "step": 11230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.256381469179763e-06,
+      "loss": 0.5524,
+      "step": 11231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.254798281606366e-06,
+      "loss": 0.5675,
+      "step": 11232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.253215404174659e-06,
+      "loss": 0.608,
+      "step": 11233
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.251632836957424e-06,
+      "loss": 0.6454,
+      "step": 11234
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2500505800274225e-06,
+      "loss": 0.5314,
+      "step": 11235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.248468633457409e-06,
+      "loss": 0.6088,
+      "step": 11236
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.24688699732012e-06,
+      "loss": 0.5454,
+      "step": 11237
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2453056716882815e-06,
+      "loss": 0.642,
+      "step": 11238
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.243724656634595e-06,
+      "loss": 0.6366,
+      "step": 11239
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2421439522317604e-06,
+      "loss": 0.5837,
+      "step": 11240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2405635585524565e-06,
+      "loss": 0.5968,
+      "step": 11241
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.238983475669353e-06,
+      "loss": 0.5483,
+      "step": 11242
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.237403703655094e-06,
+      "loss": 0.6433,
+      "step": 11243
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.235824242582322e-06,
+      "loss": 0.6287,
+      "step": 11244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.234245092523658e-06,
+      "loss": 0.5994,
+      "step": 11245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.232666253551712e-06,
+      "loss": 0.6145,
+      "step": 11246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.231087725739076e-06,
+      "loss": 0.6165,
+      "step": 11247
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.229509509158333e-06,
+      "loss": 0.6308,
+      "step": 11248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.227931603882051e-06,
+      "loss": 0.6218,
+      "step": 11249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.226354009982775e-06,
+      "loss": 0.6647,
+      "step": 11250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.224776727533044e-06,
+      "loss": 0.6206,
+      "step": 11251
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.223199756605383e-06,
+      "loss": 0.5819,
+      "step": 11252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.221623097272303e-06,
+      "loss": 0.6529,
+      "step": 11253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.220046749606291e-06,
+      "loss": 0.6564,
+      "step": 11254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.218470713679831e-06,
+      "loss": 0.645,
+      "step": 11255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.216894989565388e-06,
+      "loss": 0.5925,
+      "step": 11256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.215319577335417e-06,
+      "loss": 0.6225,
+      "step": 11257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.213744477062346e-06,
+      "loss": 0.5663,
+      "step": 11258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.212169688818603e-06,
+      "loss": 0.5886,
+      "step": 11259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.210595212676597e-06,
+      "loss": 0.6098,
+      "step": 11260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.209021048708721e-06,
+      "loss": 0.6335,
+      "step": 11261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.207447196987352e-06,
+      "loss": 0.5952,
+      "step": 11262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.205873657584857e-06,
+      "loss": 0.664,
+      "step": 11263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.20430043057359e-06,
+      "loss": 0.6327,
+      "step": 11264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.202727516025882e-06,
+      "loss": 0.6236,
+      "step": 11265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.201154914014056e-06,
+      "loss": 0.6341,
+      "step": 11266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.199582624610421e-06,
+      "loss": 0.5803,
+      "step": 11267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.198010647887272e-06,
+      "loss": 0.5693,
+      "step": 11268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.196438983916884e-06,
+      "loss": 0.5459,
+      "step": 11269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.194867632771523e-06,
+      "loss": 0.656,
+      "step": 11270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.193296594523439e-06,
+      "loss": 0.6569,
+      "step": 11271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.191725869244872e-06,
+      "loss": 0.5931,
+      "step": 11272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1901554570080345e-06,
+      "loss": 0.5823,
+      "step": 11273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.18858535788514e-06,
+      "loss": 0.5678,
+      "step": 11274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.187015571948378e-06,
+      "loss": 0.5517,
+      "step": 11275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1854460992699324e-06,
+      "loss": 0.5888,
+      "step": 11276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.183876939921958e-06,
+      "loss": 0.6039,
+      "step": 11277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1823080939766095e-06,
+      "loss": 0.62,
+      "step": 11278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.180739561506019e-06,
+      "loss": 0.6617,
+      "step": 11279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.179171342582312e-06,
+      "loss": 0.6493,
+      "step": 11280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.17760343727759e-06,
+      "loss": 0.6244,
+      "step": 11281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1760358456639455e-06,
+      "loss": 0.5649,
+      "step": 11282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.174468567813461e-06,
+      "loss": 0.5947,
+      "step": 11283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1729016037981905e-06,
+      "loss": 0.6063,
+      "step": 11284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.171334953690187e-06,
+      "loss": 0.6275,
+      "step": 11285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.169768617561484e-06,
+      "loss": 0.6385,
+      "step": 11286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.168202595484103e-06,
+      "loss": 0.6745,
+      "step": 11287
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1666368875300456e-06,
+      "loss": 0.5969,
+      "step": 11288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.165071493771304e-06,
+      "loss": 0.6121,
+      "step": 11289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.163506414279853e-06,
+      "loss": 0.6334,
+      "step": 11290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.16194164912766e-06,
+      "loss": 0.5163,
+      "step": 11291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.160377198386664e-06,
+      "loss": 0.6201,
+      "step": 11292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.158813062128802e-06,
+      "loss": 0.6084,
+      "step": 11293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.157249240425992e-06,
+      "loss": 0.6685,
+      "step": 11294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.155685733350139e-06,
+      "loss": 0.5869,
+      "step": 11295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.15412254097313e-06,
+      "loss": 0.6369,
+      "step": 11296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.152559663366842e-06,
+      "loss": 0.5848,
+      "step": 11297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.150997100603139e-06,
+      "loss": 0.6171,
+      "step": 11298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1494348527538585e-06,
+      "loss": 0.668,
+      "step": 11299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.147872919890836e-06,
+      "loss": 0.5941,
+      "step": 11300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.14631130208589e-06,
+      "loss": 0.6636,
+      "step": 11301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.144749999410825e-06,
+      "loss": 0.5948,
+      "step": 11302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.143189011937424e-06,
+      "loss": 0.6087,
+      "step": 11303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.141628339737461e-06,
+      "loss": 0.5735,
+      "step": 11304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.140067982882698e-06,
+      "loss": 0.5931,
+      "step": 11305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.138507941444882e-06,
+      "loss": 0.5891,
+      "step": 11306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.136948215495735e-06,
+      "loss": 0.5967,
+      "step": 11307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1353888051069785e-06,
+      "loss": 0.5935,
+      "step": 11308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.13382971035031e-06,
+      "loss": 0.6484,
+      "step": 11309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.132270931297423e-06,
+      "loss": 0.5941,
+      "step": 11310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.130712468019982e-06,
+      "loss": 0.654,
+      "step": 11311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.129154320589645e-06,
+      "loss": 0.6006,
+      "step": 11312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.127596489078059e-06,
+      "loss": 0.6087,
+      "step": 11313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.126038973556851e-06,
+      "loss": 0.6515,
+      "step": 11314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.124481774097634e-06,
+      "loss": 0.6912,
+      "step": 11315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.122924890772008e-06,
+      "loss": 0.6322,
+      "step": 11316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.121368323651561e-06,
+      "loss": 0.5666,
+      "step": 11317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.119812072807857e-06,
+      "loss": 0.553,
+      "step": 11318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.118256138312457e-06,
+      "loss": 0.6039,
+      "step": 11319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.116700520236898e-06,
+      "loss": 0.5758,
+      "step": 11320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.115145218652713e-06,
+      "loss": 0.6179,
+      "step": 11321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.113590233631407e-06,
+      "loss": 0.5704,
+      "step": 11322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.112035565244481e-06,
+      "loss": 0.6213,
+      "step": 11323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1104812135634175e-06,
+      "loss": 0.6175,
+      "step": 11324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.108927178659688e-06,
+      "loss": 0.6257,
+      "step": 11325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.107373460604742e-06,
+      "loss": 0.6236,
+      "step": 11326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.10582005947002e-06,
+      "loss": 0.6666,
+      "step": 11327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.104266975326946e-06,
+      "loss": 0.573,
+      "step": 11328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.102714208246933e-06,
+      "loss": 0.6534,
+      "step": 11329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.101161758301374e-06,
+      "loss": 0.6266,
+      "step": 11330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.099609625561652e-06,
+      "loss": 0.5784,
+      "step": 11331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0980578100991355e-06,
+      "loss": 0.6353,
+      "step": 11332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0965063119851696e-06,
+      "loss": 0.5991,
+      "step": 11333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.094955131291096e-06,
+      "loss": 0.5942,
+      "step": 11334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.093404268088237e-06,
+      "loss": 0.5434,
+      "step": 11335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.091853722447903e-06,
+      "loss": 0.5964,
+      "step": 11336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0903034944413825e-06,
+      "loss": 0.5713,
+      "step": 11337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.088753584139956e-06,
+      "loss": 0.6495,
+      "step": 11338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.087203991614889e-06,
+      "loss": 0.6699,
+      "step": 11339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.085654716937433e-06,
+      "loss": 0.626,
+      "step": 11340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.084105760178819e-06,
+      "loss": 0.5769,
+      "step": 11341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.082557121410267e-06,
+      "loss": 0.5879,
+      "step": 11342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.081008800702988e-06,
+      "loss": 0.6021,
+      "step": 11343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.079460798128171e-06,
+      "loss": 0.5992,
+      "step": 11344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.077913113756989e-06,
+      "loss": 0.6177,
+      "step": 11345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.076365747660607e-06,
+      "loss": 0.6449,
+      "step": 11346
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.074818699910172e-06,
+      "loss": 0.6316,
+      "step": 11347
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.073271970576816e-06,
+      "loss": 0.6376,
+      "step": 11348
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.071725559731656e-06,
+      "loss": 0.652,
+      "step": 11349
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.070179467445798e-06,
+      "loss": 0.6321,
+      "step": 11350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.068633693790333e-06,
+      "loss": 0.5765,
+      "step": 11351
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.067088238836328e-06,
+      "loss": 0.6111,
+      "step": 11352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.065543102654847e-06,
+      "loss": 0.6111,
+      "step": 11353
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.063998285316932e-06,
+      "loss": 0.5821,
+      "step": 11354
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.06245378689362e-06,
+      "loss": 0.5168,
+      "step": 11355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.060909607455916e-06,
+      "loss": 0.6935,
+      "step": 11356
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.059365747074827e-06,
+      "loss": 0.514,
+      "step": 11357
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.057822205821338e-06,
+      "loss": 0.6219,
+      "step": 11358
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.056278983766424e-06,
+      "loss": 0.6378,
+      "step": 11359
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.054736080981035e-06,
+      "loss": 0.5994,
+      "step": 11360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0531934975361166e-06,
+      "loss": 0.6846,
+      "step": 11361
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0516512335025945e-06,
+      "loss": 0.643,
+      "step": 11362
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.050109288951384e-06,
+      "loss": 0.6714,
+      "step": 11363
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.048567663953382e-06,
+      "loss": 0.5973,
+      "step": 11364
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.04702635857947e-06,
+      "loss": 0.6149,
+      "step": 11365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.045485372900522e-06,
+      "loss": 0.6607,
+      "step": 11366
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.043944706987386e-06,
+      "loss": 0.6116,
+      "step": 11367
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.042404360910903e-06,
+      "loss": 0.6275,
+      "step": 11368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.040864334741899e-06,
+      "loss": 0.5405,
+      "step": 11369
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.039324628551183e-06,
+      "loss": 0.6062,
+      "step": 11370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.037785242409549e-06,
+      "loss": 0.6688,
+      "step": 11371
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0362461763877776e-06,
+      "loss": 0.6212,
+      "step": 11372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.034707430556634e-06,
+      "loss": 0.5922,
+      "step": 11373
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.033169004986873e-06,
+      "loss": 0.6374,
+      "step": 11374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.031630899749226e-06,
+      "loss": 0.5791,
+      "step": 11375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.030093114914416e-06,
+      "loss": 0.6037,
+      "step": 11376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.02855565055315e-06,
+      "loss": 0.6586,
+      "step": 11377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.027018506736124e-06,
+      "loss": 0.574,
+      "step": 11378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.025481683534006e-06,
+      "loss": 0.6243,
+      "step": 11379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.023945181017466e-06,
+      "loss": 0.589,
+      "step": 11380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.022408999257148e-06,
+      "loss": 0.6078,
+      "step": 11381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.020873138323688e-06,
+      "loss": 0.5789,
+      "step": 11382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.019337598287701e-06,
+      "loss": 0.6239,
+      "step": 11383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.017802379219794e-06,
+      "loss": 0.5876,
+      "step": 11384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.016267481190557e-06,
+      "loss": 0.6,
+      "step": 11385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.014732904270559e-06,
+      "loss": 0.585,
+      "step": 11386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.01319864853036e-06,
+      "loss": 0.6307,
+      "step": 11387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.011664714040506e-06,
+      "loss": 0.5945,
+      "step": 11388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.01013110087153e-06,
+      "loss": 0.592,
+      "step": 11389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.00859780909394e-06,
+      "loss": 0.5747,
+      "step": 11390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.007064838778241e-06,
+      "loss": 0.6342,
+      "step": 11391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.005532189994917e-06,
+      "loss": 0.5759,
+      "step": 11392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0039998628144395e-06,
+      "loss": 0.5515,
+      "step": 11393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.002467857307262e-06,
+      "loss": 0.6068,
+      "step": 11394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.000936173543827e-06,
+      "loss": 0.672,
+      "step": 11395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.999404811594559e-06,
+      "loss": 0.6501,
+      "step": 11396
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997873771529873e-06,
+      "loss": 0.6523,
+      "step": 11397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.996343053420163e-06,
+      "loss": 0.5785,
+      "step": 11398
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.994812657335812e-06,
+      "loss": 0.6329,
+      "step": 11399
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.993282583347188e-06,
+      "loss": 0.6185,
+      "step": 11400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9917528315246385e-06,
+      "loss": 0.6045,
+      "step": 11401
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9902234019385055e-06,
+      "loss": 0.6552,
+      "step": 11402
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.98869429465911e-06,
+      "loss": 0.5528,
+      "step": 11403
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.987165509756762e-06,
+      "loss": 0.5867,
+      "step": 11404
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.98563704730175e-06,
+      "loss": 0.5957,
+      "step": 11405
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.984108907364353e-06,
+      "loss": 0.6233,
+      "step": 11406
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.982581090014838e-06,
+      "loss": 0.5768,
+      "step": 11407
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.981053595323451e-06,
+      "loss": 0.5448,
+      "step": 11408
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.979526423360426e-06,
+      "loss": 0.5922,
+      "step": 11409
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.977999574195979e-06,
+      "loss": 0.6635,
+      "step": 11410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.976473047900316e-06,
+      "loss": 0.6152,
+      "step": 11411
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.974946844543629e-06,
+      "loss": 0.6706,
+      "step": 11412
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.973420964196088e-06,
+      "loss": 0.5819,
+      "step": 11413
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.971895406927854e-06,
+      "loss": 0.6137,
+      "step": 11414
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.970370172809074e-06,
+      "loss": 0.5998,
+      "step": 11415
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.968845261909873e-06,
+      "loss": 0.5811,
+      "step": 11416
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.967320674300368e-06,
+      "loss": 0.5421,
+      "step": 11417
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.965796410050657e-06,
+      "loss": 0.6974,
+      "step": 11418
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.964272469230833e-06,
+      "loss": 0.6279,
+      "step": 11419
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9627488519109545e-06,
+      "loss": 0.6374,
+      "step": 11420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.961225558161084e-06,
+      "loss": 0.5687,
+      "step": 11421
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.95970258805126e-06,
+      "loss": 0.6051,
+      "step": 11422
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.958179941651512e-06,
+      "loss": 0.5595,
+      "step": 11423
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9566576190318425e-06,
+      "loss": 0.6287,
+      "step": 11424
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.955135620262254e-06,
+      "loss": 0.6011,
+      "step": 11425
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.953613945412724e-06,
+      "loss": 0.6477,
+      "step": 11426
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.952092594553223e-06,
+      "loss": 0.63,
+      "step": 11427
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.950571567753696e-06,
+      "loss": 0.5711,
+      "step": 11428
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.949050865084082e-06,
+      "loss": 0.6346,
+      "step": 11429
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9475304866143025e-06,
+      "loss": 0.603,
+      "step": 11430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.946010432414265e-06,
+      "loss": 0.5758,
+      "step": 11431
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.944490702553858e-06,
+      "loss": 0.6061,
+      "step": 11432
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.942971297102961e-06,
+      "loss": 0.6434,
+      "step": 11433
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.941452216131438e-06,
+      "loss": 0.5248,
+      "step": 11434
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.939933459709128e-06,
+      "loss": 0.6743,
+      "step": 11435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9384150279058674e-06,
+      "loss": 0.6058,
+      "step": 11436
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.936896920791473e-06,
+      "loss": 0.6285,
+      "step": 11437
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.93537913843575e-06,
+      "loss": 0.5492,
+      "step": 11438
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.933861680908479e-06,
+      "loss": 0.622,
+      "step": 11439
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.932344548279435e-06,
+      "loss": 0.5789,
+      "step": 11440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.930827740618374e-06,
+      "loss": 0.6486,
+      "step": 11441
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.929311257995044e-06,
+      "loss": 0.6073,
+      "step": 11442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.927795100479165e-06,
+      "loss": 0.6076,
+      "step": 11443
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.926279268140451e-06,
+      "loss": 0.5956,
+      "step": 11444
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9247637610486e-06,
+      "loss": 0.6143,
+      "step": 11445
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.923248579273295e-06,
+      "loss": 0.6795,
+      "step": 11446
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.921733722884205e-06,
+      "loss": 0.6347,
+      "step": 11447
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.920219191950979e-06,
+      "loss": 0.6678,
+      "step": 11448
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.918704986543259e-06,
+      "loss": 0.5387,
+      "step": 11449
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.917191106730664e-06,
+      "loss": 0.6132,
+      "step": 11450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9156775525828e-06,
+      "loss": 0.6083,
+      "step": 11451
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.914164324169263e-06,
+      "loss": 0.5664,
+      "step": 11452
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.912651421559633e-06,
+      "loss": 0.6881,
+      "step": 11453
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.911138844823467e-06,
+      "loss": 0.5874,
+      "step": 11454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.909626594030314e-06,
+      "loss": 0.6384,
+      "step": 11455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.908114669249707e-06,
+      "loss": 0.5587,
+      "step": 11456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9066030705511695e-06,
+      "loss": 0.6505,
+      "step": 11457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.905091798004195e-06,
+      "loss": 0.6282,
+      "step": 11458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.903580851678274e-06,
+      "loss": 0.6107,
+      "step": 11459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.902070231642883e-06,
+      "loss": 0.5821,
+      "step": 11460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.90055993796748e-06,
+      "loss": 0.5396,
+      "step": 11461
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8990499707215e-06,
+      "loss": 0.6495,
+      "step": 11462
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.897540329974378e-06,
+      "loss": 0.5871,
+      "step": 11463
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.896031015795523e-06,
+      "loss": 0.6023,
+      "step": 11464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8945220282543345e-06,
+      "loss": 0.616,
+      "step": 11465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.893013367420195e-06,
+      "loss": 0.6362,
+      "step": 11466
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8915050333624696e-06,
+      "loss": 0.6347,
+      "step": 11467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.889997026150518e-06,
+      "loss": 0.5653,
+      "step": 11468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.88848934585367e-06,
+      "loss": 0.6775,
+      "step": 11469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.88698199254125e-06,
+      "loss": 0.6186,
+      "step": 11470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.885474966282567e-06,
+      "loss": 0.5465,
+      "step": 11471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.883968267146916e-06,
+      "loss": 0.6133,
+      "step": 11472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.882461895203568e-06,
+      "loss": 0.6139,
+      "step": 11473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.880955850521789e-06,
+      "loss": 0.6253,
+      "step": 11474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.879450133170827e-06,
+      "loss": 0.5864,
+      "step": 11475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.877944743219915e-06,
+      "loss": 0.5859,
+      "step": 11476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.876439680738267e-06,
+      "loss": 0.6393,
+      "step": 11477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.874934945795086e-06,
+      "loss": 0.6314,
+      "step": 11478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.87343053845956e-06,
+      "loss": 0.6263,
+      "step": 11479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.87192645880086e-06,
+      "loss": 0.584,
+      "step": 11480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8704227068881445e-06,
+      "loss": 0.5942,
+      "step": 11481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.868919282790556e-06,
+      "loss": 0.5981,
+      "step": 11482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.867416186577222e-06,
+      "loss": 0.6176,
+      "step": 11483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.86591341831725e-06,
+      "loss": 0.5658,
+      "step": 11484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.864410978079739e-06,
+      "loss": 0.6428,
+      "step": 11485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.862908865933772e-06,
+      "loss": 0.5817,
+      "step": 11486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.861407081948415e-06,
+      "loss": 0.5888,
+      "step": 11487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.859905626192716e-06,
+      "loss": 0.5843,
+      "step": 11488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.858404498735715e-06,
+      "loss": 0.5475,
+      "step": 11489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.856903699646433e-06,
+      "loss": 0.6098,
+      "step": 11490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8554032289938775e-06,
+      "loss": 0.5878,
+      "step": 11491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.853903086847034e-06,
+      "loss": 0.627,
+      "step": 11492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8524032732748815e-06,
+      "loss": 0.6319,
+      "step": 11493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8509037883463825e-06,
+      "loss": 0.5934,
+      "step": 11494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.849404632130484e-06,
+      "loss": 0.6166,
+      "step": 11495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.847905804696111e-06,
+      "loss": 0.5848,
+      "step": 11496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.846407306112182e-06,
+      "loss": 0.6147,
+      "step": 11497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.844909136447598e-06,
+      "loss": 0.6546,
+      "step": 11498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.843411295771244e-06,
+      "loss": 0.6094,
+      "step": 11499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.841913784151988e-06,
+      "loss": 0.6286,
+      "step": 11500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.840416601658689e-06,
+      "loss": 0.6933,
+      "step": 11501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.838919748360186e-06,
+      "loss": 0.6548,
+      "step": 11502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8374232243253e-06,
+      "loss": 0.5635,
+      "step": 11503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.835927029622842e-06,
+      "loss": 0.6049,
+      "step": 11504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.834431164321608e-06,
+      "loss": 0.6158,
+      "step": 11505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.832935628490381e-06,
+      "loss": 0.5663,
+      "step": 11506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.831440422197916e-06,
+      "loss": 0.6042,
+      "step": 11507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.829945545512968e-06,
+      "loss": 0.6197,
+      "step": 11508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.828450998504268e-06,
+      "loss": 0.5943,
+      "step": 11509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.826956781240541e-06,
+      "loss": 0.6437,
+      "step": 11510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.825462893790481e-06,
+      "loss": 0.6446,
+      "step": 11511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.823969336222783e-06,
+      "loss": 0.6409,
+      "step": 11512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.822476108606118e-06,
+      "loss": 0.6525,
+      "step": 11513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.820983211009144e-06,
+      "loss": 0.6001,
+      "step": 11514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.819490643500504e-06,
+      "loss": 0.584,
+      "step": 11515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.817998406148827e-06,
+      "loss": 0.6087,
+      "step": 11516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.816506499022725e-06,
+      "loss": 0.5861,
+      "step": 11517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.815014922190793e-06,
+      "loss": 0.5917,
+      "step": 11518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8135236757216155e-06,
+      "loss": 0.6847,
+      "step": 11519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.812032759683759e-06,
+      "loss": 0.613,
+      "step": 11520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.810542174145776e-06,
+      "loss": 0.643,
+      "step": 11521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809051919176199e-06,
+      "loss": 0.6471,
+      "step": 11522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.807561994843553e-06,
+      "loss": 0.653,
+      "step": 11523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806072401216343e-06,
+      "loss": 0.5198,
+      "step": 11524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.804583138363064e-06,
+      "loss": 0.6923,
+      "step": 11525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.803094206352185e-06,
+      "loss": 0.5394,
+      "step": 11526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.801605605252168e-06,
+      "loss": 0.6895,
+      "step": 11527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.800117335131462e-06,
+      "loss": 0.6191,
+      "step": 11528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7986293960584965e-06,
+      "loss": 0.6176,
+      "step": 11529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.797141788101683e-06,
+      "loss": 0.5793,
+      "step": 11530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.795654511329421e-06,
+      "loss": 0.7023,
+      "step": 11531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.794167565810098e-06,
+      "loss": 0.6194,
+      "step": 11532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.792680951612081e-06,
+      "loss": 0.6081,
+      "step": 11533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.791194668803726e-06,
+      "loss": 0.6599,
+      "step": 11534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.78970871745337e-06,
+      "loss": 0.5667,
+      "step": 11535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.788223097629338e-06,
+      "loss": 0.5966,
+      "step": 11536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.786737809399937e-06,
+      "loss": 0.6015,
+      "step": 11537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.785252852833459e-06,
+      "loss": 0.585,
+      "step": 11538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.783768227998183e-06,
+      "loss": 0.6795,
+      "step": 11539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.782283934962374e-06,
+      "loss": 0.5293,
+      "step": 11540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7807999737942745e-06,
+      "loss": 0.5733,
+      "step": 11541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7793163445621175e-06,
+      "loss": 0.6296,
+      "step": 11542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.777833047334121e-06,
+      "loss": 0.6716,
+      "step": 11543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.776350082178488e-06,
+      "loss": 0.576,
+      "step": 11544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.774867449163401e-06,
+      "loss": 0.602,
+      "step": 11545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.773385148357032e-06,
+      "loss": 0.5764,
+      "step": 11546
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.771903179827537e-06,
+      "loss": 0.5907,
+      "step": 11547
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.770421543643057e-06,
+      "loss": 0.6292,
+      "step": 11548
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7689402398717165e-06,
+      "loss": 0.6388,
+      "step": 11549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.767459268581624e-06,
+      "loss": 0.5477,
+      "step": 11550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.76597862984088e-06,
+      "loss": 0.5745,
+      "step": 11551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.764498323717555e-06,
+      "loss": 0.6063,
+      "step": 11552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.763018350279718e-06,
+      "loss": 0.5782,
+      "step": 11553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7615387095954165e-06,
+      "loss": 0.5467,
+      "step": 11554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.760059401732685e-06,
+      "loss": 0.614,
+      "step": 11555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.758580426759538e-06,
+      "loss": 0.5648,
+      "step": 11556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.757101784743981e-06,
+      "loss": 0.5392,
+      "step": 11557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.755623475753999e-06,
+      "loss": 0.62,
+      "step": 11558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.75414549985757e-06,
+      "loss": 0.5745,
+      "step": 11559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.752667857122644e-06,
+      "loss": 0.6039,
+      "step": 11560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7511905476171635e-06,
+      "loss": 0.596,
+      "step": 11561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.749713571409058e-06,
+      "loss": 0.5892,
+      "step": 11562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.748236928566238e-06,
+      "loss": 0.5613,
+      "step": 11563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.746760619156596e-06,
+      "loss": 0.6036,
+      "step": 11564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.745284643248011e-06,
+      "loss": 0.5907,
+      "step": 11565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.743809000908351e-06,
+      "loss": 0.6434,
+      "step": 11566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.742333692205466e-06,
+      "loss": 0.6355,
+      "step": 11567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.740858717207187e-06,
+      "loss": 0.54,
+      "step": 11568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.739384075981335e-06,
+      "loss": 0.5949,
+      "step": 11569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.737909768595716e-06,
+      "loss": 0.5334,
+      "step": 11570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.736435795118113e-06,
+      "loss": 0.6613,
+      "step": 11571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.734962155616299e-06,
+      "loss": 0.6969,
+      "step": 11572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.733488850158035e-06,
+      "loss": 0.6024,
+      "step": 11573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.732015878811063e-06,
+      "loss": 0.6502,
+      "step": 11574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.730543241643104e-06,
+      "loss": 0.6297,
+      "step": 11575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.729070938721873e-06,
+      "loss": 0.6246,
+      "step": 11576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.727598970115068e-06,
+      "loss": 0.6614,
+      "step": 11577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.726127335890369e-06,
+      "loss": 0.6119,
+      "step": 11578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.724656036115437e-06,
+      "loss": 0.6389,
+      "step": 11579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.723185070857924e-06,
+      "loss": 0.5244,
+      "step": 11580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7217144401854645e-06,
+      "loss": 0.5276,
+      "step": 11581
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.720244144165678e-06,
+      "loss": 0.6172,
+      "step": 11582
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7187741828661685e-06,
+      "loss": 0.5445,
+      "step": 11583
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.717304556354523e-06,
+      "loss": 0.6671,
+      "step": 11584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7158352646983175e-06,
+      "loss": 0.6038,
+      "step": 11585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.714366307965105e-06,
+      "loss": 0.6886,
+      "step": 11586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.712897686222429e-06,
+      "loss": 0.6446,
+      "step": 11587
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.711429399537816e-06,
+      "loss": 0.6121,
+      "step": 11588
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7099614479787815e-06,
+      "loss": 0.5523,
+      "step": 11589
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.708493831612815e-06,
+      "loss": 0.6391,
+      "step": 11590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.707026550507399e-06,
+      "loss": 0.6106,
+      "step": 11591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.705559604729999e-06,
+      "loss": 0.641,
+      "step": 11592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.704092994348068e-06,
+      "loss": 0.6287,
+      "step": 11593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.702626719429034e-06,
+      "loss": 0.6329,
+      "step": 11594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.701160780040318e-06,
+      "loss": 0.6816,
+      "step": 11595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6996951762493245e-06,
+      "loss": 0.6074,
+      "step": 11596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.69822990812344e-06,
+      "loss": 0.6193,
+      "step": 11597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.696764975730038e-06,
+      "loss": 0.5979,
+      "step": 11598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.695300379136475e-06,
+      "loss": 0.624,
+      "step": 11599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.693836118410097e-06,
+      "loss": 0.6112,
+      "step": 11600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.692372193618221e-06,
+      "loss": 0.6005,
+      "step": 11601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.690908604828164e-06,
+      "loss": 0.5635,
+      "step": 11602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.68944535210722e-06,
+      "loss": 0.6559,
+      "step": 11603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.68798243552267e-06,
+      "loss": 0.6008,
+      "step": 11604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.686519855141775e-06,
+      "loss": 0.6001,
+      "step": 11605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6850576110317865e-06,
+      "loss": 0.5851,
+      "step": 11606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.683595703259936e-06,
+      "loss": 0.5669,
+      "step": 11607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.682134131893445e-06,
+      "loss": 0.5787,
+      "step": 11608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6806728969995114e-06,
+      "loss": 0.6217,
+      "step": 11609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6792119986453245e-06,
+      "loss": 0.6645,
+      "step": 11610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6777514368980565e-06,
+      "loss": 0.6314,
+      "step": 11611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.676291211824865e-06,
+      "loss": 0.6337,
+      "step": 11612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.674831323492885e-06,
+      "loss": 0.6146,
+      "step": 11613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.673371771969244e-06,
+      "loss": 0.6087,
+      "step": 11614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.671912557321053e-06,
+      "loss": 0.5975,
+      "step": 11615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6704536796154065e-06,
+      "loss": 0.5897,
+      "step": 11616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.668995138919381e-06,
+      "loss": 0.562,
+      "step": 11617
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.667536935300042e-06,
+      "loss": 0.5788,
+      "step": 11618
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6660790688244375e-06,
+      "loss": 0.6061,
+      "step": 11619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.664621539559596e-06,
+      "loss": 0.5445,
+      "step": 11620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.663164347572536e-06,
+      "loss": 0.6406,
+      "step": 11621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.661707492930258e-06,
+      "loss": 0.6213,
+      "step": 11622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.660250975699752e-06,
+      "loss": 0.606,
+      "step": 11623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.658794795947981e-06,
+      "loss": 0.6673,
+      "step": 11624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.657338953741903e-06,
+      "loss": 0.5811,
+      "step": 11625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.655883449148457e-06,
+      "loss": 0.6142,
+      "step": 11626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.65442828223457e-06,
+      "loss": 0.6463,
+      "step": 11627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.652973453067143e-06,
+      "loss": 0.5681,
+      "step": 11628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.651518961713072e-06,
+      "loss": 0.588,
+      "step": 11629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.650064808239233e-06,
+      "loss": 0.6619,
+      "step": 11630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.64861099271249e-06,
+      "loss": 0.5221,
+      "step": 11631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.647157515199687e-06,
+      "loss": 0.5923,
+      "step": 11632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.645704375767654e-06,
+      "loss": 0.6471,
+      "step": 11633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.644251574483208e-06,
+      "loss": 0.5701,
+      "step": 11634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.642799111413145e-06,
+      "loss": 0.6395,
+      "step": 11635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.64134698662425e-06,
+      "loss": 0.5899,
+      "step": 11636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.639895200183292e-06,
+      "loss": 0.5028,
+      "step": 11637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6384437521570245e-06,
+      "loss": 0.6144,
+      "step": 11638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.63699264261218e-06,
+      "loss": 0.6258,
+      "step": 11639
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.635541871615484e-06,
+      "loss": 0.6073,
+      "step": 11640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.63409143923364e-06,
+      "loss": 0.6615,
+      "step": 11641
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.632641345533343e-06,
+      "loss": 0.6354,
+      "step": 11642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.631191590581261e-06,
+      "loss": 0.6267,
+      "step": 11643
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.629742174444056e-06,
+      "loss": 0.6181,
+      "step": 11644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.628293097188373e-06,
+      "loss": 0.6178,
+      "step": 11645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.626844358880841e-06,
+      "loss": 0.6007,
+      "step": 11646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.625395959588067e-06,
+      "loss": 0.6921,
+      "step": 11647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.623947899376652e-06,
+      "loss": 0.5911,
+      "step": 11648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.622500178313177e-06,
+      "loss": 0.6522,
+      "step": 11649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.621052796464206e-06,
+      "loss": 0.5543,
+      "step": 11650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.61960575389629e-06,
+      "loss": 0.6126,
+      "step": 11651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.618159050675964e-06,
+      "loss": 0.6351,
+      "step": 11652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.616712686869749e-06,
+      "loss": 0.584,
+      "step": 11653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6152666625441425e-06,
+      "loss": 0.6038,
+      "step": 11654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.613820977765635e-06,
+      "loss": 0.6013,
+      "step": 11655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6123756326007e-06,
+      "loss": 0.6707,
+      "step": 11656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.610930627115794e-06,
+      "loss": 0.6067,
+      "step": 11657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.609485961377355e-06,
+      "loss": 0.6503,
+      "step": 11658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6080416354518075e-06,
+      "loss": 0.6275,
+      "step": 11659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.606597649405564e-06,
+      "loss": 0.5905,
+      "step": 11660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.605154003305021e-06,
+      "loss": 0.6243,
+      "step": 11661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6037106972165506e-06,
+      "loss": 0.658,
+      "step": 11662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6022677312065175e-06,
+      "loss": 0.656,
+      "step": 11663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6008251053412695e-06,
+      "loss": 0.5804,
+      "step": 11664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.599382819687137e-06,
+      "loss": 0.5739,
+      "step": 11665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.597940874310437e-06,
+      "loss": 0.6723,
+      "step": 11666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.596499269277469e-06,
+      "loss": 0.6103,
+      "step": 11667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.59505800465452e-06,
+      "loss": 0.567,
+      "step": 11668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5936170805078535e-06,
+      "loss": 0.6027,
+      "step": 11669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.592176496903727e-06,
+      "loss": 0.6628,
+      "step": 11670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.590736253908375e-06,
+      "loss": 0.6273,
+      "step": 11671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.589296351588023e-06,
+      "loss": 0.5962,
+      "step": 11672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5878567900088726e-06,
+      "loss": 0.5709,
+      "step": 11673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.586417569237116e-06,
+      "loss": 0.5651,
+      "step": 11674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5849786893389295e-06,
+      "loss": 0.6078,
+      "step": 11675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.583540150380472e-06,
+      "loss": 0.55,
+      "step": 11676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5821019524278855e-06,
+      "loss": 0.5485,
+      "step": 11677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.580664095547297e-06,
+      "loss": 0.557,
+      "step": 11678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5792265798048195e-06,
+      "loss": 0.6106,
+      "step": 11679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5777894052665545e-06,
+      "loss": 0.5871,
+      "step": 11680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.576352571998574e-06,
+      "loss": 0.6214,
+      "step": 11681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.574916080066947e-06,
+      "loss": 0.6928,
+      "step": 11682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.573479929537723e-06,
+      "loss": 0.6372,
+      "step": 11683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.572044120476937e-06,
+      "loss": 0.5442,
+      "step": 11684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5706086529506034e-06,
+      "loss": 0.5738,
+      "step": 11685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.569173527024728e-06,
+      "loss": 0.6403,
+      "step": 11686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.567738742765298e-06,
+      "loss": 0.5994,
+      "step": 11687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5663043002382786e-06,
+      "loss": 0.6719,
+      "step": 11688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.564870199509629e-06,
+      "loss": 0.629,
+      "step": 11689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.563436440645289e-06,
+      "loss": 0.6355,
+      "step": 11690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.562003023711183e-06,
+      "loss": 0.5682,
+      "step": 11691
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.560569948773214e-06,
+      "loss": 0.6094,
+      "step": 11692
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5591372158972765e-06,
+      "loss": 0.6139,
+      "step": 11693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.557704825149249e-06,
+      "loss": 0.5886,
+      "step": 11694
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.556272776594994e-06,
+      "loss": 0.5957,
+      "step": 11695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.554841070300351e-06,
+      "loss": 0.6476,
+      "step": 11696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5534097063311515e-06,
+      "loss": 0.5981,
+      "step": 11697
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.55197868475321e-06,
+      "loss": 0.6281,
+      "step": 11698
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.550548005632321e-06,
+      "loss": 0.6897,
+      "step": 11699
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5491176690342725e-06,
+      "loss": 0.5485,
+      "step": 11700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.547687675024826e-06,
+      "loss": 0.587,
+      "step": 11701
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.546258023669737e-06,
+      "loss": 0.5547,
+      "step": 11702
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.544828715034734e-06,
+      "loss": 0.7021,
+      "step": 11703
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.543399749185539e-06,
+      "loss": 0.6343,
+      "step": 11704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.541971126187855e-06,
+      "loss": 0.6976,
+      "step": 11705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.540542846107373e-06,
+      "loss": 0.5716,
+      "step": 11706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.539114909009759e-06,
+      "loss": 0.6095,
+      "step": 11707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.537687314960672e-06,
+      "loss": 0.5503,
+      "step": 11708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.536260064025751e-06,
+      "loss": 0.6924,
+      "step": 11709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5348331562706225e-06,
+      "loss": 0.5591,
+      "step": 11710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.533406591760893e-06,
+      "loss": 0.6138,
+      "step": 11711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.531980370562156e-06,
+      "loss": 0.6016,
+      "step": 11712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.530554492739988e-06,
+      "loss": 0.6165,
+      "step": 11713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.529128958359954e-06,
+      "loss": 0.6489,
+      "step": 11714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.527703767487594e-06,
+      "loss": 0.6026,
+      "step": 11715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.526278920188441e-06,
+      "loss": 0.5611,
+      "step": 11716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.524854416528007e-06,
+      "loss": 0.5841,
+      "step": 11717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5234302565717915e-06,
+      "loss": 0.6068,
+      "step": 11718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.522006440385276e-06,
+      "loss": 0.6171,
+      "step": 11719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.520582968033929e-06,
+      "loss": 0.5827,
+      "step": 11720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.519159839583201e-06,
+      "loss": 0.6333,
+      "step": 11721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5177370550985246e-06,
+      "loss": 0.624,
+      "step": 11722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.516314614645319e-06,
+      "loss": 0.5856,
+      "step": 11723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.514892518288988e-06,
+      "loss": 0.612,
+      "step": 11724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.513470766094923e-06,
+      "loss": 0.5998,
+      "step": 11725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.51204935812849e-06,
+      "loss": 0.6178,
+      "step": 11726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.510628294455045e-06,
+      "loss": 0.5421,
+      "step": 11727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.50920757513993e-06,
+      "loss": 0.5749,
+      "step": 11728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5077872002484725e-06,
+      "loss": 0.5888,
+      "step": 11729
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.506367169845974e-06,
+      "loss": 0.585,
+      "step": 11730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.50494748399773e-06,
+      "loss": 0.6195,
+      "step": 11731
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.503528142769017e-06,
+      "loss": 0.6454,
+      "step": 11732
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.502109146225095e-06,
+      "loss": 0.572,
+      "step": 11733
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.500690494431211e-06,
+      "loss": 0.6316,
+      "step": 11734
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.499272187452592e-06,
+      "loss": 0.6283,
+      "step": 11735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.497854225354456e-06,
+      "loss": 0.6309,
+      "step": 11736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4964366082019924e-06,
+      "loss": 0.5552,
+      "step": 11737
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.495019336060387e-06,
+      "loss": 0.6508,
+      "step": 11738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4936024089948045e-06,
+      "loss": 0.5653,
+      "step": 11739
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.492185827070399e-06,
+      "loss": 0.6087,
+      "step": 11740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.490769590352297e-06,
+      "loss": 0.6516,
+      "step": 11741
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.48935369890562e-06,
+      "loss": 0.6075,
+      "step": 11742
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.487938152795471e-06,
+      "loss": 0.5975,
+      "step": 11743
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.486522952086937e-06,
+      "loss": 0.6278,
+      "step": 11744
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.485108096845084e-06,
+      "loss": 0.6262,
+      "step": 11745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.48369358713497e-06,
+      "loss": 0.6138,
+      "step": 11746
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.482279423021632e-06,
+      "loss": 0.6815,
+      "step": 11747
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.480865604570099e-06,
+      "loss": 0.5962,
+      "step": 11748
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.479452131845366e-06,
+      "loss": 0.5782,
+      "step": 11749
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.478039004912432e-06,
+      "loss": 0.6745,
+      "step": 11750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.47662622383627e-06,
+      "loss": 0.6036,
+      "step": 11751
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.47521378868184e-06,
+      "loss": 0.5972,
+      "step": 11752
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4738016995140835e-06,
+      "loss": 0.6334,
+      "step": 11753
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4723899563979294e-06,
+      "loss": 0.5818,
+      "step": 11754
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.470978559398292e-06,
+      "loss": 0.5617,
+      "step": 11755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.469567508580059e-06,
+      "loss": 0.6065,
+      "step": 11756
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.468156804008115e-06,
+      "loss": 0.6052,
+      "step": 11757
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.466746445747323e-06,
+      "loss": 0.5767,
+      "step": 11758
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.465336433862533e-06,
+      "loss": 0.5877,
+      "step": 11759
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4639267684185735e-06,
+      "loss": 0.593,
+      "step": 11760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.462517449480259e-06,
+      "loss": 0.6078,
+      "step": 11761
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.461108477112394e-06,
+      "loss": 0.6811,
+      "step": 11762
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.459699851379762e-06,
+      "loss": 0.6189,
+      "step": 11763
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.458291572347126e-06,
+      "loss": 0.5653,
+      "step": 11764
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.456883640079242e-06,
+      "loss": 0.5959,
+      "step": 11765
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4554760546408445e-06,
+      "loss": 0.7057,
+      "step": 11766
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.454068816096656e-06,
+      "loss": 0.5635,
+      "step": 11767
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4526619245113785e-06,
+      "loss": 0.5782,
+      "step": 11768
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.451255379949702e-06,
+      "loss": 0.5571,
+      "step": 11769
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4498491824763e-06,
+      "loss": 0.5793,
+      "step": 11770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.448443332155826e-06,
+      "loss": 0.6026,
+      "step": 11771
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.447037829052919e-06,
+      "loss": 0.6655,
+      "step": 11772
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.445632673232208e-06,
+      "loss": 0.5737,
+      "step": 11773
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.444227864758302e-06,
+      "loss": 0.604,
+      "step": 11774
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.442823403695788e-06,
+      "loss": 0.5362,
+      "step": 11775
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.441419290109246e-06,
+      "loss": 0.5966,
+      "step": 11776
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.440015524063236e-06,
+      "loss": 0.5809,
+      "step": 11777
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.438612105622305e-06,
+      "loss": 0.5502,
+      "step": 11778
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.437209034850977e-06,
+      "loss": 0.6124,
+      "step": 11779
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.435806311813767e-06,
+      "loss": 0.6266,
+      "step": 11780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.434403936575173e-06,
+      "loss": 0.686,
+      "step": 11781
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4330019091996726e-06,
+      "loss": 0.6265,
+      "step": 11782
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.431600229751732e-06,
+      "loss": 0.6441,
+      "step": 11783
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.430198898295801e-06,
+      "loss": 0.6526,
+      "step": 11784
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.428797914896315e-06,
+      "loss": 0.5767,
+      "step": 11785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.427397279617683e-06,
+      "loss": 0.6448,
+      "step": 11786
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.42599699252431e-06,
+      "loss": 0.5654,
+      "step": 11787
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.424597053680579e-06,
+      "loss": 0.5781,
+      "step": 11788
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.423197463150865e-06,
+      "loss": 0.5858,
+      "step": 11789
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.421798220999512e-06,
+      "loss": 0.636,
+      "step": 11790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.420399327290859e-06,
+      "loss": 0.6765,
+      "step": 11791
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.419000782089228e-06,
+      "loss": 0.5882,
+      "step": 11792
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.417602585458927e-06,
+      "loss": 0.5426,
+      "step": 11793
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4162047374642375e-06,
+      "loss": 0.6385,
+      "step": 11794
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.414807238169435e-06,
+      "loss": 0.6494,
+      "step": 11795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.413410087638777e-06,
+      "loss": 0.5676,
+      "step": 11796
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.412013285936505e-06,
+      "loss": 0.6586,
+      "step": 11797
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.41061683312684e-06,
+      "loss": 0.5965,
+      "step": 11798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.40922072927399e-06,
+      "loss": 0.6055,
+      "step": 11799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.407824974442151e-06,
+      "loss": 0.6297,
+      "step": 11800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.406429568695497e-06,
+      "loss": 0.5967,
+      "step": 11801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.405034512098187e-06,
+      "loss": 0.6885,
+      "step": 11802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.403639804714369e-06,
+      "loss": 0.6214,
+      "step": 11803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.402245446608171e-06,
+      "loss": 0.6858,
+      "step": 11804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4008514378437e-06,
+      "loss": 0.6793,
+      "step": 11805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.399457778485054e-06,
+      "loss": 0.6236,
+      "step": 11806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.398064468596316e-06,
+      "loss": 0.5914,
+      "step": 11807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.396671508241549e-06,
+      "loss": 0.5431,
+      "step": 11808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.395278897484796e-06,
+      "loss": 0.6166,
+      "step": 11809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.393886636390094e-06,
+      "loss": 0.6147,
+      "step": 11810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.392494725021455e-06,
+      "loss": 0.6484,
+      "step": 11811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.391103163442882e-06,
+      "loss": 0.5666,
+      "step": 11812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.389711951718356e-06,
+      "loss": 0.5852,
+      "step": 11813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.388321089911845e-06,
+      "loss": 0.6017,
+      "step": 11814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3869305780872995e-06,
+      "loss": 0.632,
+      "step": 11815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.385540416308655e-06,
+      "loss": 0.668,
+      "step": 11816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.384150604639832e-06,
+      "loss": 0.5613,
+      "step": 11817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3827611431447325e-06,
+      "loss": 0.6426,
+      "step": 11818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.381372031887246e-06,
+      "loss": 0.6627,
+      "step": 11819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.379983270931237e-06,
+      "loss": 0.6172,
+      "step": 11820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.378594860340566e-06,
+      "loss": 0.5731,
+      "step": 11821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.377206800179068e-06,
+      "loss": 0.5958,
+      "step": 11822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.37581909051057e-06,
+      "loss": 0.5704,
+      "step": 11823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.374431731398872e-06,
+      "loss": 0.5908,
+      "step": 11824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3730447229077675e-06,
+      "loss": 0.6469,
+      "step": 11825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3716580651010305e-06,
+      "loss": 0.6209,
+      "step": 11826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.370271758042424e-06,
+      "loss": 0.595,
+      "step": 11827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.368885801795679e-06,
+      "loss": 0.6118,
+      "step": 11828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.367500196424529e-06,
+      "loss": 0.5485,
+      "step": 11829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.366114941992681e-06,
+      "loss": 0.6144,
+      "step": 11830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.364730038563832e-06,
+      "loss": 0.6076,
+      "step": 11831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.363345486201654e-06,
+      "loss": 0.6108,
+      "step": 11832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.36196128496981e-06,
+      "loss": 0.6404,
+      "step": 11833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.360577434931947e-06,
+      "loss": 0.5957,
+      "step": 11834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.359193936151691e-06,
+      "loss": 0.5603,
+      "step": 11835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.357810788692657e-06,
+      "loss": 0.6053,
+      "step": 11836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.356427992618441e-06,
+      "loss": 0.5967,
+      "step": 11837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.355045547992625e-06,
+      "loss": 0.5943,
+      "step": 11838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.353663454878769e-06,
+      "loss": 0.5991,
+      "step": 11839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.352281713340424e-06,
+      "loss": 0.6939,
+      "step": 11840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.350900323441121e-06,
+      "loss": 0.6493,
+      "step": 11841
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.349519285244379e-06,
+      "loss": 0.5727,
+      "step": 11842
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.348138598813693e-06,
+      "loss": 0.5949,
+      "step": 11843
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3467582642125475e-06,
+      "loss": 0.6439,
+      "step": 11844
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.34537828150441e-06,
+      "loss": 0.6179,
+      "step": 11845
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3439986507527355e-06,
+      "loss": 0.6879,
+      "step": 11846
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.342619372020953e-06,
+      "loss": 0.5858,
+      "step": 11847
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.341240445372483e-06,
+      "loss": 0.6071,
+      "step": 11848
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.339861870870728e-06,
+      "loss": 0.6168,
+      "step": 11849
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.338483648579077e-06,
+      "loss": 0.6088,
+      "step": 11850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.337105778560896e-06,
+      "loss": 0.6961,
+      "step": 11851
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3357282608795415e-06,
+      "loss": 0.561,
+      "step": 11852
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.334351095598352e-06,
+      "loss": 0.5473,
+      "step": 11853
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.332974282780648e-06,
+      "loss": 0.6435,
+      "step": 11854
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.331597822489733e-06,
+      "loss": 0.6253,
+      "step": 11855
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.330221714788895e-06,
+      "loss": 0.6095,
+      "step": 11856
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.328845959741416e-06,
+      "loss": 0.6021,
+      "step": 11857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.327470557410541e-06,
+      "loss": 0.5517,
+      "step": 11858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.326095507859517e-06,
+      "loss": 0.6033,
+      "step": 11859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.324720811151566e-06,
+      "loss": 0.6698,
+      "step": 11860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.323346467349901e-06,
+      "loss": 0.6075,
+      "step": 11861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.321972476517706e-06,
+      "loss": 0.7116,
+      "step": 11862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3205988387181586e-06,
+      "loss": 0.6546,
+      "step": 11863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.31922555401442e-06,
+      "loss": 0.6379,
+      "step": 11864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3178526224696375e-06,
+      "loss": 0.5734,
+      "step": 11865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.316480044146929e-06,
+      "loss": 0.6017,
+      "step": 11866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.31510781910941e-06,
+      "loss": 0.5958,
+      "step": 11867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.313735947420175e-06,
+      "loss": 0.628,
+      "step": 11868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.312364429142301e-06,
+      "loss": 0.6717,
+      "step": 11869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.31099326433885e-06,
+      "loss": 0.5961,
+      "step": 11870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.309622453072867e-06,
+      "loss": 0.643,
+      "step": 11871
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.308251995407387e-06,
+      "loss": 0.6567,
+      "step": 11872
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.306881891405414e-06,
+      "loss": 0.6125,
+      "step": 11873
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3055121411299505e-06,
+      "loss": 0.5996,
+      "step": 11874
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3041427446439755e-06,
+      "loss": 0.6222,
+      "step": 11875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.302773702010456e-06,
+      "loss": 0.5446,
+      "step": 11876
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.301405013292335e-06,
+      "loss": 0.6545,
+      "step": 11877
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.300036678552547e-06,
+      "loss": 0.541,
+      "step": 11878
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.298668697854007e-06,
+      "loss": 0.6009,
+      "step": 11879
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.297301071259618e-06,
+      "loss": 0.5915,
+      "step": 11880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.295933798832255e-06,
+      "loss": 0.6425,
+      "step": 11881
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2945668806347915e-06,
+      "loss": 0.5178,
+      "step": 11882
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.293200316730075e-06,
+      "loss": 0.6611,
+      "step": 11883
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.291834107180939e-06,
+      "loss": 0.5883,
+      "step": 11884
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.290468252050204e-06,
+      "loss": 0.5517,
+      "step": 11885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.289102751400667e-06,
+      "loss": 0.5793,
+      "step": 11886
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.287737605295122e-06,
+      "loss": 0.6245,
+      "step": 11887
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.286372813796328e-06,
+      "loss": 0.6045,
+      "step": 11888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.28500837696704e-06,
+      "loss": 0.5942,
+      "step": 11889
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.283644294869998e-06,
+      "loss": 0.6441,
+      "step": 11890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.282280567567921e-06,
+      "loss": 0.6581,
+      "step": 11891
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.280917195123509e-06,
+      "loss": 0.6374,
+      "step": 11892
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.279554177599451e-06,
+      "loss": 0.6681,
+      "step": 11893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2781915150584205e-06,
+      "loss": 0.6072,
+      "step": 11894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.276829207563071e-06,
+      "loss": 0.5201,
+      "step": 11895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2754672551760375e-06,
+      "loss": 0.6091,
+      "step": 11896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.274105657959946e-06,
+      "loss": 0.5565,
+      "step": 11897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2727444159774e-06,
+      "loss": 0.5909,
+      "step": 11898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.271383529290992e-06,
+      "loss": 0.6593,
+      "step": 11899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.270022997963292e-06,
+      "loss": 0.5971,
+      "step": 11900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.268662822056855e-06,
+      "loss": 0.5486,
+      "step": 11901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.267303001634225e-06,
+      "loss": 0.5659,
+      "step": 11902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.265943536757925e-06,
+      "loss": 0.5741,
+      "step": 11903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.264584427490463e-06,
+      "loss": 0.6552,
+      "step": 11904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.263225673894329e-06,
+      "loss": 0.5483,
+      "step": 11905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.261867276032002e-06,
+      "loss": 0.6166,
+      "step": 11906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2605092339659336e-06,
+      "loss": 0.6071,
+      "step": 11907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.25915154775857e-06,
+      "loss": 0.6089,
+      "step": 11908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2577942174723376e-06,
+      "loss": 0.6061,
+      "step": 11909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.256437243169647e-06,
+      "loss": 0.6248,
+      "step": 11910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.255080624912887e-06,
+      "loss": 0.6114,
+      "step": 11911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.253724362764437e-06,
+      "loss": 0.6067,
+      "step": 11912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.252368456786658e-06,
+      "loss": 0.6111,
+      "step": 11913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.251012907041894e-06,
+      "loss": 0.6387,
+      "step": 11914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.24965771359247e-06,
+      "loss": 0.5038,
+      "step": 11915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2483028765007e-06,
+      "loss": 0.6617,
+      "step": 11916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.246948395828877e-06,
+      "loss": 0.5834,
+      "step": 11917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.245594271639282e-06,
+      "loss": 0.606,
+      "step": 11918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.244240503994174e-06,
+      "loss": 0.6608,
+      "step": 11919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.242887092955801e-06,
+      "loss": 0.6333,
+      "step": 11920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.241534038586394e-06,
+      "loss": 0.6371,
+      "step": 11921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.240181340948161e-06,
+      "loss": 0.5732,
+      "step": 11922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.238829000103301e-06,
+      "loss": 0.6078,
+      "step": 11923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.237477016113994e-06,
+      "loss": 0.6007,
+      "step": 11924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2361253890424055e-06,
+      "loss": 0.567,
+      "step": 11925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.234774118950679e-06,
+      "loss": 0.6055,
+      "step": 11926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2334232059009485e-06,
+      "loss": 0.5861,
+      "step": 11927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.232072649955327e-06,
+      "loss": 0.6761,
+      "step": 11928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2307224511759146e-06,
+      "loss": 0.6025,
+      "step": 11929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.229372609624789e-06,
+      "loss": 0.6704,
+      "step": 11930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.228023125364017e-06,
+      "loss": 0.6691,
+      "step": 11931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.226673998455648e-06,
+      "loss": 0.5858,
+      "step": 11932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.225325228961718e-06,
+      "loss": 0.646,
+      "step": 11933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.223976816944235e-06,
+      "loss": 0.6253,
+      "step": 11934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.222628762465204e-06,
+      "loss": 0.5999,
+      "step": 11935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.221281065586606e-06,
+      "loss": 0.5761,
+      "step": 11936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.219933726370407e-06,
+      "loss": 0.6514,
+      "step": 11937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.218586744878559e-06,
+      "loss": 0.6124,
+      "step": 11938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.217240121172995e-06,
+      "loss": 0.6066,
+      "step": 11939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.215893855315636e-06,
+      "loss": 0.5833,
+      "step": 11940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.214547947368375e-06,
+      "loss": 0.6015,
+      "step": 11941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2132023973931e-06,
+      "loss": 0.5853,
+      "step": 11942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.211857205451679e-06,
+      "loss": 0.528,
+      "step": 11943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.210512371605967e-06,
+      "loss": 0.6221,
+      "step": 11944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.209167895917792e-06,
+      "loss": 0.6221,
+      "step": 11945
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.207823778448976e-06,
+      "loss": 0.5507,
+      "step": 11946
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.206480019261321e-06,
+      "loss": 0.6188,
+      "step": 11947
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.205136618416615e-06,
+      "loss": 0.5303,
+      "step": 11948
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.203793575976622e-06,
+      "loss": 0.6385,
+      "step": 11949
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.202450892003096e-06,
+      "loss": 0.6251,
+      "step": 11950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.201108566557775e-06,
+      "loss": 0.588,
+      "step": 11951
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.199766599702376e-06,
+      "loss": 0.699,
+      "step": 11952
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.198424991498605e-06,
+      "loss": 0.586,
+      "step": 11953
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.197083742008148e-06,
+      "loss": 0.6124,
+      "step": 11954
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.195742851292675e-06,
+      "loss": 0.5318,
+      "step": 11955
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.194402319413838e-06,
+      "loss": 0.6397,
+      "step": 11956
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.193062146433275e-06,
+      "loss": 0.6045,
+      "step": 11957
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.191722332412607e-06,
+      "loss": 0.6535,
+      "step": 11958
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.19038287741344e-06,
+      "loss": 0.5794,
+      "step": 11959
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.189043781497357e-06,
+      "loss": 0.61,
+      "step": 11960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.187705044725932e-06,
+      "loss": 0.6058,
+      "step": 11961
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.186366667160718e-06,
+      "loss": 0.6383,
+      "step": 11962
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.185028648863258e-06,
+      "loss": 0.5741,
+      "step": 11963
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.183690989895067e-06,
+      "loss": 0.621,
+      "step": 11964
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.182353690317652e-06,
+      "loss": 0.6221,
+      "step": 11965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.181016750192504e-06,
+      "loss": 0.6038,
+      "step": 11966
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.179680169581094e-06,
+      "loss": 0.614,
+      "step": 11967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1783439485448735e-06,
+      "loss": 0.6175,
+      "step": 11968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.177008087145286e-06,
+      "loss": 0.6456,
+      "step": 11969
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.175672585443755e-06,
+      "loss": 0.5632,
+      "step": 11970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.174337443501682e-06,
+      "loss": 0.6335,
+      "step": 11971
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.173002661380458e-06,
+      "loss": 0.6393,
+      "step": 11972
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.171668239141455e-06,
+      "loss": 0.6192,
+      "step": 11973
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1703341768460335e-06,
+      "loss": 0.6257,
+      "step": 11974
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.169000474555528e-06,
+      "loss": 0.5619,
+      "step": 11975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1676671323312624e-06,
+      "loss": 0.5733,
+      "step": 11976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.166334150234545e-06,
+      "loss": 0.6224,
+      "step": 11977
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.165001528326666e-06,
+      "loss": 0.6269,
+      "step": 11978
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.163669266668897e-06,
+      "loss": 0.6105,
+      "step": 11979
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.162337365322495e-06,
+      "loss": 0.6164,
+      "step": 11980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1610058243487e-06,
+      "loss": 0.5764,
+      "step": 11981
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.159674643808741e-06,
+      "loss": 0.5667,
+      "step": 11982
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.158343823763818e-06,
+      "loss": 0.5827,
+      "step": 11983
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.157013364275125e-06,
+      "loss": 0.6345,
+      "step": 11984
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.155683265403834e-06,
+      "loss": 0.5892,
+      "step": 11985
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1543535272111035e-06,
+      "loss": 0.665,
+      "step": 11986
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.153024149758076e-06,
+      "loss": 0.643,
+      "step": 11987
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1516951331058745e-06,
+      "loss": 0.5787,
+      "step": 11988
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.150366477315609e-06,
+      "loss": 0.6135,
+      "step": 11989
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1490381824483636e-06,
+      "loss": 0.5931,
+      "step": 11990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.14771024856522e-06,
+      "loss": 0.5474,
+      "step": 11991
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.146382675727233e-06,
+      "loss": 0.5657,
+      "step": 11992
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.145055463995445e-06,
+      "loss": 0.5988,
+      "step": 11993
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.143728613430878e-06,
+      "loss": 0.6006,
+      "step": 11994
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.142402124094543e-06,
+      "loss": 0.6085,
+      "step": 11995
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.14107599604743e-06,
+      "loss": 0.5198,
+      "step": 11996
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.139750229350518e-06,
+      "loss": 0.5624,
+      "step": 11997
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.138424824064758e-06,
+      "loss": 0.5366,
+      "step": 11998
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.137099780251094e-06,
+      "loss": 0.5749,
+      "step": 11999
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.135775097970453e-06,
+      "loss": 0.6047,
+      "step": 12000
+    },
+    {
+      "epoch": 0.79,
+      "eval_text_loss": 0.6544542908668518,
+      "eval_text_runtime": 1.8289,
+      "eval_text_samples_per_second": 40.46,
+      "eval_text_steps_per_second": 1.094,
+      "step": 12000
+    },
+    {
+      "epoch": 0.79,
+      "eval_svit_loss": 0.7631040215492249,
+      "eval_svit_runtime": 6.6706,
+      "eval_svit_samples_per_second": 122.328,
+      "eval_svit_steps_per_second": 1.949,
+      "step": 12000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.134450777283743e-06,
+      "loss": 0.6258,
+      "step": 12001
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.133126818251854e-06,
+      "loss": 0.6237,
+      "step": 12002
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1318032209356625e-06,
+      "loss": 0.6275,
+      "step": 12003
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.130479985396031e-06,
+      "loss": 0.6393,
+      "step": 12004
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.129157111693793e-06,
+      "loss": 0.6599,
+      "step": 12005
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.127834599889778e-06,
+      "loss": 0.6034,
+      "step": 12006
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.126512450044795e-06,
+      "loss": 0.6206,
+      "step": 12007
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.125190662219638e-06,
+      "loss": 0.6537,
+      "step": 12008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1238692364750765e-06,
+      "loss": 0.6125,
+      "step": 12009
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.122548172871874e-06,
+      "loss": 0.5736,
+      "step": 12010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.121227471470768e-06,
+      "loss": 0.5591,
+      "step": 12011
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.11990713233249e-06,
+      "loss": 0.5927,
+      "step": 12012
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.118587155517743e-06,
+      "loss": 0.575,
+      "step": 12013
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.117267541087221e-06,
+      "loss": 0.655,
+      "step": 12014
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1159482891016e-06,
+      "loss": 0.5765,
+      "step": 12015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.114629399621541e-06,
+      "loss": 0.5979,
+      "step": 12016
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1133108727076805e-06,
+      "loss": 0.5997,
+      "step": 12017
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.111992708420646e-06,
+      "loss": 0.6142,
+      "step": 12018
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.110674906821047e-06,
+      "loss": 0.6439,
+      "step": 12019
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.109357467969475e-06,
+      "loss": 0.6604,
+      "step": 12020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.108040391926507e-06,
+      "loss": 0.6141,
+      "step": 12021
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.106723678752699e-06,
+      "loss": 0.6,
+      "step": 12022
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.105407328508597e-06,
+      "loss": 0.6137,
+      "step": 12023
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.10409134125472e-06,
+      "loss": 0.6362,
+      "step": 12024
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.102775717051582e-06,
+      "loss": 0.6589,
+      "step": 12025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.10146045595967e-06,
+      "loss": 0.5925,
+      "step": 12026
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.100145558039467e-06,
+      "loss": 0.6222,
+      "step": 12027
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.098831023351425e-06,
+      "loss": 0.5843,
+      "step": 12028
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.097516851955984e-06,
+      "loss": 0.6127,
+      "step": 12029
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.096203043913574e-06,
+      "loss": 0.6294,
+      "step": 12030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.094889599284604e-06,
+      "loss": 0.6912,
+      "step": 12031
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.093576518129462e-06,
+      "loss": 0.6331,
+      "step": 12032
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.092263800508524e-06,
+      "loss": 0.6155,
+      "step": 12033
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.090951446482148e-06,
+      "loss": 0.5914,
+      "step": 12034
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.089639456110676e-06,
+      "loss": 0.6961,
+      "step": 12035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.088327829454433e-06,
+      "loss": 0.6044,
+      "step": 12036
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.087016566573727e-06,
+      "loss": 0.6648,
+      "step": 12037
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.085705667528852e-06,
+      "loss": 0.5404,
+      "step": 12038
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.084395132380078e-06,
+      "loss": 0.5991,
+      "step": 12039
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.083084961187665e-06,
+      "loss": 0.629,
+      "step": 12040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0817751540118524e-06,
+      "loss": 0.5879,
+      "step": 12041
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.08046571091287e-06,
+      "loss": 0.5149,
+      "step": 12042
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.079156631950919e-06,
+      "loss": 0.6042,
+      "step": 12043
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.077847917186192e-06,
+      "loss": 0.6334,
+      "step": 12044
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.076539566678866e-06,
+      "loss": 0.6042,
+      "step": 12045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.075231580489098e-06,
+      "loss": 0.6278,
+      "step": 12046
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.073923958677024e-06,
+      "loss": 0.6342,
+      "step": 12047
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.072616701302773e-06,
+      "loss": 0.6225,
+      "step": 12048
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.071309808426449e-06,
+      "loss": 0.6443,
+      "step": 12049
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0700032801081475e-06,
+      "loss": 0.6204,
+      "step": 12050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.068697116407935e-06,
+      "loss": 0.6728,
+      "step": 12051
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.067391317385873e-06,
+      "loss": 0.5444,
+      "step": 12052
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.066085883101999e-06,
+      "loss": 0.5515,
+      "step": 12053
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.064780813616338e-06,
+      "loss": 0.6,
+      "step": 12054
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.063476108988898e-06,
+      "loss": 0.5787,
+      "step": 12055
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.062171769279666e-06,
+      "loss": 0.6335,
+      "step": 12056
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.060867794548619e-06,
+      "loss": 0.5744,
+      "step": 12057
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.059564184855708e-06,
+      "loss": 0.5929,
+      "step": 12058
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.058260940260874e-06,
+      "loss": 0.6443,
+      "step": 12059
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.056958060824041e-06,
+      "loss": 0.5594,
+      "step": 12060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.055655546605119e-06,
+      "loss": 0.5795,
+      "step": 12061
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.054353397663987e-06,
+      "loss": 0.5443,
+      "step": 12062
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.053051614060525e-06,
+      "loss": 0.6531,
+      "step": 12063
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.051750195854585e-06,
+      "loss": 0.5946,
+      "step": 12064
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.05044914310601e-06,
+      "loss": 0.6231,
+      "step": 12065
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.049148455874617e-06,
+      "loss": 0.6419,
+      "step": 12066
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.047848134220212e-06,
+      "loss": 0.5621,
+      "step": 12067
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.046548178202585e-06,
+      "loss": 0.609,
+      "step": 12068
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.045248587881506e-06,
+      "loss": 0.5629,
+      "step": 12069
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.043949363316732e-06,
+      "loss": 0.5529,
+      "step": 12070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.042650504567998e-06,
+      "loss": 0.6459,
+      "step": 12071
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.041352011695028e-06,
+      "loss": 0.6349,
+      "step": 12072
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.040053884757523e-06,
+      "loss": 0.6189,
+      "step": 12073
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.038756123815172e-06,
+      "loss": 0.7029,
+      "step": 12074
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0374587289276435e-06,
+      "loss": 0.6342,
+      "step": 12075
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.036161700154597e-06,
+      "loss": 0.5689,
+      "step": 12076
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.034865037555663e-06,
+      "loss": 0.6013,
+      "step": 12077
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.033568741190464e-06,
+      "loss": 0.6107,
+      "step": 12078
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.032272811118602e-06,
+      "loss": 0.6322,
+      "step": 12079
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0309772473996684e-06,
+      "loss": 0.6234,
+      "step": 12080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.029682050093224e-06,
+      "loss": 0.6428,
+      "step": 12081
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0283872192588284e-06,
+      "loss": 0.5942,
+      "step": 12082
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.027092754956014e-06,
+      "loss": 0.6084,
+      "step": 12083
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.025798657244303e-06,
+      "loss": 0.5543,
+      "step": 12084
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.024504926183193e-06,
+      "loss": 0.6165,
+      "step": 12085
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.023211561832173e-06,
+      "loss": 0.5923,
+      "step": 12086
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.02191856425071e-06,
+      "loss": 0.6347,
+      "step": 12087
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0206259334982545e-06,
+      "loss": 0.6147,
+      "step": 12088
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.019333669634242e-06,
+      "loss": 0.6019,
+      "step": 12089
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0180417727180915e-06,
+      "loss": 0.5959,
+      "step": 12090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.016750242809206e-06,
+      "loss": 0.5566,
+      "step": 12091
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.015459079966963e-06,
+      "loss": 0.5539,
+      "step": 12092
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.014168284250732e-06,
+      "loss": 0.6197,
+      "step": 12093
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.012877855719865e-06,
+      "loss": 0.5348,
+      "step": 12094
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.011587794433698e-06,
+      "loss": 0.6487,
+      "step": 12095
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.010298100451542e-06,
+      "loss": 0.6377,
+      "step": 12096
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.009008773832697e-06,
+      "loss": 0.6712,
+      "step": 12097
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.007719814636449e-06,
+      "loss": 0.6138,
+      "step": 12098
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.006431222922066e-06,
+      "loss": 0.5959,
+      "step": 12099
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.005142998748789e-06,
+      "loss": 0.5822,
+      "step": 12100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.003855142175855e-06,
+      "loss": 0.5918,
+      "step": 12101
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.002567653262479e-06,
+      "loss": 0.6102,
+      "step": 12102
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.001280532067859e-06,
+      "loss": 0.6192,
+      "step": 12103
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.999993778651174e-06,
+      "loss": 0.5792,
+      "step": 12104
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9987073930715925e-06,
+      "loss": 0.5959,
+      "step": 12105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.997421375388262e-06,
+      "loss": 0.5891,
+      "step": 12106
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.996135725660308e-06,
+      "loss": 0.5844,
+      "step": 12107
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9948504439468486e-06,
+      "loss": 0.589,
+      "step": 12108
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.993565530306978e-06,
+      "loss": 0.607,
+      "step": 12109
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.992280984799781e-06,
+      "loss": 0.581,
+      "step": 12110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.990996807484312e-06,
+      "loss": 0.5668,
+      "step": 12111
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.989712998419623e-06,
+      "loss": 0.6781,
+      "step": 12112
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.988429557664741e-06,
+      "loss": 0.5544,
+      "step": 12113
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.987146485278681e-06,
+      "loss": 0.5429,
+      "step": 12114
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.985863781320435e-06,
+      "loss": 0.6229,
+      "step": 12115
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9845814458489806e-06,
+      "loss": 0.6372,
+      "step": 12116
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.983299478923282e-06,
+      "loss": 0.6092,
+      "step": 12117
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.982017880602284e-06,
+      "loss": 0.591,
+      "step": 12118
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.980736650944909e-06,
+      "loss": 0.6122,
+      "step": 12119
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.979455790010071e-06,
+      "loss": 0.6101,
+      "step": 12120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.978175297856664e-06,
+      "loss": 0.6032,
+      "step": 12121
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.976895174543562e-06,
+      "loss": 0.5905,
+      "step": 12122
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.975615420129627e-06,
+      "loss": 0.6209,
+      "step": 12123
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.974336034673701e-06,
+      "loss": 0.5383,
+      "step": 12124
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.973057018234611e-06,
+      "loss": 0.5486,
+      "step": 12125
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9717783708711614e-06,
+      "loss": 0.6067,
+      "step": 12126
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.970500092642147e-06,
+      "loss": 0.62,
+      "step": 12127
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.969222183606342e-06,
+      "loss": 0.6302,
+      "step": 12128
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.967944643822506e-06,
+      "loss": 0.6017,
+      "step": 12129
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.966667473349376e-06,
+      "loss": 0.6289,
+      "step": 12130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.965390672245678e-06,
+      "loss": 0.5813,
+      "step": 12131
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.964114240570118e-06,
+      "loss": 0.6457,
+      "step": 12132
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.96283817838139e-06,
+      "loss": 0.5883,
+      "step": 12133
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.961562485738159e-06,
+      "loss": 0.6007,
+      "step": 12134
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.960287162699086e-06,
+      "loss": 0.5929,
+      "step": 12135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.959012209322809e-06,
+      "loss": 0.6219,
+      "step": 12136
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.95773762566795e-06,
+      "loss": 0.5486,
+      "step": 12137
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.956463411793113e-06,
+      "loss": 0.5963,
+      "step": 12138
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.955189567756886e-06,
+      "loss": 0.5686,
+      "step": 12139
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.953916093617844e-06,
+      "loss": 0.5905,
+      "step": 12140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.952642989434534e-06,
+      "loss": 0.5982,
+      "step": 12141
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.951370255265495e-06,
+      "loss": 0.6108,
+      "step": 12142
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.950097891169249e-06,
+      "loss": 0.6601,
+      "step": 12143
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9488258972043e-06,
+      "loss": 0.6258,
+      "step": 12144
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.947554273429127e-06,
+      "loss": 0.5862,
+      "step": 12145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9462830199022054e-06,
+      "loss": 0.6029,
+      "step": 12146
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.945012136681984e-06,
+      "loss": 0.5583,
+      "step": 12147
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.943741623826901e-06,
+      "loss": 0.5996,
+      "step": 12148
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.942471481395367e-06,
+      "loss": 0.564,
+      "step": 12149
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.941201709445787e-06,
+      "loss": 0.604,
+      "step": 12150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.939932308036546e-06,
+      "loss": 0.6605,
+      "step": 12151
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.93866327722601e-06,
+      "loss": 0.6254,
+      "step": 12152
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.937394617072526e-06,
+      "loss": 0.6106,
+      "step": 12153
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.936126327634425e-06,
+      "loss": 0.571,
+      "step": 12154
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.934858408970031e-06,
+      "loss": 0.592,
+      "step": 12155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.933590861137634e-06,
+      "loss": 0.6131,
+      "step": 12156
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9323236841955185e-06,
+      "loss": 0.5612,
+      "step": 12157
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.931056878201948e-06,
+      "loss": 0.6026,
+      "step": 12158
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.929790443215172e-06,
+      "loss": 0.6271,
+      "step": 12159
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.928524379293418e-06,
+      "loss": 0.5777,
+      "step": 12160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.927258686494899e-06,
+      "loss": 0.5987,
+      "step": 12161
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.925993364877813e-06,
+      "loss": 0.6375,
+      "step": 12162
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.92472841450034e-06,
+      "loss": 0.6058,
+      "step": 12163
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9234638354206366e-06,
+      "loss": 0.5801,
+      "step": 12164
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.922199627696852e-06,
+      "loss": 0.5861,
+      "step": 12165
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9209357913871135e-06,
+      "loss": 0.6472,
+      "step": 12166
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9196723265495335e-06,
+      "loss": 0.6392,
+      "step": 12167
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9184092332422e-06,
+      "loss": 0.599,
+      "step": 12168
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.917146511523194e-06,
+      "loss": 0.6048,
+      "step": 12169
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.915884161450573e-06,
+      "loss": 0.6223,
+      "step": 12170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.91462218308238e-06,
+      "loss": 0.6004,
+      "step": 12171
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9133605764766405e-06,
+      "loss": 0.5975,
+      "step": 12172
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.912099341691363e-06,
+      "loss": 0.5917,
+      "step": 12173
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.910838478784541e-06,
+      "loss": 0.6146,
+      "step": 12174
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.909577987814142e-06,
+      "loss": 0.6119,
+      "step": 12175
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9083178688381284e-06,
+      "loss": 0.6486,
+      "step": 12176
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9070581219144376e-06,
+      "loss": 0.5216,
+      "step": 12177
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.905798747100995e-06,
+      "loss": 0.6621,
+      "step": 12178
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.904539744455701e-06,
+      "loss": 0.6172,
+      "step": 12179
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.903281114036447e-06,
+      "loss": 0.6764,
+      "step": 12180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.902022855901105e-06,
+      "loss": 0.5962,
+      "step": 12181
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.900764970107531e-06,
+      "loss": 0.5426,
+      "step": 12182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.899507456713556e-06,
+      "loss": 0.5975,
+      "step": 12183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.898250315777004e-06,
+      "loss": 0.6065,
+      "step": 12184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.896993547355679e-06,
+      "loss": 0.5845,
+      "step": 12185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.895737151507363e-06,
+      "loss": 0.5432,
+      "step": 12186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8944811282898274e-06,
+      "loss": 0.5705,
+      "step": 12187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.893225477760824e-06,
+      "loss": 0.5933,
+      "step": 12188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.891970199978087e-06,
+      "loss": 0.5682,
+      "step": 12189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.890715294999332e-06,
+      "loss": 0.578,
+      "step": 12190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.889460762882257e-06,
+      "loss": 0.5272,
+      "step": 12191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.888206603684549e-06,
+      "loss": 0.6887,
+      "step": 12192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.886952817463875e-06,
+      "loss": 0.5662,
+      "step": 12193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.885699404277878e-06,
+      "loss": 0.5688,
+      "step": 12194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.884446364184193e-06,
+      "loss": 0.6176,
+      "step": 12195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.883193697240433e-06,
+      "loss": 0.5891,
+      "step": 12196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8819414035041985e-06,
+      "loss": 0.6068,
+      "step": 12197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.880689483033064e-06,
+      "loss": 0.556,
+      "step": 12198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.879437935884594e-06,
+      "loss": 0.5936,
+      "step": 12199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.878186762116336e-06,
+      "loss": 0.5644,
+      "step": 12200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.876935961785821e-06,
+      "loss": 0.6229,
+      "step": 12201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.875685534950554e-06,
+      "loss": 0.604,
+      "step": 12202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.874435481668031e-06,
+      "loss": 0.6281,
+      "step": 12203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.873185801995731e-06,
+      "loss": 0.5896,
+      "step": 12204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8719364959911125e-06,
+      "loss": 0.574,
+      "step": 12205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.870687563711619e-06,
+      "loss": 0.6632,
+      "step": 12206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.869439005214673e-06,
+      "loss": 0.5796,
+      "step": 12207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8681908205576905e-06,
+      "loss": 0.6048,
+      "step": 12208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.866943009798053e-06,
+      "loss": 0.545,
+      "step": 12209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.865695572993139e-06,
+      "loss": 0.6381,
+      "step": 12210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.864448510200304e-06,
+      "loss": 0.6116,
+      "step": 12211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.863201821476891e-06,
+      "loss": 0.6109,
+      "step": 12212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.861955506880215e-06,
+      "loss": 0.6039,
+      "step": 12213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.860709566467586e-06,
+      "loss": 0.6011,
+      "step": 12214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.859464000296293e-06,
+      "loss": 0.627,
+      "step": 12215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8582188084236065e-06,
+      "loss": 0.5702,
+      "step": 12216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.856973990906774e-06,
+      "loss": 0.617,
+      "step": 12217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.855729547803038e-06,
+      "loss": 0.656,
+      "step": 12218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.854485479169614e-06,
+      "loss": 0.6596,
+      "step": 12219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.853241785063705e-06,
+      "loss": 0.5988,
+      "step": 12220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.851998465542497e-06,
+      "loss": 0.57,
+      "step": 12221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.850755520663156e-06,
+      "loss": 0.5689,
+      "step": 12222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.849512950482833e-06,
+      "loss": 0.5778,
+      "step": 12223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.848270755058659e-06,
+      "loss": 0.6021,
+      "step": 12224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.84702893444775e-06,
+      "loss": 0.598,
+      "step": 12225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.845787488707206e-06,
+      "loss": 0.5289,
+      "step": 12226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.844546417894109e-06,
+      "loss": 0.5551,
+      "step": 12227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.843305722065519e-06,
+      "loss": 0.5597,
+      "step": 12228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.842065401278484e-06,
+      "loss": 0.5256,
+      "step": 12229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.840825455590036e-06,
+      "loss": 0.5952,
+      "step": 12230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.839585885057188e-06,
+      "loss": 0.6101,
+      "step": 12231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.838346689736929e-06,
+      "loss": 0.6185,
+      "step": 12232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.83710786968624e-06,
+      "loss": 0.5523,
+      "step": 12233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.835869424962083e-06,
+      "loss": 0.5621,
+      "step": 12234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.834631355621402e-06,
+      "loss": 0.6368,
+      "step": 12235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.833393661721119e-06,
+      "loss": 0.6216,
+      "step": 12236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.832156343318144e-06,
+      "loss": 0.5912,
+      "step": 12237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.830919400469369e-06,
+      "loss": 0.6355,
+      "step": 12238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.829682833231668e-06,
+      "loss": 0.6515,
+      "step": 12239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.828446641661898e-06,
+      "loss": 0.6392,
+      "step": 12240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.827210825816898e-06,
+      "loss": 0.6143,
+      "step": 12241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8259753857534945e-06,
+      "loss": 0.6861,
+      "step": 12242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.824740321528485e-06,
+      "loss": 0.6171,
+      "step": 12243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.823505633198662e-06,
+      "loss": 0.6001,
+      "step": 12244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.822271320820795e-06,
+      "loss": 0.5999,
+      "step": 12245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.82103738445164e-06,
+      "loss": 0.6269,
+      "step": 12246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.819803824147929e-06,
+      "loss": 0.548,
+      "step": 12247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.818570639966381e-06,
+      "loss": 0.6448,
+      "step": 12248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.817337831963697e-06,
+      "loss": 0.6582,
+      "step": 12249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.816105400196565e-06,
+      "loss": 0.6462,
+      "step": 12250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.814873344721647e-06,
+      "loss": 0.5931,
+      "step": 12251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8136416655955947e-06,
+      "loss": 0.6466,
+      "step": 12252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8124103628750386e-06,
+      "loss": 0.5504,
+      "step": 12253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8111794366165968e-06,
+      "loss": 0.6049,
+      "step": 12254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.809948886876863e-06,
+      "loss": 0.6092,
+      "step": 12255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8087187137124196e-06,
+      "loss": 0.6041,
+      "step": 12256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8074889171798318e-06,
+      "loss": 0.5936,
+      "step": 12257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8062594973356396e-06,
+      "loss": 0.5328,
+      "step": 12258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.805030454236375e-06,
+      "loss": 0.563,
+      "step": 12259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.803801787938546e-06,
+      "loss": 0.5877,
+      "step": 12260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8025734984986517e-06,
+      "loss": 0.6176,
+      "step": 12261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.801345585973163e-06,
+      "loss": 0.5277,
+      "step": 12262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.80011805041854e-06,
+      "loss": 0.6065,
+      "step": 12263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.798890891891225e-06,
+      "loss": 0.617,
+      "step": 12264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7976641104476443e-06,
+      "loss": 0.6054,
+      "step": 12265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7964377061442005e-06,
+      "loss": 0.5946,
+      "step": 12266
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.795211679037285e-06,
+      "loss": 0.5878,
+      "step": 12267
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7939860291832713e-06,
+      "loss": 0.6113,
+      "step": 12268
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.792760756638514e-06,
+      "loss": 0.5496,
+      "step": 12269
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.791535861459349e-06,
+      "loss": 0.5965,
+      "step": 12270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.790311343702098e-06,
+      "loss": 0.6236,
+      "step": 12271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.789087203423062e-06,
+      "loss": 0.6053,
+      "step": 12272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7878634406785286e-06,
+      "loss": 0.5636,
+      "step": 12273
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7866400555247653e-06,
+      "loss": 0.5901,
+      "step": 12274
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7854170480180234e-06,
+      "loss": 0.5939,
+      "step": 12275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7841944182145365e-06,
+      "loss": 0.6036,
+      "step": 12276
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.782972166170518e-06,
+      "loss": 0.5603,
+      "step": 12277
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7817502919421693e-06,
+      "loss": 0.5942,
+      "step": 12278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.78052879558567e-06,
+      "loss": 0.5571,
+      "step": 12279
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7793076771571858e-06,
+      "loss": 0.637,
+      "step": 12280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7780869367128616e-06,
+      "loss": 0.5825,
+      "step": 12281
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7768665743088264e-06,
+      "loss": 0.5943,
+      "step": 12282
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.775646590001193e-06,
+      "loss": 0.6001,
+      "step": 12283
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.774426983846058e-06,
+      "loss": 0.5922,
+      "step": 12284
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.773207755899493e-06,
+      "loss": 0.5895,
+      "step": 12285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.771988906217561e-06,
+      "loss": 0.5629,
+      "step": 12286
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7707704348563034e-06,
+      "loss": 0.636,
+      "step": 12287
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7695523418717452e-06,
+      "loss": 0.644,
+      "step": 12288
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.768334627319895e-06,
+      "loss": 0.5829,
+      "step": 12289
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7671172912567412e-06,
+      "loss": 0.6181,
+      "step": 12290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7659003337382596e-06,
+      "loss": 0.562,
+      "step": 12291
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7646837548203996e-06,
+      "loss": 0.5954,
+      "step": 12292
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.763467554559104e-06,
+      "loss": 0.5751,
+      "step": 12293
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7622517330102903e-06,
+      "loss": 0.5718,
+      "step": 12294
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7610362902298663e-06,
+      "loss": 0.6716,
+      "step": 12295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.75982122627371e-06,
+      "loss": 0.5808,
+      "step": 12296
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7586065411976947e-06,
+      "loss": 0.5659,
+      "step": 12297
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.757392235057669e-06,
+      "loss": 0.5706,
+      "step": 12298
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7561783079094706e-06,
+      "loss": 0.5792,
+      "step": 12299
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.754964759808909e-06,
+      "loss": 0.6364,
+      "step": 12300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7537515908117853e-06,
+      "loss": 0.5418,
+      "step": 12301
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.752538800973882e-06,
+      "loss": 0.5324,
+      "step": 12302
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7513263903509624e-06,
+      "loss": 0.5516,
+      "step": 12303
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.750114358998771e-06,
+      "loss": 0.6488,
+      "step": 12304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.748902706973037e-06,
+      "loss": 0.5882,
+      "step": 12305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7476914343294714e-06,
+      "loss": 0.5844,
+      "step": 12306
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7464805411237685e-06,
+      "loss": 0.5911,
+      "step": 12307
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.745270027411605e-06,
+      "loss": 0.6117,
+      "step": 12308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7440598932486393e-06,
+      "loss": 0.6092,
+      "step": 12309
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.742850138690517e-06,
+      "loss": 0.6343,
+      "step": 12310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7416407637928555e-06,
+      "loss": 0.6669,
+      "step": 12311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7404317686112637e-06,
+      "loss": 0.5863,
+      "step": 12312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7392231532013314e-06,
+      "loss": 0.6102,
+      "step": 12313
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7380149176186325e-06,
+      "loss": 0.6326,
+      "step": 12314
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.736807061918717e-06,
+      "loss": 0.6755,
+      "step": 12315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7355995861571228e-06,
+      "loss": 0.606,
+      "step": 12316
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7343924903893707e-06,
+      "loss": 0.6352,
+      "step": 12317
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7331857746709623e-06,
+      "loss": 0.6311,
+      "step": 12318
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7319794390573806e-06,
+      "loss": 0.657,
+      "step": 12319
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7307734836040924e-06,
+      "loss": 0.5717,
+      "step": 12320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7295679083665475e-06,
+      "loss": 0.5845,
+      "step": 12321
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7283627134001777e-06,
+      "loss": 0.6404,
+      "step": 12322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7271578987603976e-06,
+      "loss": 0.5903,
+      "step": 12323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.725953464502604e-06,
+      "loss": 0.6844,
+      "step": 12324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7247494106821804e-06,
+      "loss": 0.5971,
+      "step": 12325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.723545737354482e-06,
+      "loss": 0.6848,
+      "step": 12326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7223424445748547e-06,
+      "loss": 0.5924,
+      "step": 12327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.721139532398627e-06,
+      "loss": 0.6081,
+      "step": 12328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.719937000881111e-06,
+      "loss": 0.6094,
+      "step": 12329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.718734850077592e-06,
+      "loss": 0.6453,
+      "step": 12330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7175330800433496e-06,
+      "loss": 0.6707,
+      "step": 12331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.716331690833639e-06,
+      "loss": 0.5976,
+      "step": 12332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7151306825037016e-06,
+      "loss": 0.6449,
+      "step": 12333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7139300551087544e-06,
+      "loss": 0.6381,
+      "step": 12334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.712729808704007e-06,
+      "loss": 0.574,
+      "step": 12335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.711529943344644e-06,
+      "loss": 0.5336,
+      "step": 12336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7103304590858336e-06,
+      "loss": 0.5619,
+      "step": 12337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7091313559827294e-06,
+      "loss": 0.5867,
+      "step": 12338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7079326340904632e-06,
+      "loss": 0.6384,
+      "step": 12339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.706734293464155e-06,
+      "loss": 0.5826,
+      "step": 12340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7055363341589028e-06,
+      "loss": 0.5701,
+      "step": 12341
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7043387562297884e-06,
+      "loss": 0.645,
+      "step": 12342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7031415597318744e-06,
+      "loss": 0.6022,
+      "step": 12343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.701944744720212e-06,
+      "loss": 0.6015,
+      "step": 12344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7007483112498243e-06,
+      "loss": 0.5896,
+      "step": 12345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6995522593757263e-06,
+      "loss": 0.5948,
+      "step": 12346
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6983565891529106e-06,
+      "loss": 0.6211,
+      "step": 12347
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6971613006363574e-06,
+      "loss": 0.6137,
+      "step": 12348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.69596639388102e-06,
+      "loss": 0.5651,
+      "step": 12349
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.694771868941841e-06,
+      "loss": 0.5891,
+      "step": 12350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.693577725873746e-06,
+      "loss": 0.5839,
+      "step": 12351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6923839647316428e-06,
+      "loss": 0.5443,
+      "step": 12352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.691190585570417e-06,
+      "loss": 0.6095,
+      "step": 12353
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.689997588444939e-06,
+      "loss": 0.6426,
+      "step": 12354
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.688804973410064e-06,
+      "loss": 0.6042,
+      "step": 12355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.687612740520629e-06,
+      "loss": 0.5145,
+      "step": 12356
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6864208898314503e-06,
+      "loss": 0.5807,
+      "step": 12357
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6852294213973294e-06,
+      "loss": 0.5996,
+      "step": 12358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6840383352730526e-06,
+      "loss": 0.6124,
+      "step": 12359
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.682847631513381e-06,
+      "loss": 0.5854,
+      "step": 12360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6816573101730634e-06,
+      "loss": 0.581,
+      "step": 12361
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.680467371306832e-06,
+      "loss": 0.595,
+      "step": 12362
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.679277814969402e-06,
+      "loss": 0.5373,
+      "step": 12363
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6780886412154625e-06,
+      "loss": 0.5825,
+      "step": 12364
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6768998500996945e-06,
+      "loss": 0.5904,
+      "step": 12365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6757114416767595e-06,
+      "loss": 0.6095,
+      "step": 12366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6745234160013e-06,
+      "loss": 0.5547,
+      "step": 12367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6733357731279377e-06,
+      "loss": 0.6576,
+      "step": 12368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6721485131112816e-06,
+      "loss": 0.5778,
+      "step": 12369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.670961636005922e-06,
+      "loss": 0.5974,
+      "step": 12370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6697751418664308e-06,
+      "loss": 0.6726,
+      "step": 12371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6685890307473627e-06,
+      "loss": 0.5745,
+      "step": 12372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6674033027032543e-06,
+      "loss": 0.6413,
+      "step": 12373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6662179577886283e-06,
+      "loss": 0.6096,
+      "step": 12374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.665032996057979e-06,
+      "loss": 0.6186,
+      "step": 12375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.663848417565796e-06,
+      "loss": 0.5585,
+      "step": 12376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6626642223665447e-06,
+      "loss": 0.6531,
+      "step": 12377
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6614804105146753e-06,
+      "loss": 0.5725,
+      "step": 12378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.660296982064614e-06,
+      "loss": 0.6823,
+      "step": 12379
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6591139370707796e-06,
+      "loss": 0.6687,
+      "step": 12380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.657931275587564e-06,
+      "loss": 0.5787,
+      "step": 12381
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6567489976693515e-06,
+      "loss": 0.6078,
+      "step": 12382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6555671033704964e-06,
+      "loss": 0.6217,
+      "step": 12383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6543855927453445e-06,
+      "loss": 0.6334,
+      "step": 12384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6532044658482192e-06,
+      "loss": 0.7017,
+      "step": 12385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.652023722733434e-06,
+      "loss": 0.6106,
+      "step": 12386
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6508433634552723e-06,
+      "loss": 0.5848,
+      "step": 12387
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.649663388068009e-06,
+      "loss": 0.6536,
+      "step": 12388
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.648483796625898e-06,
+      "loss": 0.6208,
+      "step": 12389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6473045891831796e-06,
+      "loss": 0.5758,
+      "step": 12390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6461257657940696e-06,
+      "loss": 0.6276,
+      "step": 12391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.644947326512772e-06,
+      "loss": 0.5697,
+      "step": 12392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643769271393473e-06,
+      "loss": 0.5936,
+      "step": 12393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6425916004903343e-06,
+      "loss": 0.5651,
+      "step": 12394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641414313857506e-06,
+      "loss": 0.577,
+      "step": 12395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.64023741154912e-06,
+      "loss": 0.5818,
+      "step": 12396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.639060893619294e-06,
+      "loss": 0.5636,
+      "step": 12397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637884760122117e-06,
+      "loss": 0.5558,
+      "step": 12398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6367090111116703e-06,
+      "loss": 0.6263,
+      "step": 12399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6355336466420125e-06,
+      "loss": 0.5853,
+      "step": 12400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6343586667671917e-06,
+      "loss": 0.6091,
+      "step": 12401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633184071541226e-06,
+      "loss": 0.6513,
+      "step": 12402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6320098610181276e-06,
+      "loss": 0.5897,
+      "step": 12403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6308360352518834e-06,
+      "loss": 0.6002,
+      "step": 12404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6296625942964672e-06,
+      "loss": 0.6132,
+      "step": 12405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.628489538205833e-06,
+      "loss": 0.5929,
+      "step": 12406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6273168670339168e-06,
+      "loss": 0.6147,
+      "step": 12407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.626144580834642e-06,
+      "loss": 0.5803,
+      "step": 12408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6249726796619026e-06,
+      "loss": 0.6093,
+      "step": 12409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6238011635695848e-06,
+      "loss": 0.5647,
+      "step": 12410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6226300326115565e-06,
+      "loss": 0.5676,
+      "step": 12411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6214592868416656e-06,
+      "loss": 0.5526,
+      "step": 12412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.62028892631374e-06,
+      "loss": 0.6455,
+      "step": 12413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6191189510815943e-06,
+      "loss": 0.5847,
+      "step": 12414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6179493611990218e-06,
+      "loss": 0.6012,
+      "step": 12415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6167801567198038e-06,
+      "loss": 0.5732,
+      "step": 12416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6156113376976944e-06,
+      "loss": 0.553,
+      "step": 12417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6144429041864394e-06,
+      "loss": 0.545,
+      "step": 12418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6132748562397603e-06,
+      "loss": 0.5958,
+      "step": 12419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6121071939113675e-06,
+      "loss": 0.6535,
+      "step": 12420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6109399172549454e-06,
+      "loss": 0.6619,
+      "step": 12421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.609773026324165e-06,
+      "loss": 0.557,
+      "step": 12422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.608606521172682e-06,
+      "loss": 0.5535,
+      "step": 12423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.60744040185413e-06,
+      "loss": 0.6699,
+      "step": 12424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6062746684221287e-06,
+      "loss": 0.5676,
+      "step": 12425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.605109320930276e-06,
+      "loss": 0.6417,
+      "step": 12426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.603944359432158e-06,
+      "loss": 0.5449,
+      "step": 12427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.602779783981333e-06,
+      "loss": 0.6356,
+      "step": 12428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6016155946313504e-06,
+      "loss": 0.6196,
+      "step": 12429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6004517914357407e-06,
+      "loss": 0.6194,
+      "step": 12430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.599288374448016e-06,
+      "loss": 0.5763,
+      "step": 12431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.598125343721665e-06,
+      "loss": 0.5375,
+      "step": 12432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.596962699310166e-06,
+      "loss": 0.5912,
+      "step": 12433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5958004412669778e-06,
+      "loss": 0.6328,
+      "step": 12434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5946385696455423e-06,
+      "loss": 0.5793,
+      "step": 12435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5934770844992762e-06,
+      "loss": 0.6198,
+      "step": 12436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5923159858815883e-06,
+      "loss": 0.5768,
+      "step": 12437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5911552738458643e-06,
+      "loss": 0.6098,
+      "step": 12438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5899949484454735e-06,
+      "loss": 0.6099,
+      "step": 12439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5888350097337665e-06,
+      "loss": 0.6442,
+      "step": 12440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5876754577640786e-06,
+      "loss": 0.5629,
+      "step": 12441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.586516292589726e-06,
+      "loss": 0.6468,
+      "step": 12442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.585357514264003e-06,
+      "loss": 0.5764,
+      "step": 12443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.584199122840192e-06,
+      "loss": 0.6439,
+      "step": 12444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.583041118371554e-06,
+      "loss": 0.66,
+      "step": 12445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5818835009113387e-06,
+      "loss": 0.4851,
+      "step": 12446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5807262705127667e-06,
+      "loss": 0.6221,
+      "step": 12447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5795694272290485e-06,
+      "loss": 0.5865,
+      "step": 12448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5784129711133763e-06,
+      "loss": 0.6026,
+      "step": 12449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.577256902218925e-06,
+      "loss": 0.6185,
+      "step": 12450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5761012205988478e-06,
+      "loss": 0.5753,
+      "step": 12451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5749459263062814e-06,
+      "loss": 0.6054,
+      "step": 12452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5737910193943494e-06,
+      "loss": 0.6734,
+      "step": 12453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5726364999161536e-06,
+      "loss": 0.5358,
+      "step": 12454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5714823679247733e-06,
+      "loss": 0.6147,
+      "step": 12455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5703286234732806e-06,
+      "loss": 0.6174,
+      "step": 12456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.569175266614721e-06,
+      "loss": 0.5717,
+      "step": 12457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5680222974021266e-06,
+      "loss": 0.5809,
+      "step": 12458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5668697158885105e-06,
+      "loss": 0.591,
+      "step": 12459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.565717522126868e-06,
+      "loss": 0.5817,
+      "step": 12460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5645657161701785e-06,
+      "loss": 0.6213,
+      "step": 12461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5634142980713973e-06,
+      "loss": 0.5993,
+      "step": 12462
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.562263267883468e-06,
+      "loss": 0.5923,
+      "step": 12463
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5611126256593157e-06,
+      "loss": 0.603,
+      "step": 12464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5599623714518474e-06,
+      "loss": 0.6142,
+      "step": 12465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.558812505313947e-06,
+      "loss": 0.6321,
+      "step": 12466
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.557663027298488e-06,
+      "loss": 0.5853,
+      "step": 12467
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5565139374583217e-06,
+      "loss": 0.6305,
+      "step": 12468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.555365235846287e-06,
+      "loss": 0.6273,
+      "step": 12469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5542169225151925e-06,
+      "loss": 0.5471,
+      "step": 12470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5530689975178438e-06,
+      "loss": 0.6097,
+      "step": 12471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.55192146090702e-06,
+      "loss": 0.5707,
+      "step": 12472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5507743127354837e-06,
+      "loss": 0.567,
+      "step": 12473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5496275530559814e-06,
+      "loss": 0.661,
+      "step": 12474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.548481181921241e-06,
+      "loss": 0.6025,
+      "step": 12475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.547335199383972e-06,
+      "loss": 0.5473,
+      "step": 12476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.546189605496866e-06,
+      "loss": 0.5533,
+      "step": 12477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.545044400312595e-06,
+      "loss": 0.6379,
+      "step": 12478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.543899583883818e-06,
+      "loss": 0.5825,
+      "step": 12479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5427551562631724e-06,
+      "loss": 0.6173,
+      "step": 12480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.541611117503278e-06,
+      "loss": 0.6074,
+      "step": 12481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5404674676567365e-06,
+      "loss": 0.5523,
+      "step": 12482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5393242067761326e-06,
+      "loss": 0.5727,
+      "step": 12483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.538181334914037e-06,
+      "loss": 0.6381,
+      "step": 12484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5370388521229927e-06,
+      "loss": 0.5465,
+      "step": 12485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.535896758455533e-06,
+      "loss": 0.5868,
+      "step": 12486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5347550539641733e-06,
+      "loss": 0.5754,
+      "step": 12487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.533613738701404e-06,
+      "loss": 0.6021,
+      "step": 12488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.532472812719705e-06,
+      "loss": 0.5602,
+      "step": 12489
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5313322760715355e-06,
+      "loss": 0.5891,
+      "step": 12490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5301921288093365e-06,
+      "loss": 0.6132,
+      "step": 12491
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5290523709855318e-06,
+      "loss": 0.6054,
+      "step": 12492
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5279130026525267e-06,
+      "loss": 0.59,
+      "step": 12493
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5267740238627104e-06,
+      "loss": 0.6309,
+      "step": 12494
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.525635434668453e-06,
+      "loss": 0.6046,
+      "step": 12495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5244972351221033e-06,
+      "loss": 0.6195,
+      "step": 12496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.523359425275996e-06,
+      "loss": 0.5924,
+      "step": 12497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5222220051824484e-06,
+      "loss": 0.5829,
+      "step": 12498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5210849748937603e-06,
+      "loss": 0.6546,
+      "step": 12499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.519948334462208e-06,
+      "loss": 0.6128,
+      "step": 12500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5188120839400554e-06,
+      "loss": 0.6425,
+      "step": 12501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.517676223379547e-06,
+      "loss": 0.6142,
+      "step": 12502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.516540752832911e-06,
+      "loss": 0.6065,
+      "step": 12503
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5154056723523535e-06,
+      "loss": 0.6045,
+      "step": 12504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5142709819900647e-06,
+      "loss": 0.5981,
+      "step": 12505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5131366817982183e-06,
+      "loss": 0.5537,
+      "step": 12506
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5120027718289687e-06,
+      "loss": 0.5269,
+      "step": 12507
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5108692521344525e-06,
+      "loss": 0.6524,
+      "step": 12508
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5097361227667893e-06,
+      "loss": 0.6204,
+      "step": 12509
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.508603383778082e-06,
+      "loss": 0.6661,
+      "step": 12510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.507471035220408e-06,
+      "loss": 0.5863,
+      "step": 12511
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5063390771458353e-06,
+      "loss": 0.5148,
+      "step": 12512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.505207509606411e-06,
+      "loss": 0.6587,
+      "step": 12513
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5040763326541662e-06,
+      "loss": 0.584,
+      "step": 12514
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.502945546341108e-06,
+      "loss": 0.598,
+      "step": 12515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5018151507192303e-06,
+      "loss": 0.6042,
+      "step": 12516
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.500685145840509e-06,
+      "loss": 0.6198,
+      "step": 12517
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4995555317569045e-06,
+      "loss": 0.5393,
+      "step": 12518
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4984263085203497e-06,
+      "loss": 0.6061,
+      "step": 12519
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4972974761827687e-06,
+      "loss": 0.557,
+      "step": 12520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4961690347960676e-06,
+      "loss": 0.6253,
+      "step": 12521
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.495040984412128e-06,
+      "loss": 0.6008,
+      "step": 12522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4939133250828174e-06,
+      "loss": 0.5513,
+      "step": 12523
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.492786056859986e-06,
+      "loss": 0.6265,
+      "step": 12524
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.491659179795466e-06,
+      "loss": 0.5472,
+      "step": 12525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.490532693941069e-06,
+      "loss": 0.5574,
+      "step": 12526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4894065993485917e-06,
+      "loss": 0.6248,
+      "step": 12527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.488280896069811e-06,
+      "loss": 0.6457,
+      "step": 12528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4871555841564886e-06,
+      "loss": 0.6919,
+      "step": 12529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4860306636603627e-06,
+      "loss": 0.6248,
+      "step": 12530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.484906134633156e-06,
+      "loss": 0.5181,
+      "step": 12531
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4837819971265766e-06,
+      "loss": 0.5625,
+      "step": 12532
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4826582511923137e-06,
+      "loss": 0.5283,
+      "step": 12533
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4815348968820306e-06,
+      "loss": 0.5753,
+      "step": 12534
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.480411934247382e-06,
+      "loss": 0.6114,
+      "step": 12535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.479289363340002e-06,
+      "loss": 0.5349,
+      "step": 12536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4781671842115062e-06,
+      "loss": 0.619,
+      "step": 12537
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.47704539691349e-06,
+      "loss": 0.5825,
+      "step": 12538
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4759240014975338e-06,
+      "loss": 0.5647,
+      "step": 12539
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4748029980151977e-06,
+      "loss": 0.5884,
+      "step": 12540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4736823865180273e-06,
+      "loss": 0.5937,
+      "step": 12541
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.472562167057545e-06,
+      "loss": 0.5834,
+      "step": 12542
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4714423396852614e-06,
+      "loss": 0.6006,
+      "step": 12543
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4703229044526655e-06,
+      "loss": 0.6205,
+      "step": 12544
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4692038614112257e-06,
+      "loss": 0.5832,
+      "step": 12545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4680852106123957e-06,
+      "loss": 0.5579,
+      "step": 12546
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4669669521076113e-06,
+      "loss": 0.5377,
+      "step": 12547
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4658490859482923e-06,
+      "loss": 0.6136,
+      "step": 12548
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.464731612185832e-06,
+      "loss": 0.575,
+      "step": 12549
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4636145308716153e-06,
+      "loss": 0.5872,
+      "step": 12550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4624978420570044e-06,
+      "loss": 0.557,
+      "step": 12551
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4613815457933466e-06,
+      "loss": 0.5996,
+      "step": 12552
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4602656421319637e-06,
+      "loss": 0.5724,
+      "step": 12553
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4591501311241673e-06,
+      "loss": 0.6417,
+      "step": 12554
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4580350128212485e-06,
+      "loss": 0.5447,
+      "step": 12555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.456920287274479e-06,
+      "loss": 0.5606,
+      "step": 12556
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.455805954535114e-06,
+      "loss": 0.5867,
+      "step": 12557
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.45469201465439e-06,
+      "loss": 0.5994,
+      "step": 12558
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4535784676835274e-06,
+      "loss": 0.5653,
+      "step": 12559
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.452465313673723e-06,
+      "loss": 0.6479,
+      "step": 12560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4513525526761605e-06,
+      "loss": 0.6314,
+      "step": 12561
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4502401847420056e-06,
+      "loss": 0.5846,
+      "step": 12562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4491282099224044e-06,
+      "loss": 0.6416,
+      "step": 12563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4480166282684836e-06,
+      "loss": 0.6122,
+      "step": 12564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.446905439831353e-06,
+      "loss": 0.6381,
+      "step": 12565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.445794644662105e-06,
+      "loss": 0.6246,
+      "step": 12566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.444684242811818e-06,
+      "loss": 0.6077,
+      "step": 12567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4435742343315404e-06,
+      "loss": 0.6365,
+      "step": 12568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4424646192723136e-06,
+      "loss": 0.6155,
+      "step": 12569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.441355397685157e-06,
+      "loss": 0.6071,
+      "step": 12570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.440246569621074e-06,
+      "loss": 0.5876,
+      "step": 12571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.439138135131045e-06,
+      "loss": 0.643,
+      "step": 12572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4380300942660374e-06,
+      "loss": 0.6841,
+      "step": 12573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.436922447076996e-06,
+      "loss": 0.5788,
+      "step": 12574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4358151936148533e-06,
+      "loss": 0.5766,
+      "step": 12575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4347083339305177e-06,
+      "loss": 0.6067,
+      "step": 12576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.433601868074884e-06,
+      "loss": 0.5993,
+      "step": 12577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4324957960988286e-06,
+      "loss": 0.5555,
+      "step": 12578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.431390118053203e-06,
+      "loss": 0.5605,
+      "step": 12579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4302848339888495e-06,
+      "loss": 0.5868,
+      "step": 12580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.429179943956588e-06,
+      "loss": 0.6167,
+      "step": 12581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4280754480072236e-06,
+      "loss": 0.6063,
+      "step": 12582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4269713461915345e-06,
+      "loss": 0.6122,
+      "step": 12583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.425867638560291e-06,
+      "loss": 0.5977,
+      "step": 12584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4247643251642404e-06,
+      "loss": 0.5832,
+      "step": 12585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4236614060541143e-06,
+      "loss": 0.5453,
+      "step": 12586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4225588812806215e-06,
+      "loss": 0.5408,
+      "step": 12587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.421456750894455e-06,
+      "loss": 0.5754,
+      "step": 12588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.420355014946294e-06,
+      "loss": 0.6037,
+      "step": 12589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4192536734867937e-06,
+      "loss": 0.5967,
+      "step": 12590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.418152726566594e-06,
+      "loss": 0.6203,
+      "step": 12591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4170521742363152e-06,
+      "loss": 0.6092,
+      "step": 12592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.415952016546562e-06,
+      "loss": 0.6002,
+      "step": 12593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.414852253547918e-06,
+      "loss": 0.6653,
+      "step": 12594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4137528852909485e-06,
+      "loss": 0.532,
+      "step": 12595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.412653911826204e-06,
+      "loss": 0.6318,
+      "step": 12596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4115553332042156e-06,
+      "loss": 0.6205,
+      "step": 12597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4104571494754926e-06,
+      "loss": 0.5269,
+      "step": 12598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4093593606905308e-06,
+      "loss": 0.6313,
+      "step": 12599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.408261966899805e-06,
+      "loss": 0.5625,
+      "step": 12600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.407164968153777e-06,
+      "loss": 0.6204,
+      "step": 12601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4060683645028804e-06,
+      "loss": 0.5591,
+      "step": 12602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4049721559975404e-06,
+      "loss": 0.5337,
+      "step": 12603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4038763426881584e-06,
+      "loss": 0.55,
+      "step": 12604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4027809246251223e-06,
+      "loss": 0.6248,
+      "step": 12605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4016859018587957e-06,
+      "loss": 0.5547,
+      "step": 12606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4005912744395275e-06,
+      "loss": 0.5483,
+      "step": 12607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3994970424176502e-06,
+      "loss": 0.6281,
+      "step": 12608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3984032058434746e-06,
+      "loss": 0.5658,
+      "step": 12609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.397309764767296e-06,
+      "loss": 0.5798,
+      "step": 12610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3962167192393893e-06,
+      "loss": 0.5683,
+      "step": 12611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3951240693100165e-06,
+      "loss": 0.6442,
+      "step": 12612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3940318150294106e-06,
+      "loss": 0.5976,
+      "step": 12613
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3929399564477952e-06,
+      "loss": 0.5982,
+      "step": 12614
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3918484936153753e-06,
+      "loss": 0.5896,
+      "step": 12615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3907574265823374e-06,
+      "loss": 0.6225,
+      "step": 12616
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.389666755398843e-06,
+      "loss": 0.5921,
+      "step": 12617
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3885764801150447e-06,
+      "loss": 0.5547,
+      "step": 12618
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3874866007810705e-06,
+      "loss": 0.6354,
+      "step": 12619
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.386397117447038e-06,
+      "loss": 0.5684,
+      "step": 12620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3853080301630336e-06,
+      "loss": 0.588,
+      "step": 12621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.384219338979137e-06,
+      "loss": 0.609,
+      "step": 12622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.383131043945406e-06,
+      "loss": 0.5984,
+      "step": 12623
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3820431451118788e-06,
+      "loss": 0.6173,
+      "step": 12624
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3809556425285775e-06,
+      "loss": 0.5897,
+      "step": 12625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3798685362455043e-06,
+      "loss": 0.5723,
+      "step": 12626
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.378781826312647e-06,
+      "loss": 0.5952,
+      "step": 12627
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.377695512779967e-06,
+      "loss": 0.6298,
+      "step": 12628
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3766095956974158e-06,
+      "loss": 0.5973,
+      "step": 12629
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3755240751149224e-06,
+      "loss": 0.6002,
+      "step": 12630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.374438951082401e-06,
+      "loss": 0.55,
+      "step": 12631
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.373354223649741e-06,
+      "loss": 0.606,
+      "step": 12632
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3722698928668187e-06,
+      "loss": 0.6424,
+      "step": 12633
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.371185958783493e-06,
+      "loss": 0.6055,
+      "step": 12634
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.370102421449605e-06,
+      "loss": 0.5232,
+      "step": 12635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.369019280914969e-06,
+      "loss": 0.5678,
+      "step": 12636
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3679365372293916e-06,
+      "loss": 0.5707,
+      "step": 12637
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3668541904426577e-06,
+      "loss": 0.6526,
+      "step": 12638
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3657722406045292e-06,
+      "loss": 0.5908,
+      "step": 12639
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.364690687764757e-06,
+      "loss": 0.5745,
+      "step": 12640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.363609531973068e-06,
+      "loss": 0.6148,
+      "step": 12641
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3625287732791758e-06,
+      "loss": 0.6083,
+      "step": 12642
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3614484117327717e-06,
+      "loss": 0.5926,
+      "step": 12643
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.360368447383532e-06,
+      "loss": 0.5918,
+      "step": 12644
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3592888802811112e-06,
+      "loss": 0.5662,
+      "step": 12645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.35820971047515e-06,
+      "loss": 0.5047,
+      "step": 12646
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.357130938015264e-06,
+      "loss": 0.641,
+      "step": 12647
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3560525629510567e-06,
+      "loss": 0.6055,
+      "step": 12648
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.354974585332113e-06,
+      "loss": 0.6182,
+      "step": 12649
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3538970052079976e-06,
+      "loss": 0.5667,
+      "step": 12650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.352819822628254e-06,
+      "loss": 0.6074,
+      "step": 12651
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3517430376424133e-06,
+      "loss": 0.6012,
+      "step": 12652
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.350666650299985e-06,
+      "loss": 0.6204,
+      "step": 12653
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3495906606504625e-06,
+      "loss": 0.6298,
+      "step": 12654
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3485150687433165e-06,
+      "loss": 0.6019,
+      "step": 12655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3474398746280037e-06,
+      "loss": 0.6339,
+      "step": 12656
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3463650783539615e-06,
+      "loss": 0.6219,
+      "step": 12657
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.345290679970608e-06,
+      "loss": 0.6256,
+      "step": 12658
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3442166795273434e-06,
+      "loss": 0.6177,
+      "step": 12659
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3431430770735497e-06,
+      "loss": 0.6129,
+      "step": 12660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3420698726585945e-06,
+      "loss": 0.6473,
+      "step": 12661
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3409970663318184e-06,
+      "loss": 0.5351,
+      "step": 12662
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3399246581425496e-06,
+      "loss": 0.5877,
+      "step": 12663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3388526481400964e-06,
+      "loss": 0.5463,
+      "step": 12664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.337781036373754e-06,
+      "loss": 0.5581,
+      "step": 12665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3367098228927897e-06,
+      "loss": 0.5447,
+      "step": 12666
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3356390077464584e-06,
+      "loss": 0.6276,
+      "step": 12667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3345685909839978e-06,
+      "loss": 0.5586,
+      "step": 12668
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.333498572654624e-06,
+      "loss": 0.5445,
+      "step": 12669
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.332428952807535e-06,
+      "loss": 0.6047,
+      "step": 12670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3313597314919124e-06,
+      "loss": 0.5922,
+      "step": 12671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3302909087569215e-06,
+      "loss": 0.5791,
+      "step": 12672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3292224846516996e-06,
+      "loss": 0.6732,
+      "step": 12673
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3281544592253767e-06,
+      "loss": 0.5092,
+      "step": 12674
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3270868325270616e-06,
+      "loss": 0.5868,
+      "step": 12675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.326019604605839e-06,
+      "loss": 0.5381,
+      "step": 12676
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3249527755107836e-06,
+      "loss": 0.6224,
+      "step": 12677
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3238863452909456e-06,
+      "loss": 0.5935,
+      "step": 12678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.32282031399536e-06,
+      "loss": 0.689,
+      "step": 12679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3217546816730443e-06,
+      "loss": 0.5188,
+      "step": 12680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3206894483729927e-06,
+      "loss": 0.5848,
+      "step": 12681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3196246141441837e-06,
+      "loss": 0.5784,
+      "step": 12682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.318560179035581e-06,
+      "loss": 0.5468,
+      "step": 12683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3174961430961265e-06,
+      "loss": 0.5556,
+      "step": 12684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3164325063747406e-06,
+      "loss": 0.5852,
+      "step": 12685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.315369268920332e-06,
+      "loss": 0.6033,
+      "step": 12686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3143064307817873e-06,
+      "loss": 0.575,
+      "step": 12687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3132439920079776e-06,
+      "loss": 0.538,
+      "step": 12688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3121819526477485e-06,
+      "loss": 0.5629,
+      "step": 12689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.311120312749935e-06,
+      "loss": 0.5947,
+      "step": 12690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.310059072363351e-06,
+      "loss": 0.5612,
+      "step": 12691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.308998231536792e-06,
+      "loss": 0.6533,
+      "step": 12692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3079377903190347e-06,
+      "loss": 0.58,
+      "step": 12693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3068777487588367e-06,
+      "loss": 0.5688,
+      "step": 12694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3058181069049425e-06,
+      "loss": 0.619,
+      "step": 12695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3047588648060683e-06,
+      "loss": 0.6818,
+      "step": 12696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3037000225109206e-06,
+      "loss": 0.6331,
+      "step": 12697
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3026415800681836e-06,
+      "loss": 0.6114,
+      "step": 12698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3015835375265273e-06,
+      "loss": 0.5939,
+      "step": 12699
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3005258949345942e-06,
+      "loss": 0.5966,
+      "step": 12700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.299468652341018e-06,
+      "loss": 0.5722,
+      "step": 12701
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2984118097944098e-06,
+      "loss": 0.5647,
+      "step": 12702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2973553673433647e-06,
+      "loss": 0.607,
+      "step": 12703
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.296299325036454e-06,
+      "loss": 0.6009,
+      "step": 12704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2952436829222347e-06,
+      "loss": 0.573,
+      "step": 12705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2941884410492487e-06,
+      "loss": 0.6056,
+      "step": 12706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.29313359946601e-06,
+      "loss": 0.6297,
+      "step": 12707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2920791582210224e-06,
+      "loss": 0.5297,
+      "step": 12708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2910251173627684e-06,
+      "loss": 0.5615,
+      "step": 12709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.289971476939714e-06,
+      "loss": 0.6339,
+      "step": 12710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2889182370003028e-06,
+      "loss": 0.6092,
+      "step": 12711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2878653975929646e-06,
+      "loss": 0.5777,
+      "step": 12712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.286812958766106e-06,
+      "loss": 0.612,
+      "step": 12713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.285760920568122e-06,
+      "loss": 0.6442,
+      "step": 12714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2847092830473796e-06,
+      "loss": 0.5412,
+      "step": 12715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2836580462522338e-06,
+      "loss": 0.6098,
+      "step": 12716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.282607210231024e-06,
+      "loss": 0.5783,
+      "step": 12717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2815567750320645e-06,
+      "loss": 0.6288,
+      "step": 12718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2805067407036528e-06,
+      "loss": 0.5771,
+      "step": 12719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2794571072940696e-06,
+      "loss": 0.6309,
+      "step": 12720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.278407874851577e-06,
+      "loss": 0.5663,
+      "step": 12721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.277359043424422e-06,
+      "loss": 0.6211,
+      "step": 12722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2763106130608226e-06,
+      "loss": 0.5425,
+      "step": 12723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2752625838089894e-06,
+      "loss": 0.6046,
+      "step": 12724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.274214955717109e-06,
+      "loss": 0.5828,
+      "step": 12725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.273167728833351e-06,
+      "loss": 0.4862,
+      "step": 12726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2721209032058676e-06,
+      "loss": 0.5819,
+      "step": 12727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.27107447888279e-06,
+      "loss": 0.532,
+      "step": 12728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2700284559122363e-06,
+      "loss": 0.6684,
+      "step": 12729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.268982834342296e-06,
+      "loss": 0.6131,
+      "step": 12730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.267937614221051e-06,
+      "loss": 0.654,
+      "step": 12731
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.266892795596557e-06,
+      "loss": 0.5732,
+      "step": 12732
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2658483785168583e-06,
+      "loss": 0.617,
+      "step": 12733
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.264804363029972e-06,
+      "loss": 0.6347,
+      "step": 12734
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.263760749183904e-06,
+      "loss": 0.6332,
+      "step": 12735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2627175370266397e-06,
+      "loss": 0.5736,
+      "step": 12736
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2616747266061466e-06,
+      "loss": 0.5423,
+      "step": 12737
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2606323179703686e-06,
+      "loss": 0.6088,
+      "step": 12738
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.259590311167238e-06,
+      "loss": 0.5618,
+      "step": 12739
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.258548706244667e-06,
+      "loss": 0.5372,
+      "step": 12740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2575075032505442e-06,
+      "loss": 0.5818,
+      "step": 12741
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2564667022327483e-06,
+      "loss": 0.6027,
+      "step": 12742
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2554263032391322e-06,
+      "loss": 0.6541,
+      "step": 12743
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2543863063175355e-06,
+      "loss": 0.6051,
+      "step": 12744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2533467115157723e-06,
+      "loss": 0.614,
+      "step": 12745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2523075188816467e-06,
+      "loss": 0.5178,
+      "step": 12746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2512687284629397e-06,
+      "loss": 0.6094,
+      "step": 12747
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2502303403074155e-06,
+      "loss": 0.6384,
+      "step": 12748
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2491923544628147e-06,
+      "loss": 0.6462,
+      "step": 12749
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.248154770976866e-06,
+      "loss": 0.5376,
+      "step": 12750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2471175898972772e-06,
+      "loss": 0.6466,
+      "step": 12751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2460808112717393e-06,
+      "loss": 0.6284,
+      "step": 12752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2450444351479197e-06,
+      "loss": 0.5385,
+      "step": 12753
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2440084615734704e-06,
+      "loss": 0.6116,
+      "step": 12754
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.242972890596028e-06,
+      "loss": 0.6367,
+      "step": 12755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2419377222632075e-06,
+      "loss": 0.6056,
+      "step": 12756
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2409029566226026e-06,
+      "loss": 0.6297,
+      "step": 12757
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2398685937217923e-06,
+      "loss": 0.555,
+      "step": 12758
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2388346336083378e-06,
+      "loss": 0.5098,
+      "step": 12759
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2378010763297785e-06,
+      "loss": 0.6395,
+      "step": 12760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.236767921933638e-06,
+      "loss": 0.5793,
+      "step": 12761
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2357351704674204e-06,
+      "loss": 0.6308,
+      "step": 12762
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2347028219786124e-06,
+      "loss": 0.6241,
+      "step": 12763
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2336708765146774e-06,
+      "loss": 0.502,
+      "step": 12764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2326393341230658e-06,
+      "loss": 0.5657,
+      "step": 12765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2316081948512084e-06,
+      "loss": 0.5782,
+      "step": 12766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2305774587465166e-06,
+      "loss": 0.5903,
+      "step": 12767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2295471258563816e-06,
+      "loss": 0.6069,
+      "step": 12768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.228517196228177e-06,
+      "loss": 0.6195,
+      "step": 12769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.227487669909262e-06,
+      "loss": 0.5599,
+      "step": 12770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2264585469469724e-06,
+      "loss": 0.6201,
+      "step": 12771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.225429827388624e-06,
+      "loss": 0.5767,
+      "step": 12772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.224401511281521e-06,
+      "loss": 0.6046,
+      "step": 12773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.223373598672942e-06,
+      "loss": 0.6075,
+      "step": 12774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.222346089610151e-06,
+      "loss": 0.6507,
+      "step": 12775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2213189841403926e-06,
+      "loss": 0.5604,
+      "step": 12776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2202922823108933e-06,
+      "loss": 0.5595,
+      "step": 12777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2192659841688625e-06,
+      "loss": 0.5943,
+      "step": 12778
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.218240089761483e-06,
+      "loss": 0.5782,
+      "step": 12779
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2172145991359297e-06,
+      "loss": 0.6425,
+      "step": 12780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2161895123393518e-06,
+      "loss": 0.612,
+      "step": 12781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2151648294188865e-06,
+      "loss": 0.5311,
+      "step": 12782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.214140550421643e-06,
+      "loss": 0.6327,
+      "step": 12783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2131166753947184e-06,
+      "loss": 0.5522,
+      "step": 12784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.212093204385193e-06,
+      "loss": 0.6079,
+      "step": 12785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2110701374401256e-06,
+      "loss": 0.5708,
+      "step": 12786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2100474746065525e-06,
+      "loss": 0.553,
+      "step": 12787
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.209025215931497e-06,
+      "loss": 0.6714,
+      "step": 12788
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2080033614619623e-06,
+      "loss": 0.5885,
+      "step": 12789
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2069819112449366e-06,
+      "loss": 0.6304,
+      "step": 12790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.205960865327379e-06,
+      "loss": 0.5951,
+      "step": 12791
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2049402237562415e-06,
+      "loss": 0.5632,
+      "step": 12792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.20391998657845e-06,
+      "loss": 0.6082,
+      "step": 12793
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.202900153840917e-06,
+      "loss": 0.6349,
+      "step": 12794
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.201880725590533e-06,
+      "loss": 0.6022,
+      "step": 12795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.200861701874171e-06,
+      "loss": 0.5694,
+      "step": 12796
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.199843082738686e-06,
+      "loss": 0.6409,
+      "step": 12797
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.198824868230912e-06,
+      "loss": 0.6217,
+      "step": 12798
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1978070583976677e-06,
+      "loss": 0.5667,
+      "step": 12799
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.19678965328575e-06,
+      "loss": 0.5775,
+      "step": 12800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1957726529419424e-06,
+      "loss": 0.5224,
+      "step": 12801
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.194756057413001e-06,
+      "loss": 0.5713,
+      "step": 12802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.193739866745671e-06,
+      "loss": 0.6054,
+      "step": 12803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1927240809866765e-06,
+      "loss": 0.6126,
+      "step": 12804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1917087001827254e-06,
+      "loss": 0.6024,
+      "step": 12805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1906937243804994e-06,
+      "loss": 0.6305,
+      "step": 12806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.189679153626669e-06,
+      "loss": 0.5523,
+      "step": 12807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.188664987967885e-06,
+      "loss": 0.6079,
+      "step": 12808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1876512274507755e-06,
+      "loss": 0.6039,
+      "step": 12809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.186637872121956e-06,
+      "loss": 0.5966,
+      "step": 12810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.185624922028018e-06,
+      "loss": 0.6147,
+      "step": 12811
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1846123772155405e-06,
+      "loss": 0.5815,
+      "step": 12812
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.183600237731074e-06,
+      "loss": 0.5599,
+      "step": 12813
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.182588503621158e-06,
+      "loss": 0.6058,
+      "step": 12814
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1815771749323145e-06,
+      "loss": 0.6259,
+      "step": 12815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.180566251711043e-06,
+      "loss": 0.6042,
+      "step": 12816
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1795557340038232e-06,
+      "loss": 0.6397,
+      "step": 12817
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.178545621857119e-06,
+      "loss": 0.6278,
+      "step": 12818
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.177535915317376e-06,
+      "loss": 0.6003,
+      "step": 12819
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1765266144310205e-06,
+      "loss": 0.5455,
+      "step": 12820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.175517719244458e-06,
+      "loss": 0.5608,
+      "step": 12821
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1745092298040784e-06,
+      "loss": 0.5942,
+      "step": 12822
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1735011461562516e-06,
+      "loss": 0.5086,
+      "step": 12823
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1724934683473277e-06,
+      "loss": 0.6311,
+      "step": 12824
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.17148619642364e-06,
+      "loss": 0.553,
+      "step": 12825
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.170479330431503e-06,
+      "loss": 0.5393,
+      "step": 12826
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.169472870417211e-06,
+      "loss": 0.6035,
+      "step": 12827
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.168466816427042e-06,
+      "loss": 0.5902,
+      "step": 12828
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.167461168507252e-06,
+      "loss": 0.6467,
+      "step": 12829
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1664559267040823e-06,
+      "loss": 0.62,
+      "step": 12830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.165451091063755e-06,
+      "loss": 0.5641,
+      "step": 12831
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1644466616324667e-06,
+      "loss": 0.5924,
+      "step": 12832
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1634426384564046e-06,
+      "loss": 0.6268,
+      "step": 12833
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1624390215817316e-06,
+      "loss": 0.5123,
+      "step": 12834
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1614358110545975e-06,
+      "loss": 0.6057,
+      "step": 12835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1604330069211233e-06,
+      "loss": 0.6254,
+      "step": 12836
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1594306092274217e-06,
+      "loss": 0.6225,
+      "step": 12837
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1584286180195812e-06,
+      "loss": 0.5537,
+      "step": 12838
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.157427033343674e-06,
+      "loss": 0.6054,
+      "step": 12839
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1564258552457516e-06,
+      "loss": 0.5596,
+      "step": 12840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1554250837718473e-06,
+      "loss": 0.654,
+      "step": 12841
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.154424718967977e-06,
+      "loss": 0.6153,
+      "step": 12842
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1534247608801373e-06,
+      "loss": 0.619,
+      "step": 12843
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1524252095543042e-06,
+      "loss": 0.6165,
+      "step": 12844
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1514260650364394e-06,
+      "loss": 0.5418,
+      "step": 12845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.150427327372483e-06,
+      "loss": 0.622,
+      "step": 12846
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.149428996608354e-06,
+      "loss": 0.6243,
+      "step": 12847
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.148431072789956e-06,
+      "loss": 0.5518,
+      "step": 12848
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1474335559631734e-06,
+      "loss": 0.5905,
+      "step": 12849
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.146436446173875e-06,
+      "loss": 0.5551,
+      "step": 12850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.145439743467902e-06,
+      "loss": 0.6593,
+      "step": 12851
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1444434478910847e-06,
+      "loss": 0.6416,
+      "step": 12852
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1434475594892323e-06,
+      "loss": 0.5759,
+      "step": 12853
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1424520783081383e-06,
+      "loss": 0.5888,
+      "step": 12854
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.141457004393569e-06,
+      "loss": 0.5475,
+      "step": 12855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1404623377912803e-06,
+      "loss": 0.606,
+      "step": 12856
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.139468078547009e-06,
+      "loss": 0.5231,
+      "step": 12857
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.138474226706466e-06,
+      "loss": 0.5973,
+      "step": 12858
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.137480782315351e-06,
+      "loss": 0.5654,
+      "step": 12859
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.136487745419341e-06,
+      "loss": 0.5938,
+      "step": 12860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.135495116064097e-06,
+      "loss": 0.5569,
+      "step": 12861
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1345028942952587e-06,
+      "loss": 0.5302,
+      "step": 12862
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.133511080158448e-06,
+      "loss": 0.6447,
+      "step": 12863
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1325196736992682e-06,
+      "loss": 0.5747,
+      "step": 12864
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.131528674963307e-06,
+      "loss": 0.6349,
+      "step": 12865
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.130538083996123e-06,
+      "loss": 0.6059,
+      "step": 12866
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1295479008432687e-06,
+      "loss": 0.6471,
+      "step": 12867
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1285581255502708e-06,
+      "loss": 0.5494,
+      "step": 12868
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1275687581626404e-06,
+      "loss": 0.5488,
+      "step": 12869
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1265797987258654e-06,
+      "loss": 0.633,
+      "step": 12870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1255912472854182e-06,
+      "loss": 0.5788,
+      "step": 12871
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.124603103886752e-06,
+      "loss": 0.5402,
+      "step": 12872
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.123615368575306e-06,
+      "loss": 0.5322,
+      "step": 12873
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1226280413964884e-06,
+      "loss": 0.6129,
+      "step": 12874
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1216411223957e-06,
+      "loss": 0.549,
+      "step": 12875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.120654611618319e-06,
+      "loss": 0.5468,
+      "step": 12876
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1196685091097034e-06,
+      "loss": 0.5834,
+      "step": 12877
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1186828149151962e-06,
+      "loss": 0.5518,
+      "step": 12878
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.117697529080117e-06,
+      "loss": 0.568,
+      "step": 12879
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1167126516497717e-06,
+      "loss": 0.5986,
+      "step": 12880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.11572818266944e-06,
+      "loss": 0.6151,
+      "step": 12881
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1147441221843904e-06,
+      "loss": 0.6371,
+      "step": 12882
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1137604702398684e-06,
+      "loss": 0.6061,
+      "step": 12883
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.112777226881105e-06,
+      "loss": 0.5831,
+      "step": 12884
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1117943921533058e-06,
+      "loss": 0.6196,
+      "step": 12885
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1108119661016612e-06,
+      "loss": 0.6059,
+      "step": 12886
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1098299487713447e-06,
+      "loss": 0.6001,
+      "step": 12887
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1088483402075097e-06,
+      "loss": 0.6217,
+      "step": 12888
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1078671404552876e-06,
+      "loss": 0.6443,
+      "step": 12889
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1068863495597937e-06,
+      "loss": 0.5407,
+      "step": 12890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1059059675661275e-06,
+      "loss": 0.6054,
+      "step": 12891
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1049259945193622e-06,
+      "loss": 0.5913,
+      "step": 12892
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1039464304645596e-06,
+      "loss": 0.6169,
+      "step": 12893
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.102967275446759e-06,
+      "loss": 0.6472,
+      "step": 12894
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1019885295109797e-06,
+      "loss": 0.6074,
+      "step": 12895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1010101927022277e-06,
+      "loss": 0.5439,
+      "step": 12896
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1000322650654837e-06,
+      "loss": 0.5872,
+      "step": 12897
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.099054746645713e-06,
+      "loss": 0.6139,
+      "step": 12898
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0980776374878653e-06,
+      "loss": 0.592,
+      "step": 12899
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.097100937636861e-06,
+      "loss": 0.5445,
+      "step": 12900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0961246471376115e-06,
+      "loss": 0.6005,
+      "step": 12901
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0951487660350065e-06,
+      "loss": 0.5692,
+      "step": 12902
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.094173294373919e-06,
+      "loss": 0.5587,
+      "step": 12903
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.093198232199196e-06,
+      "loss": 0.5244,
+      "step": 12904
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.092223579555672e-06,
+      "loss": 0.5782,
+      "step": 12905
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0912493364881624e-06,
+      "loss": 0.5661,
+      "step": 12906
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0902755030414644e-06,
+      "loss": 0.6243,
+      "step": 12907
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0893020792603502e-06,
+      "loss": 0.6209,
+      "step": 12908
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.088329065189579e-06,
+      "loss": 0.5094,
+      "step": 12909
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0873564608738893e-06,
+      "loss": 0.5832,
+      "step": 12910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.086384266358003e-06,
+      "loss": 0.5763,
+      "step": 12911
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0854124816866185e-06,
+      "loss": 0.5385,
+      "step": 12912
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.08444110690442e-06,
+      "loss": 0.5497,
+      "step": 12913
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0834701420560724e-06,
+      "loss": 0.6147,
+      "step": 12914
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0824995871862164e-06,
+      "loss": 0.5446,
+      "step": 12915
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0815294423394797e-06,
+      "loss": 0.6281,
+      "step": 12916
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0805597075604676e-06,
+      "loss": 0.5788,
+      "step": 12917
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0795903828937737e-06,
+      "loss": 0.5615,
+      "step": 12918
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.07862146838396e-06,
+      "loss": 0.5158,
+      "step": 12919
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0776529640755793e-06,
+      "loss": 0.6268,
+      "step": 12920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0766848700131647e-06,
+      "loss": 0.6607,
+      "step": 12921
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0757171862412294e-06,
+      "loss": 0.6436,
+      "step": 12922
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0747499128042626e-06,
+      "loss": 0.6136,
+      "step": 12923
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0737830497467425e-06,
+      "loss": 0.6169,
+      "step": 12924
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0728165971131242e-06,
+      "loss": 0.5908,
+      "step": 12925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.071850554947846e-06,
+      "loss": 0.6365,
+      "step": 12926
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.070884923295324e-06,
+      "loss": 0.5479,
+      "step": 12927
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0699197021999596e-06,
+      "loss": 0.5634,
+      "step": 12928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0689548917061342e-06,
+      "loss": 0.5097,
+      "step": 12929
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.067990491858206e-06,
+      "loss": 0.5595,
+      "step": 12930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0670265027005185e-06,
+      "loss": 0.5392,
+      "step": 12931
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0660629242773977e-06,
+      "loss": 0.58,
+      "step": 12932
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0650997566331487e-06,
+      "loss": 0.6279,
+      "step": 12933
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.064136999812054e-06,
+      "loss": 0.6193,
+      "step": 12934
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.063174653858384e-06,
+      "loss": 0.5871,
+      "step": 12935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0622127188163846e-06,
+      "loss": 0.6344,
+      "step": 12936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0612511947302894e-06,
+      "loss": 0.6111,
+      "step": 12937
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.060290081644303e-06,
+      "loss": 0.601,
+      "step": 12938
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0593293796026204e-06,
+      "loss": 0.5298,
+      "step": 12939
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0583690886494146e-06,
+      "loss": 0.5534,
+      "step": 12940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.05740920882884e-06,
+      "loss": 0.576,
+      "step": 12941
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.056449740185028e-06,
+      "loss": 0.588,
+      "step": 12942
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.055490682762098e-06,
+      "loss": 0.6274,
+      "step": 12943
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0545320366041444e-06,
+      "loss": 0.5611,
+      "step": 12944
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0535738017552476e-06,
+      "loss": 0.6067,
+      "step": 12945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.052615978259466e-06,
+      "loss": 0.6144,
+      "step": 12946
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0516585661608403e-06,
+      "loss": 0.5062,
+      "step": 12947
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0507015655033925e-06,
+      "loss": 0.5595,
+      "step": 12948
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.049744976331124e-06,
+      "loss": 0.6019,
+      "step": 12949
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0487887986880175e-06,
+      "loss": 0.5987,
+      "step": 12950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0478330326180387e-06,
+      "loss": 0.5879,
+      "step": 12951
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0468776781651356e-06,
+      "loss": 0.5967,
+      "step": 12952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0459227353732314e-06,
+      "loss": 0.6025,
+      "step": 12953
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0449682042862352e-06,
+      "loss": 0.597,
+      "step": 12954
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0440140849480364e-06,
+      "loss": 0.5508,
+      "step": 12955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.043060377402506e-06,
+      "loss": 0.5869,
+      "step": 12956
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0421070816934928e-06,
+      "loss": 0.6159,
+      "step": 12957
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0411541978648295e-06,
+      "loss": 0.563,
+      "step": 12958
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0402017259603303e-06,
+      "loss": 0.5896,
+      "step": 12959
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0392496660237894e-06,
+      "loss": 0.536,
+      "step": 12960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.038298018098981e-06,
+      "loss": 0.5693,
+      "step": 12961
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0373467822296635e-06,
+      "loss": 0.5838,
+      "step": 12962
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.036395958459574e-06,
+      "loss": 0.6069,
+      "step": 12963
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.035445546832429e-06,
+      "loss": 0.5569,
+      "step": 12964
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0344955473919293e-06,
+      "loss": 0.6507,
+      "step": 12965
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0335459601817543e-06,
+      "loss": 0.6329,
+      "step": 12966
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0325967852455697e-06,
+      "loss": 0.5841,
+      "step": 12967
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.031648022627013e-06,
+      "loss": 0.5697,
+      "step": 12968
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0306996723697105e-06,
+      "loss": 0.6228,
+      "step": 12969
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.029751734517266e-06,
+      "loss": 0.5993,
+      "step": 12970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0288042091132687e-06,
+      "loss": 0.5809,
+      "step": 12971
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.02785709620128e-06,
+      "loss": 0.5499,
+      "step": 12972
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0269103958248516e-06,
+      "loss": 0.5546,
+      "step": 12973
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0259641080275118e-06,
+      "loss": 0.5809,
+      "step": 12974
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.025018232852769e-06,
+      "loss": 0.6524,
+      "step": 12975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0240727703441144e-06,
+      "loss": 0.572,
+      "step": 12976
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0231277205450217e-06,
+      "loss": 0.6086,
+      "step": 12977
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.022183083498942e-06,
+      "loss": 0.5114,
+      "step": 12978
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.02123885924931e-06,
+      "loss": 0.5546,
+      "step": 12979
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0202950478395423e-06,
+      "loss": 0.5836,
+      "step": 12980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.019351649313034e-06,
+      "loss": 0.6338,
+      "step": 12981
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.018408663713163e-06,
+      "loss": 0.5925,
+      "step": 12982
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0174660910832843e-06,
+      "loss": 0.5844,
+      "step": 12983
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0165239314667396e-06,
+      "loss": 0.564,
+      "step": 12984
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.015582184906848e-06,
+      "loss": 0.6298,
+      "step": 12985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.014640851446913e-06,
+      "loss": 0.5604,
+      "step": 12986
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.013699931130213e-06,
+      "loss": 0.5692,
+      "step": 12987
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.012759424000014e-06,
+      "loss": 0.5405,
+      "step": 12988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.011819330099558e-06,
+      "loss": 0.6038,
+      "step": 12989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.010879649472073e-06,
+      "loss": 0.6108,
+      "step": 12990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.009940382160762e-06,
+      "loss": 0.5311,
+      "step": 12991
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.009001528208814e-06,
+      "loss": 0.6391,
+      "step": 12992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0080630876593953e-06,
+      "loss": 0.6045,
+      "step": 12993
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0071250605556572e-06,
+      "loss": 0.6466,
+      "step": 12994
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0061874469407297e-06,
+      "loss": 0.6058,
+      "step": 12995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0052502468577216e-06,
+      "loss": 0.604,
+      "step": 12996
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.004313460349729e-06,
+      "loss": 0.6092,
+      "step": 12997
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0033770874598223e-06,
+      "loss": 0.533,
+      "step": 12998
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.002441128231054e-06,
+      "loss": 0.6313,
+      "step": 12999
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.001505582706462e-06,
+      "loss": 0.5785,
+      "step": 13000
+    },
+    {
+      "epoch": 0.86,
+      "eval_text_loss": 0.6473656296730042,
+      "eval_text_runtime": 1.7719,
+      "eval_text_samples_per_second": 41.763,
+      "eval_text_steps_per_second": 1.129,
+      "step": 13000
+    },
+    {
+      "epoch": 0.86,
+      "eval_svit_loss": 0.7559177279472351,
+      "eval_svit_runtime": 6.6068,
+      "eval_svit_samples_per_second": 123.509,
+      "eval_svit_steps_per_second": 1.968,
+      "step": 13000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0005704509290624e-06,
+      "loss": 0.6298,
+      "step": 13001
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9996357329418495e-06,
+      "loss": 0.611,
+      "step": 13002
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9987014287878028e-06,
+      "loss": 0.6881,
+      "step": 13003
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9977675385098818e-06,
+      "loss": 0.5456,
+      "step": 13004
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9968340621510263e-06,
+      "loss": 0.5425,
+      "step": 13005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9959009997541557e-06,
+      "loss": 0.5404,
+      "step": 13006
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9949683513621733e-06,
+      "loss": 0.6754,
+      "step": 13007
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9940361170179635e-06,
+      "loss": 0.5606,
+      "step": 13008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9931042967643855e-06,
+      "loss": 0.6328,
+      "step": 13009
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.992172890644286e-06,
+      "loss": 0.6364,
+      "step": 13010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9912418987004924e-06,
+      "loss": 0.6177,
+      "step": 13011
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9903113209758094e-06,
+      "loss": 0.578,
+      "step": 13012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.989381157513025e-06,
+      "loss": 0.5922,
+      "step": 13013
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9884514083549087e-06,
+      "loss": 0.6207,
+      "step": 13014
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.987522073544209e-06,
+      "loss": 0.583,
+      "step": 13015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9865931531236596e-06,
+      "loss": 0.5529,
+      "step": 13016
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9856646471359664e-06,
+      "loss": 0.5717,
+      "step": 13017
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9847365556238255e-06,
+      "loss": 0.6128,
+      "step": 13018
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.983808878629907e-06,
+      "loss": 0.5563,
+      "step": 13019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9828816161968713e-06,
+      "loss": 0.4983,
+      "step": 13020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9819547683673457e-06,
+      "loss": 0.5693,
+      "step": 13021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.981028335183951e-06,
+      "loss": 0.592,
+      "step": 13022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9801023166892832e-06,
+      "loss": 0.6199,
+      "step": 13023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.979176712925922e-06,
+      "loss": 0.6016,
+      "step": 13024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.978251523936422e-06,
+      "loss": 0.5366,
+      "step": 13025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9773267497633264e-06,
+      "loss": 0.5855,
+      "step": 13026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9764023904491557e-06,
+      "loss": 0.6215,
+      "step": 13027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9754784460364107e-06,
+      "loss": 0.5704,
+      "step": 13028
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9745549165675737e-06,
+      "loss": 0.6543,
+      "step": 13029
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9736318020851096e-06,
+      "loss": 0.5854,
+      "step": 13030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9727091026314636e-06,
+      "loss": 0.5872,
+      "step": 13031
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9717868182490577e-06,
+      "loss": 0.5387,
+      "step": 13032
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9708649489803e-06,
+      "loss": 0.5644,
+      "step": 13033
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.969943494867578e-06,
+      "loss": 0.5655,
+      "step": 13034
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.969022455953261e-06,
+      "loss": 0.6298,
+      "step": 13035
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9681018322796952e-06,
+      "loss": 0.5416,
+      "step": 13036
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9671816238892106e-06,
+      "loss": 0.5604,
+      "step": 13037
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.96626183082412e-06,
+      "loss": 0.566,
+      "step": 13038
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.965342453126716e-06,
+      "loss": 0.6333,
+      "step": 13039
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9644234908392693e-06,
+      "loss": 0.562,
+      "step": 13040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.963504944004032e-06,
+      "loss": 0.6564,
+      "step": 13041
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.962586812663242e-06,
+      "loss": 0.5823,
+      "step": 13042
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9616690968591113e-06,
+      "loss": 0.5799,
+      "step": 13043
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9607517966338374e-06,
+      "loss": 0.6183,
+      "step": 13044
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.959834912029598e-06,
+      "loss": 0.6168,
+      "step": 13045
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.958918443088551e-06,
+      "loss": 0.5665,
+      "step": 13046
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9580023898528347e-06,
+      "loss": 0.6198,
+      "step": 13047
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9570867523645687e-06,
+      "loss": 0.6147,
+      "step": 13048
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9561715306658547e-06,
+      "loss": 0.6228,
+      "step": 13049
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9552567247987754e-06,
+      "loss": 0.6103,
+      "step": 13050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9543423348053893e-06,
+      "loss": 0.6197,
+      "step": 13051
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9534283607277427e-06,
+      "loss": 0.5975,
+      "step": 13052
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9525148026078578e-06,
+      "loss": 0.529,
+      "step": 13053
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.951601660487743e-06,
+      "loss": 0.6184,
+      "step": 13054
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9506889344093806e-06,
+      "loss": 0.5699,
+      "step": 13055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.949776624414738e-06,
+      "loss": 0.4995,
+      "step": 13056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9488647305457632e-06,
+      "loss": 0.4909,
+      "step": 13057
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.947953252844388e-06,
+      "loss": 0.5434,
+      "step": 13058
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9470421913525162e-06,
+      "loss": 0.5705,
+      "step": 13059
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.946131546112041e-06,
+      "loss": 0.531,
+      "step": 13060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.945221317164834e-06,
+      "loss": 0.6041,
+      "step": 13061
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.944311504552745e-06,
+      "loss": 0.5881,
+      "step": 13062
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.943402108317609e-06,
+      "loss": 0.6652,
+      "step": 13063
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9424931285012384e-06,
+      "loss": 0.5349,
+      "step": 13064
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.94158456514543e-06,
+      "loss": 0.5859,
+      "step": 13065
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.940676418291957e-06,
+      "loss": 0.5964,
+      "step": 13066
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.939768687982575e-06,
+      "loss": 0.6119,
+      "step": 13067
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.938861374259021e-06,
+      "loss": 0.6375,
+      "step": 13068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.937954477163018e-06,
+      "loss": 0.5991,
+      "step": 13069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9370479967362576e-06,
+      "loss": 0.6466,
+      "step": 13070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9361419330204233e-06,
+      "loss": 0.5771,
+      "step": 13071
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.935236286057174e-06,
+      "loss": 0.5585,
+      "step": 13072
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9343310558881544e-06,
+      "loss": 0.5805,
+      "step": 13073
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.933426242554982e-06,
+      "loss": 0.5581,
+      "step": 13074
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9325218460992622e-06,
+      "loss": 0.6217,
+      "step": 13075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9316178665625796e-06,
+      "loss": 0.5898,
+      "step": 13076
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.930714303986496e-06,
+      "loss": 0.563,
+      "step": 13077
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9298111584125587e-06,
+      "loss": 0.5364,
+      "step": 13078
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9289084298822937e-06,
+      "loss": 0.608,
+      "step": 13079
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.928006118437208e-06,
+      "loss": 0.5744,
+      "step": 13080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.92710422411879e-06,
+      "loss": 0.6301,
+      "step": 13081
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9262027469685085e-06,
+      "loss": 0.5735,
+      "step": 13082
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.925301687027812e-06,
+      "loss": 0.6087,
+      "step": 13083
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9244010443381333e-06,
+      "loss": 0.6158,
+      "step": 13084
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9235008189408802e-06,
+      "loss": 0.5412,
+      "step": 13085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9226010108774463e-06,
+      "loss": 0.4908,
+      "step": 13086
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.921701620189205e-06,
+      "loss": 0.6306,
+      "step": 13087
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.92080264691751e-06,
+      "loss": 0.5524,
+      "step": 13088
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9199040911036937e-06,
+      "loss": 0.6031,
+      "step": 13089
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9190059527890738e-06,
+      "loss": 0.5796,
+      "step": 13090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9181082320149445e-06,
+      "loss": 0.5876,
+      "step": 13091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9172109288225847e-06,
+      "loss": 0.5896,
+      "step": 13092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9163140432532503e-06,
+      "loss": 0.5864,
+      "step": 13093
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9154175753481797e-06,
+      "loss": 0.5887,
+      "step": 13094
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9145215251485914e-06,
+      "loss": 0.5546,
+      "step": 13095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9136258926956888e-06,
+      "loss": 0.6289,
+      "step": 13096
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9127306780306497e-06,
+      "loss": 0.6669,
+      "step": 13097
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.911835881194637e-06,
+      "loss": 0.6393,
+      "step": 13098
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9109415022287937e-06,
+      "loss": 0.6473,
+      "step": 13099
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9100475411742403e-06,
+      "loss": 0.5391,
+      "step": 13100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.909153998072083e-06,
+      "loss": 0.6529,
+      "step": 13101
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.908260872963407e-06,
+      "loss": 0.5358,
+      "step": 13102
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9073681658892774e-06,
+      "loss": 0.6574,
+      "step": 13103
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9064758768907375e-06,
+      "loss": 0.6249,
+      "step": 13104
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.905584006008818e-06,
+      "loss": 0.6404,
+      "step": 13105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9046925532845263e-06,
+      "loss": 0.615,
+      "step": 13106
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.903801518758851e-06,
+      "loss": 0.538,
+      "step": 13107
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.90291090247276e-06,
+      "loss": 0.5846,
+      "step": 13108
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.902020704467205e-06,
+      "loss": 0.565,
+      "step": 13109
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9011309247831167e-06,
+      "loss": 0.5883,
+      "step": 13110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9002415634614024e-06,
+      "loss": 0.6296,
+      "step": 13111
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8993526205429627e-06,
+      "loss": 0.6411,
+      "step": 13112
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8984640960686652e-06,
+      "loss": 0.5782,
+      "step": 13113
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8975759900793678e-06,
+      "loss": 0.5429,
+      "step": 13114
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8966883026159012e-06,
+      "loss": 0.5487,
+      "step": 13115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8958010337190824e-06,
+      "loss": 0.5853,
+      "step": 13116
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.894914183429708e-06,
+      "loss": 0.5799,
+      "step": 13117
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8940277517885565e-06,
+      "loss": 0.5866,
+      "step": 13118
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8931417388363827e-06,
+      "loss": 0.6265,
+      "step": 13119
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8922561446139265e-06,
+      "loss": 0.5767,
+      "step": 13120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.891370969161908e-06,
+      "loss": 0.5068,
+      "step": 13121
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8904862125210266e-06,
+      "loss": 0.6012,
+      "step": 13122
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8896018747319623e-06,
+      "loss": 0.5927,
+      "step": 13123
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8887179558353773e-06,
+      "loss": 0.5648,
+      "step": 13124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8878344558719152e-06,
+      "loss": 0.6079,
+      "step": 13125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8869513748821955e-06,
+      "loss": 0.5547,
+      "step": 13126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8860687129068244e-06,
+      "loss": 0.5436,
+      "step": 13127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8851864699863854e-06,
+      "loss": 0.5473,
+      "step": 13128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8843046461614447e-06,
+      "loss": 0.6114,
+      "step": 13129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.883423241472547e-06,
+      "loss": 0.6241,
+      "step": 13130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8825422559602203e-06,
+      "loss": 0.5502,
+      "step": 13131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.881661689664971e-06,
+      "loss": 0.6127,
+      "step": 13132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.880781542627289e-06,
+      "loss": 0.5113,
+      "step": 13133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8799018148876403e-06,
+      "loss": 0.6228,
+      "step": 13134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8790225064864756e-06,
+      "loss": 0.5899,
+      "step": 13135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8781436174642255e-06,
+      "loss": 0.6325,
+      "step": 13136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8772651478613015e-06,
+      "loss": 0.5479,
+      "step": 13137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.876387097718093e-06,
+      "loss": 0.5618,
+      "step": 13138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.875509467074975e-06,
+      "loss": 0.6036,
+      "step": 13139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8746322559722983e-06,
+      "loss": 0.6404,
+      "step": 13140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8737554644503997e-06,
+      "loss": 0.5501,
+      "step": 13141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.872879092549591e-06,
+      "loss": 0.5526,
+      "step": 13142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.872003140310168e-06,
+      "loss": 0.5988,
+      "step": 13143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.871127607772407e-06,
+      "loss": 0.548,
+      "step": 13144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8702524949765643e-06,
+      "loss": 0.5165,
+      "step": 13145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8693778019628785e-06,
+      "loss": 0.6366,
+      "step": 13146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.868503528771566e-06,
+      "loss": 0.6435,
+      "step": 13147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.867629675442828e-06,
+      "loss": 0.5956,
+      "step": 13148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8667562420168404e-06,
+      "loss": 0.5392,
+      "step": 13149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.865883228533766e-06,
+      "loss": 0.5601,
+      "step": 13150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.865010635033744e-06,
+      "loss": 0.5394,
+      "step": 13151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8641384615568982e-06,
+      "loss": 0.6108,
+      "step": 13152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8632667081433274e-06,
+      "loss": 0.5626,
+      "step": 13153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.862395374833117e-06,
+      "loss": 0.5087,
+      "step": 13154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8615244616663296e-06,
+      "loss": 0.5822,
+      "step": 13155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8606539686830117e-06,
+      "loss": 0.5413,
+      "step": 13156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.859783895923184e-06,
+      "loss": 0.5413,
+      "step": 13157
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8589142434268537e-06,
+      "loss": 0.5635,
+      "step": 13158
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.85804501123401e-06,
+      "loss": 0.5479,
+      "step": 13159
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.857176199384616e-06,
+      "loss": 0.6176,
+      "step": 13160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8563078079186205e-06,
+      "loss": 0.5787,
+      "step": 13161
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.855439836875953e-06,
+      "loss": 0.5903,
+      "step": 13162
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.854572286296521e-06,
+      "loss": 0.5903,
+      "step": 13163
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8537051562202145e-06,
+      "loss": 0.5925,
+      "step": 13164
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8528384466869043e-06,
+      "loss": 0.5753,
+      "step": 13165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8519721577364414e-06,
+      "loss": 0.5901,
+      "step": 13166
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8511062894086586e-06,
+      "loss": 0.5981,
+      "step": 13167
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.850240841743365e-06,
+      "loss": 0.5929,
+      "step": 13168
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8493758147803544e-06,
+      "loss": 0.5331,
+      "step": 13169
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8485112085594017e-06,
+      "loss": 0.605,
+      "step": 13170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8476470231202627e-06,
+      "loss": 0.5838,
+      "step": 13171
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.846783258502669e-06,
+      "loss": 0.5788,
+      "step": 13172
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.845919914746337e-06,
+      "loss": 0.6014,
+      "step": 13173
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8450569918909635e-06,
+      "loss": 0.6076,
+      "step": 13174
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8441944899762267e-06,
+      "loss": 0.567,
+      "step": 13175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8433324090417804e-06,
+      "loss": 0.5931,
+      "step": 13176
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8424707491272654e-06,
+      "loss": 0.5557,
+      "step": 13177
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8416095102723008e-06,
+      "loss": 0.566,
+      "step": 13178
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.840748692516484e-06,
+      "loss": 0.5787,
+      "step": 13179
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.839888295899397e-06,
+      "loss": 0.5414,
+      "step": 13180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8390283204605997e-06,
+      "loss": 0.5627,
+      "step": 13181
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8381687662396343e-06,
+      "loss": 0.651,
+      "step": 13182
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.837309633276021e-06,
+      "loss": 0.6115,
+      "step": 13183
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.836450921609264e-06,
+      "loss": 0.6172,
+      "step": 13184
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8355926312788454e-06,
+      "loss": 0.6657,
+      "step": 13185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8347347623242305e-06,
+      "loss": 0.5742,
+      "step": 13186
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.833877314784862e-06,
+      "loss": 0.5539,
+      "step": 13187
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8330202887001656e-06,
+      "loss": 0.5905,
+      "step": 13188
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.832163684109547e-06,
+      "loss": 0.5477,
+      "step": 13189
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.831307501052394e-06,
+      "loss": 0.5796,
+      "step": 13190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.830451739568071e-06,
+      "loss": 0.571,
+      "step": 13191
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8295963996959275e-06,
+      "loss": 0.6354,
+      "step": 13192
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8287414814752924e-06,
+      "loss": 0.5696,
+      "step": 13193
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8278869849454716e-06,
+      "loss": 0.5694,
+      "step": 13194
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8270329101457553e-06,
+      "loss": 0.6442,
+      "step": 13195
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8261792571154148e-06,
+      "loss": 0.5707,
+      "step": 13196
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8253260258937e-06,
+      "loss": 0.5805,
+      "step": 13197
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8244732165198432e-06,
+      "loss": 0.6091,
+      "step": 13198
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.823620829033055e-06,
+      "loss": 0.5548,
+      "step": 13199
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.822768863472527e-06,
+      "loss": 0.6207,
+      "step": 13200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8219173198774366e-06,
+      "loss": 0.5701,
+      "step": 13201
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.821066198286933e-06,
+      "loss": 0.5981,
+      "step": 13202
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.82021549874015e-06,
+      "loss": 0.5719,
+      "step": 13203
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.819365221276206e-06,
+      "loss": 0.5967,
+      "step": 13204
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8185153659341947e-06,
+      "loss": 0.552,
+      "step": 13205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.817665932753191e-06,
+      "loss": 0.6144,
+      "step": 13206
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8168169217722532e-06,
+      "loss": 0.5653,
+      "step": 13207
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.815968333030417e-06,
+      "loss": 0.5182,
+      "step": 13208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.815120166566702e-06,
+      "loss": 0.6445,
+      "step": 13209
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8142724224201046e-06,
+      "loss": 0.624,
+      "step": 13210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8134251006296047e-06,
+      "loss": 0.6139,
+      "step": 13211
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8125782012341622e-06,
+      "loss": 0.5489,
+      "step": 13212
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.811731724272715e-06,
+      "loss": 0.5529,
+      "step": 13213
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.810885669784187e-06,
+      "loss": 0.5391,
+      "step": 13214
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.810040037807478e-06,
+      "loss": 0.6364,
+      "step": 13215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.809194828381471e-06,
+      "loss": 0.6539,
+      "step": 13216
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.808350041545025e-06,
+      "loss": 0.5484,
+      "step": 13217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.807505677336986e-06,
+      "loss": 0.5301,
+      "step": 13218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.806661735796177e-06,
+      "loss": 0.5749,
+      "step": 13219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8058182169614026e-06,
+      "loss": 0.6044,
+      "step": 13220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.804975120871446e-06,
+      "loss": 0.58,
+      "step": 13221
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8041324475650722e-06,
+      "loss": 0.5839,
+      "step": 13222
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.803290197081028e-06,
+      "loss": 0.5579,
+      "step": 13223
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8024483694580408e-06,
+      "loss": 0.5607,
+      "step": 13224
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8016069647348157e-06,
+      "loss": 0.5703,
+      "step": 13225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.80076598295004e-06,
+      "loss": 0.5394,
+      "step": 13226
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7999254241423845e-06,
+      "loss": 0.5376,
+      "step": 13227
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7990852883504938e-06,
+      "loss": 0.5812,
+      "step": 13228
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.798245575612999e-06,
+      "loss": 0.5968,
+      "step": 13229
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7974062859685093e-06,
+      "loss": 0.566,
+      "step": 13230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7965674194556158e-06,
+      "loss": 0.6348,
+      "step": 13231
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.795728976112888e-06,
+      "loss": 0.5741,
+      "step": 13232
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7948909559788796e-06,
+      "loss": 0.6226,
+      "step": 13233
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7940533590921195e-06,
+      "loss": 0.5698,
+      "step": 13234
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7932161854911236e-06,
+      "loss": 0.626,
+      "step": 13235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7923794352143796e-06,
+      "loss": 0.5242,
+      "step": 13236
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.791543108300364e-06,
+      "loss": 0.5391,
+      "step": 13237
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.790707204787532e-06,
+      "loss": 0.5946,
+      "step": 13238
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7898717247143182e-06,
+      "loss": 0.553,
+      "step": 13239
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.789036668119134e-06,
+      "loss": 0.5648,
+      "step": 13240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7882020350403773e-06,
+      "loss": 0.5028,
+      "step": 13241
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.787367825516424e-06,
+      "loss": 0.6516,
+      "step": 13242
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7865340395856325e-06,
+      "loss": 0.5614,
+      "step": 13243
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.785700677286337e-06,
+      "loss": 0.6295,
+      "step": 13244
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7848677386568557e-06,
+      "loss": 0.6096,
+      "step": 13245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7840352237354875e-06,
+      "loss": 0.6183,
+      "step": 13246
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7832031325605116e-06,
+      "loss": 0.6435,
+      "step": 13247
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7823714651701862e-06,
+      "loss": 0.5228,
+      "step": 13248
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.781540221602752e-06,
+      "loss": 0.6227,
+      "step": 13249
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.780709401896431e-06,
+      "loss": 0.5866,
+      "step": 13250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.779879006089419e-06,
+      "loss": 0.5769,
+      "step": 13251
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7790490342199005e-06,
+      "loss": 0.6269,
+      "step": 13252
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.778219486326037e-06,
+      "loss": 0.5731,
+      "step": 13253
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.777390362445972e-06,
+      "loss": 0.591,
+      "step": 13254
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.776561662617826e-06,
+      "loss": 0.5937,
+      "step": 13255
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7757333868797023e-06,
+      "loss": 0.5582,
+      "step": 13256
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7749055352696853e-06,
+      "loss": 0.5854,
+      "step": 13257
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7740781078258416e-06,
+      "loss": 0.5538,
+      "step": 13258
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7732511045862123e-06,
+      "loss": 0.4907,
+      "step": 13259
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.772424525588825e-06,
+      "loss": 0.5379,
+      "step": 13260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7715983708716855e-06,
+      "loss": 0.5835,
+      "step": 13261
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.770772640472778e-06,
+      "loss": 0.5536,
+      "step": 13262
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.76994733443007e-06,
+      "loss": 0.6104,
+      "step": 13263
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.76912245278151e-06,
+      "loss": 0.6122,
+      "step": 13264
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.768297995565025e-06,
+      "loss": 0.5608,
+      "step": 13265
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7674739628185235e-06,
+      "loss": 0.6189,
+      "step": 13266
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7666503545798942e-06,
+      "loss": 0.6214,
+      "step": 13267
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7658271708870056e-06,
+      "loss": 0.5992,
+      "step": 13268
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.765004411777709e-06,
+      "loss": 0.5851,
+      "step": 13269
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.764182077289832e-06,
+      "loss": 0.594,
+      "step": 13270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7633601674611867e-06,
+      "loss": 0.5358,
+      "step": 13271
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.762538682329564e-06,
+      "loss": 0.6152,
+      "step": 13272
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.761717621932738e-06,
+      "loss": 0.6092,
+      "step": 13273
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.760896986308456e-06,
+      "loss": 0.5456,
+      "step": 13274
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7600767754944527e-06,
+      "loss": 0.667,
+      "step": 13275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.759256989528443e-06,
+      "loss": 0.6688,
+      "step": 13276
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7584376284481163e-06,
+      "loss": 0.578,
+      "step": 13277
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7576186922911486e-06,
+      "loss": 0.5671,
+      "step": 13278
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.756800181095194e-06,
+      "loss": 0.5793,
+      "step": 13279
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.755982094897889e-06,
+      "loss": 0.6228,
+      "step": 13280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.755164433736847e-06,
+      "loss": 0.6049,
+      "step": 13281
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.754347197649663e-06,
+      "loss": 0.5491,
+      "step": 13282
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.753530386673915e-06,
+      "loss": 0.5558,
+      "step": 13283
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.752714000847161e-06,
+      "loss": 0.5612,
+      "step": 13284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.751898040206935e-06,
+      "loss": 0.5219,
+      "step": 13285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7510825047907555e-06,
+      "loss": 0.5821,
+      "step": 13286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7502673946361203e-06,
+      "loss": 0.5633,
+      "step": 13287
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7494527097805103e-06,
+      "loss": 0.6267,
+      "step": 13288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7486384502613806e-06,
+      "loss": 0.6232,
+      "step": 13289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7478246161161715e-06,
+      "loss": 0.5707,
+      "step": 13290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7470112073823043e-06,
+      "loss": 0.5235,
+      "step": 13291
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7461982240971796e-06,
+      "loss": 0.5751,
+      "step": 13292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7453856662981762e-06,
+      "loss": 0.588,
+      "step": 13293
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7445735340226546e-06,
+      "loss": 0.58,
+      "step": 13294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7437618273079604e-06,
+      "loss": 0.5845,
+      "step": 13295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7429505461914098e-06,
+      "loss": 0.5977,
+      "step": 13296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.742139690710307e-06,
+      "loss": 0.5847,
+      "step": 13297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7413292609019382e-06,
+      "loss": 0.5443,
+      "step": 13298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7405192568035656e-06,
+      "loss": 0.5945,
+      "step": 13299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7397096784524303e-06,
+      "loss": 0.5254,
+      "step": 13300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.738900525885757e-06,
+      "loss": 0.5145,
+      "step": 13301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.738091799140752e-06,
+      "loss": 0.6148,
+      "step": 13302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7372834982546004e-06,
+      "loss": 0.6426,
+      "step": 13303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.736475623264464e-06,
+      "loss": 0.641,
+      "step": 13304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.735668174207491e-06,
+      "loss": 0.5893,
+      "step": 13305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.734861151120808e-06,
+      "loss": 0.5629,
+      "step": 13306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.734054554041523e-06,
+      "loss": 0.5207,
+      "step": 13307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7332483830067193e-06,
+      "loss": 0.5637,
+      "step": 13308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7324426380534653e-06,
+      "loss": 0.6108,
+      "step": 13309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.731637319218812e-06,
+      "loss": 0.5581,
+      "step": 13310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7308324265397835e-06,
+      "loss": 0.5381,
+      "step": 13311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.730027960053391e-06,
+      "loss": 0.6402,
+      "step": 13312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.729223919796621e-06,
+      "loss": 0.6341,
+      "step": 13313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7284203058064464e-06,
+      "loss": 0.5754,
+      "step": 13314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.727617118119814e-06,
+      "loss": 0.6172,
+      "step": 13315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7268143567736573e-06,
+      "loss": 0.5931,
+      "step": 13316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.726012021804884e-06,
+      "loss": 0.5338,
+      "step": 13317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7252101132503875e-06,
+      "loss": 0.5812,
+      "step": 13318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7244086311470366e-06,
+      "loss": 0.5643,
+      "step": 13319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.723607575531685e-06,
+      "loss": 0.5121,
+      "step": 13320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7228069464411636e-06,
+      "loss": 0.6497,
+      "step": 13321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.722006743912288e-06,
+      "loss": 0.5523,
+      "step": 13322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7212069679818466e-06,
+      "loss": 0.5312,
+      "step": 13323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.720407618686616e-06,
+      "loss": 0.4888,
+      "step": 13324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7196086960633496e-06,
+      "loss": 0.5616,
+      "step": 13325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7188102001487816e-06,
+      "loss": 0.6026,
+      "step": 13326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7180121309796248e-06,
+      "loss": 0.5879,
+      "step": 13327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.717214488592576e-06,
+      "loss": 0.5146,
+      "step": 13328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7164172730243092e-06,
+      "loss": 0.5787,
+      "step": 13329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.715620484311481e-06,
+      "loss": 0.4966,
+      "step": 13330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.714824122490726e-06,
+      "loss": 0.6599,
+      "step": 13331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.714028187598663e-06,
+      "loss": 0.5505,
+      "step": 13332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.713232679671888e-06,
+      "loss": 0.5683,
+      "step": 13333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7124375987469766e-06,
+      "loss": 0.5843,
+      "step": 13334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7116429448604876e-06,
+      "loss": 0.5222,
+      "step": 13335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7108487180489575e-06,
+      "loss": 0.6274,
+      "step": 13336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.710054918348908e-06,
+      "loss": 0.5927,
+      "step": 13337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.709261545796834e-06,
+      "loss": 0.6138,
+      "step": 13338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7084686004292153e-06,
+      "loss": 0.6062,
+      "step": 13339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.707676082282511e-06,
+      "loss": 0.6334,
+      "step": 13340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7068839913931647e-06,
+      "loss": 0.5769,
+      "step": 13341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7060923277975904e-06,
+      "loss": 0.5606,
+      "step": 13342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7053010915321912e-06,
+      "loss": 0.625,
+      "step": 13343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7045102826333493e-06,
+      "loss": 0.607,
+      "step": 13344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.703719901137424e-06,
+      "loss": 0.5482,
+      "step": 13345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.702929947080756e-06,
+      "loss": 0.6027,
+      "step": 13346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7021404204996686e-06,
+      "loss": 0.6285,
+      "step": 13347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.701351321430462e-06,
+      "loss": 0.5732,
+      "step": 13348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.700562649909422e-06,
+      "loss": 0.563,
+      "step": 13349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6997744059728082e-06,
+      "loss": 0.5327,
+      "step": 13350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.698986589656866e-06,
+      "loss": 0.5681,
+      "step": 13351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.69819920099782e-06,
+      "loss": 0.594,
+      "step": 13352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6974122400318693e-06,
+      "loss": 0.5656,
+      "step": 13353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.696625706795201e-06,
+      "loss": 0.559,
+      "step": 13354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6958396013239787e-06,
+      "loss": 0.6289,
+      "step": 13355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.695053923654351e-06,
+      "loss": 0.5642,
+      "step": 13356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6942686738224368e-06,
+      "loss": 0.5853,
+      "step": 13357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.693483851864345e-06,
+      "loss": 0.5835,
+      "step": 13358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6926994578161604e-06,
+      "loss": 0.6381,
+      "step": 13359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6919154917139526e-06,
+      "loss": 0.5582,
+      "step": 13360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.691131953593763e-06,
+      "loss": 0.5317,
+      "step": 13361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6903488434916206e-06,
+      "loss": 0.5485,
+      "step": 13362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6895661614435323e-06,
+      "loss": 0.5845,
+      "step": 13363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.688783907485486e-06,
+      "loss": 0.626,
+      "step": 13364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6880020816534492e-06,
+      "loss": 0.5918,
+      "step": 13365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6872206839833693e-06,
+      "loss": 0.5544,
+      "step": 13366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6864397145111777e-06,
+      "loss": 0.5767,
+      "step": 13367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.685659173272778e-06,
+      "loss": 0.5796,
+      "step": 13368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.684879060304062e-06,
+      "loss": 0.6827,
+      "step": 13369
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6840993756408997e-06,
+      "loss": 0.5316,
+      "step": 13370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6833201193191396e-06,
+      "loss": 0.5221,
+      "step": 13371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.682541291374611e-06,
+      "loss": 0.5581,
+      "step": 13372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6817628918431252e-06,
+      "loss": 0.6243,
+      "step": 13373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6809849207604723e-06,
+      "loss": 0.5922,
+      "step": 13374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.680207378162424e-06,
+      "loss": 0.5905,
+      "step": 13375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6794302640847294e-06,
+      "loss": 0.6243,
+      "step": 13376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.678653578563121e-06,
+      "loss": 0.5524,
+      "step": 13377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.677877321633312e-06,
+      "loss": 0.5501,
+      "step": 13378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6771014933309916e-06,
+      "loss": 0.5831,
+      "step": 13379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6763260936918326e-06,
+      "loss": 0.5886,
+      "step": 13380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6755511227514887e-06,
+      "loss": 0.5732,
+      "step": 13381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.674776580545592e-06,
+      "loss": 0.6005,
+      "step": 13382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6740024671097567e-06,
+      "loss": 0.5794,
+      "step": 13383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6732287824795758e-06,
+      "loss": 0.6172,
+      "step": 13384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6724555266906216e-06,
+      "loss": 0.5933,
+      "step": 13385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.671682699778452e-06,
+      "loss": 0.563,
+      "step": 13386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6709103017785966e-06,
+      "loss": 0.5389,
+      "step": 13387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.670138332726571e-06,
+      "loss": 0.6341,
+      "step": 13388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.669366792657872e-06,
+      "loss": 0.5439,
+      "step": 13389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6685956816079753e-06,
+      "loss": 0.5952,
+      "step": 13390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.667824999612332e-06,
+      "loss": 0.5514,
+      "step": 13391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6670547467063817e-06,
+      "loss": 0.6172,
+      "step": 13392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.666284922925537e-06,
+      "loss": 0.5704,
+      "step": 13393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6655155283051992e-06,
+      "loss": 0.5155,
+      "step": 13394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.664746562880739e-06,
+      "loss": 0.5869,
+      "step": 13395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.663978026687517e-06,
+      "loss": 0.5769,
+      "step": 13396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.663209919760869e-06,
+      "loss": 0.5184,
+      "step": 13397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.662442242136111e-06,
+      "loss": 0.5848,
+      "step": 13398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.661674993848542e-06,
+      "loss": 0.5821,
+      "step": 13399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.66090817493344e-06,
+      "loss": 0.5468,
+      "step": 13400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.660141785426064e-06,
+      "loss": 0.5772,
+      "step": 13401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.659375825361649e-06,
+      "loss": 0.5313,
+      "step": 13402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6586102947754147e-06,
+      "loss": 0.5524,
+      "step": 13403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6578451937025606e-06,
+      "loss": 0.5887,
+      "step": 13404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.657080522178268e-06,
+      "loss": 0.5286,
+      "step": 13405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.656316280237692e-06,
+      "loss": 0.5629,
+      "step": 13406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.655552467915974e-06,
+      "loss": 0.4604,
+      "step": 13407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.654789085248234e-06,
+      "loss": 0.59,
+      "step": 13408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6540261322695727e-06,
+      "loss": 0.5387,
+      "step": 13409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.653263609015068e-06,
+      "loss": 0.6426,
+      "step": 13410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6525015155197823e-06,
+      "loss": 0.581,
+      "step": 13411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.651739851818758e-06,
+      "loss": 0.6175,
+      "step": 13412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.650978617947011e-06,
+      "loss": 0.5642,
+      "step": 13413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6502178139395465e-06,
+      "loss": 0.6388,
+      "step": 13414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6494574398313456e-06,
+      "loss": 0.65,
+      "step": 13415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6486974956573688e-06,
+      "loss": 0.5836,
+      "step": 13416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.647937981452559e-06,
+      "loss": 0.5961,
+      "step": 13417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6471788972518367e-06,
+      "loss": 0.6451,
+      "step": 13418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6464202430901066e-06,
+      "loss": 0.5725,
+      "step": 13419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.645662019002252e-06,
+      "loss": 0.558,
+      "step": 13420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6449042250231327e-06,
+      "loss": 0.5937,
+      "step": 13421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6441468611875935e-06,
+      "loss": 0.628,
+      "step": 13422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6433899275304556e-06,
+      "loss": 0.5499,
+      "step": 13423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.642633424086527e-06,
+      "loss": 0.5772,
+      "step": 13424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6418773508905872e-06,
+      "loss": 0.6189,
+      "step": 13425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.641121707977401e-06,
+      "loss": 0.5474,
+      "step": 13426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6403664953817157e-06,
+      "loss": 0.5443,
+      "step": 13427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6396117131382507e-06,
+      "loss": 0.5715,
+      "step": 13428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.638857361281714e-06,
+      "loss": 0.6001,
+      "step": 13429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.638103439846788e-06,
+      "loss": 0.6118,
+      "step": 13430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.63734994886814e-06,
+      "loss": 0.5729,
+      "step": 13431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.636596888380414e-06,
+      "loss": 0.5733,
+      "step": 13432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6358442584182366e-06,
+      "loss": 0.6146,
+      "step": 13433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.635092059016211e-06,
+      "loss": 0.545,
+      "step": 13434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.634340290208928e-06,
+      "loss": 0.5907,
+      "step": 13435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.633588952030948e-06,
+      "loss": 0.5401,
+      "step": 13436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6328380445168192e-06,
+      "loss": 0.5545,
+      "step": 13437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6320875677010685e-06,
+      "loss": 0.5297,
+      "step": 13438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6313375216182038e-06,
+      "loss": 0.5912,
+      "step": 13439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6305879063027086e-06,
+      "loss": 0.567,
+      "step": 13440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6298387217890522e-06,
+      "loss": 0.6258,
+      "step": 13441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6290899681116813e-06,
+      "loss": 0.6519,
+      "step": 13442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.628341645305025e-06,
+      "loss": 0.5555,
+      "step": 13443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.627593753403488e-06,
+      "loss": 0.5933,
+      "step": 13444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6268462924414585e-06,
+      "loss": 0.5324,
+      "step": 13445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6260992624533077e-06,
+      "loss": 0.5348,
+      "step": 13446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.62535266347338e-06,
+      "loss": 0.5886,
+      "step": 13447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6246064955360043e-06,
+      "loss": 0.5588,
+      "step": 13448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6238607586754894e-06,
+      "loss": 0.5808,
+      "step": 13449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.623115452926125e-06,
+      "loss": 0.5616,
+      "step": 13450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.62237057832218e-06,
+      "loss": 0.5989,
+      "step": 13451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6216261348979025e-06,
+      "loss": 0.6004,
+      "step": 13452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6208821226875217e-06,
+      "loss": 0.5706,
+      "step": 13453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.620138541725249e-06,
+      "loss": 0.5668,
+      "step": 13454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.619395392045271e-06,
+      "loss": 0.6151,
+      "step": 13455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6186526736817588e-06,
+      "loss": 0.5665,
+      "step": 13456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.617910386668862e-06,
+      "loss": 0.5437,
+      "step": 13457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6171685310407114e-06,
+      "loss": 0.5811,
+      "step": 13458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6164271068314165e-06,
+      "loss": 0.5967,
+      "step": 13459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.615686114075068e-06,
+      "loss": 0.5527,
+      "step": 13460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.614945552805737e-06,
+      "loss": 0.5546,
+      "step": 13461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6142054230574735e-06,
+      "loss": 0.5967,
+      "step": 13462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.613465724864307e-06,
+      "loss": 0.6277,
+      "step": 13463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6127264582602496e-06,
+      "loss": 0.619,
+      "step": 13464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.611987623279294e-06,
+      "loss": 0.5628,
+      "step": 13465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6112492199554096e-06,
+      "loss": 0.5276,
+      "step": 13466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6105112483225493e-06,
+      "loss": 0.5753,
+      "step": 13467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6097737084146437e-06,
+      "loss": 0.5437,
+      "step": 13468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6090366002656076e-06,
+      "loss": 0.6172,
+      "step": 13469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.608299923909328e-06,
+      "loss": 0.5504,
+      "step": 13470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.607563679379679e-06,
+      "loss": 0.5665,
+      "step": 13471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.606827866710513e-06,
+      "loss": 0.5214,
+      "step": 13472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6060924859356653e-06,
+      "loss": 0.5957,
+      "step": 13473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.605357537088943e-06,
+      "loss": 0.5721,
+      "step": 13474
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6046230202041414e-06,
+      "loss": 0.6308,
+      "step": 13475
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6038889353150344e-06,
+      "loss": 0.5491,
+      "step": 13476
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6031552824553743e-06,
+      "loss": 0.6062,
+      "step": 13477
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6024220616588927e-06,
+      "loss": 0.5726,
+      "step": 13478
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.601689272959304e-06,
+      "loss": 0.5501,
+      "step": 13479
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6009569163903015e-06,
+      "loss": 0.6264,
+      "step": 13480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6002249919855582e-06,
+      "loss": 0.5166,
+      "step": 13481
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.599493499778727e-06,
+      "loss": 0.5125,
+      "step": 13482
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.598762439803444e-06,
+      "loss": 0.5746,
+      "step": 13483
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5980318120933203e-06,
+      "loss": 0.6273,
+      "step": 13484
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.597301616681952e-06,
+      "loss": 0.5795,
+      "step": 13485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.596571853602913e-06,
+      "loss": 0.6219,
+      "step": 13486
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5958425228897557e-06,
+      "loss": 0.5888,
+      "step": 13487
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.595113624576018e-06,
+      "loss": 0.5527,
+      "step": 13488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5943851586952097e-06,
+      "loss": 0.5955,
+      "step": 13489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5936571252808287e-06,
+      "loss": 0.5892,
+      "step": 13490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.592929524366348e-06,
+      "loss": 0.5031,
+      "step": 13491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5922023559852252e-06,
+      "loss": 0.6125,
+      "step": 13492
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5914756201708916e-06,
+      "loss": 0.6281,
+      "step": 13493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5907493169567633e-06,
+      "loss": 0.5599,
+      "step": 13494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5900234463762367e-06,
+      "loss": 0.5809,
+      "step": 13495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5892980084626853e-06,
+      "loss": 0.5512,
+      "step": 13496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.588573003249466e-06,
+      "loss": 0.5883,
+      "step": 13497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.587848430769913e-06,
+      "loss": 0.5893,
+      "step": 13498
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5871242910573428e-06,
+      "loss": 0.5827,
+      "step": 13499
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.58640058414505e-06,
+      "loss": 0.5495,
+      "step": 13500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5856773100663114e-06,
+      "loss": 0.5788,
+      "step": 13501
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.584954468854383e-06,
+      "loss": 0.5923,
+      "step": 13502
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5842320605425013e-06,
+      "loss": 0.5483,
+      "step": 13503
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5835100851638797e-06,
+      "loss": 0.5395,
+      "step": 13504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5827885427517153e-06,
+      "loss": 0.5664,
+      "step": 13505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.582067433339186e-06,
+      "loss": 0.5376,
+      "step": 13506
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5813467569594486e-06,
+      "loss": 0.5061,
+      "step": 13507
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5806265136456367e-06,
+      "loss": 0.6348,
+      "step": 13508
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.579906703430867e-06,
+      "loss": 0.5838,
+      "step": 13509
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5791873263482383e-06,
+      "loss": 0.5502,
+      "step": 13510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5784683824308266e-06,
+      "loss": 0.5786,
+      "step": 13511
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.577749871711688e-06,
+      "loss": 0.5261,
+      "step": 13512
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.577031794223859e-06,
+      "loss": 0.5394,
+      "step": 13513
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5763141500003567e-06,
+      "loss": 0.5209,
+      "step": 13514
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.575596939074179e-06,
+      "loss": 0.5331,
+      "step": 13515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5748801614783026e-06,
+      "loss": 0.6069,
+      "step": 13516
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5741638172456842e-06,
+      "loss": 0.5303,
+      "step": 13517
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5734479064092623e-06,
+      "loss": 0.5659,
+      "step": 13518
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.572732429001953e-06,
+      "loss": 0.5801,
+      "step": 13519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5720173850566533e-06,
+      "loss": 0.5757,
+      "step": 13520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5713027746062405e-06,
+      "loss": 0.6064,
+      "step": 13521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5705885976835755e-06,
+      "loss": 0.5423,
+      "step": 13522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5698748543214903e-06,
+      "loss": 0.5788,
+      "step": 13523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5691615445528065e-06,
+      "loss": 0.5243,
+      "step": 13524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5684486684103207e-06,
+      "loss": 0.5366,
+      "step": 13525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5677362259268124e-06,
+      "loss": 0.5792,
+      "step": 13526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.567024217135036e-06,
+      "loss": 0.4844,
+      "step": 13527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.56631264206773e-06,
+      "loss": 0.5231,
+      "step": 13528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5656015007576163e-06,
+      "loss": 0.6135,
+      "step": 13529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.564890793237388e-06,
+      "loss": 0.5367,
+      "step": 13530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5641805195397253e-06,
+      "loss": 0.5753,
+      "step": 13531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5634706796972866e-06,
+      "loss": 0.5791,
+      "step": 13532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.562761273742709e-06,
+      "loss": 0.572,
+      "step": 13533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.562052301708611e-06,
+      "loss": 0.5616,
+      "step": 13534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.561343763627592e-06,
+      "loss": 0.5199,
+      "step": 13535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5606356595322293e-06,
+      "loss": 0.5483,
+      "step": 13536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5599279894550825e-06,
+      "loss": 0.6143,
+      "step": 13537
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5592207534286887e-06,
+      "loss": 0.601,
+      "step": 13538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.558513951485565e-06,
+      "loss": 0.5634,
+      "step": 13539
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.557807583658212e-06,
+      "loss": 0.6255,
+      "step": 13540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.557101649979109e-06,
+      "loss": 0.5028,
+      "step": 13541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5563961504807124e-06,
+      "loss": 0.5309,
+      "step": 13542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.555691085195461e-06,
+      "loss": 0.5672,
+      "step": 13543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5549864541557735e-06,
+      "loss": 0.5819,
+      "step": 13544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.554282257394052e-06,
+      "loss": 0.6134,
+      "step": 13545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.553578494942669e-06,
+      "loss": 0.6085,
+      "step": 13546
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5528751668339883e-06,
+      "loss": 0.6047,
+      "step": 13547
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.552172273100345e-06,
+      "loss": 0.5559,
+      "step": 13548
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5514698137740603e-06,
+      "loss": 0.5963,
+      "step": 13549
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.550767788887433e-06,
+      "loss": 0.5417,
+      "step": 13550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5500661984727405e-06,
+      "loss": 0.5638,
+      "step": 13551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.549365042562244e-06,
+      "loss": 0.6318,
+      "step": 13552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.54866432118818e-06,
+      "loss": 0.5903,
+      "step": 13553
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.547964034382767e-06,
+      "loss": 0.542,
+      "step": 13554
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5472641821782065e-06,
+      "loss": 0.5874,
+      "step": 13555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.546564764606677e-06,
+      "loss": 0.5375,
+      "step": 13556
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5458657817003352e-06,
+      "loss": 0.5923,
+      "step": 13557
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5451672334913213e-06,
+      "loss": 0.6041,
+      "step": 13558
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5444691200117536e-06,
+      "loss": 0.5716,
+      "step": 13559
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5437714412937343e-06,
+      "loss": 0.5961,
+      "step": 13560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5430741973693374e-06,
+      "loss": 0.5576,
+      "step": 13561
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.542377388270625e-06,
+      "loss": 0.5531,
+      "step": 13562
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.541681014029637e-06,
+      "loss": 0.561,
+      "step": 13563
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5409850746783893e-06,
+      "loss": 0.5076,
+      "step": 13564
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.540289570248882e-06,
+      "loss": 0.5044,
+      "step": 13565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.539594500773095e-06,
+      "loss": 0.5531,
+      "step": 13566
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5388998662829866e-06,
+      "loss": 0.5507,
+      "step": 13567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.538205666810497e-06,
+      "loss": 0.5153,
+      "step": 13568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.537511902387544e-06,
+      "loss": 0.5531,
+      "step": 13569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5368185730460273e-06,
+      "loss": 0.5384,
+      "step": 13570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.536125678817826e-06,
+      "loss": 0.4984,
+      "step": 13571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5354332197347975e-06,
+      "loss": 0.5308,
+      "step": 13572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.534741195828783e-06,
+      "loss": 0.6105,
+      "step": 13573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5340496071316004e-06,
+      "loss": 0.5733,
+      "step": 13574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.533358453675049e-06,
+      "loss": 0.5887,
+      "step": 13575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.532667735490907e-06,
+      "loss": 0.5742,
+      "step": 13576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.531977452610933e-06,
+      "loss": 0.5532,
+      "step": 13577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.531287605066869e-06,
+      "loss": 0.6157,
+      "step": 13578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.53059819289043e-06,
+      "loss": 0.5788,
+      "step": 13579
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5299092161133166e-06,
+      "loss": 0.5554,
+      "step": 13580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.529220674767208e-06,
+      "loss": 0.5329,
+      "step": 13581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5285325688837616e-06,
+      "loss": 0.5204,
+      "step": 13582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5278448984946183e-06,
+      "loss": 0.5381,
+      "step": 13583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5271576636313965e-06,
+      "loss": 0.6494,
+      "step": 13584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.526470864325693e-06,
+      "loss": 0.5979,
+      "step": 13585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.525784500609091e-06,
+      "loss": 0.5668,
+      "step": 13586
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.525098572513144e-06,
+      "loss": 0.5562,
+      "step": 13587
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5244130800693923e-06,
+      "loss": 0.6068,
+      "step": 13588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5237280233093565e-06,
+      "loss": 0.5464,
+      "step": 13589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5230434022645347e-06,
+      "loss": 0.5632,
+      "step": 13590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5223592169664037e-06,
+      "loss": 0.5535,
+      "step": 13591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5216754674464228e-06,
+      "loss": 0.5498,
+      "step": 13592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.520992153736031e-06,
+      "loss": 0.5197,
+      "step": 13593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5203092758666482e-06,
+      "loss": 0.5612,
+      "step": 13594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5196268338696697e-06,
+      "loss": 0.563,
+      "step": 13595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.518944827776475e-06,
+      "loss": 0.5963,
+      "step": 13596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5182632576184252e-06,
+      "loss": 0.5632,
+      "step": 13597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5175821234268554e-06,
+      "loss": 0.5444,
+      "step": 13598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.516901425233084e-06,
+      "loss": 0.5543,
+      "step": 13599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.51622116306841e-06,
+      "loss": 0.6334,
+      "step": 13600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5155413369641114e-06,
+      "loss": 0.6239,
+      "step": 13601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.514861946951448e-06,
+      "loss": 0.5714,
+      "step": 13602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.514182993061656e-06,
+      "loss": 0.5274,
+      "step": 13603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.513504475325953e-06,
+      "loss": 0.5647,
+      "step": 13604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5128263937755406e-06,
+      "loss": 0.5937,
+      "step": 13605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.512148748441592e-06,
+      "loss": 0.5618,
+      "step": 13606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.511471539355267e-06,
+      "loss": 0.531,
+      "step": 13607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.510794766547703e-06,
+      "loss": 0.6223,
+      "step": 13608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5101184300500207e-06,
+      "loss": 0.5713,
+      "step": 13609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5094425298933136e-06,
+      "loss": 0.5073,
+      "step": 13610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.508767066108662e-06,
+      "loss": 0.5473,
+      "step": 13611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5080920387271226e-06,
+      "loss": 0.5399,
+      "step": 13612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.507417447779731e-06,
+      "loss": 0.5769,
+      "step": 13613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.506743293297508e-06,
+      "loss": 0.583,
+      "step": 13614
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5060695753114484e-06,
+      "loss": 0.6216,
+      "step": 13615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.50539629385253e-06,
+      "loss": 0.5405,
+      "step": 13616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5047234489517117e-06,
+      "loss": 0.5671,
+      "step": 13617
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.504051040639929e-06,
+      "loss": 0.6355,
+      "step": 13618
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5033790689480987e-06,
+      "loss": 0.6159,
+      "step": 13619
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5027075339071205e-06,
+      "loss": 0.5455,
+      "step": 13620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5020364355478678e-06,
+      "loss": 0.6178,
+      "step": 13621
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5013657739011987e-06,
+      "loss": 0.577,
+      "step": 13622
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5006955489979505e-06,
+      "loss": 0.6229,
+      "step": 13623
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5000257608689407e-06,
+      "loss": 0.5377,
+      "step": 13624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4993564095449644e-06,
+      "loss": 0.6171,
+      "step": 13625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4986874950567977e-06,
+      "loss": 0.6095,
+      "step": 13626
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4980190174351974e-06,
+      "loss": 0.5241,
+      "step": 13627
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.497350976710902e-06,
+      "loss": 0.5791,
+      "step": 13628
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4966833729146236e-06,
+      "loss": 0.5465,
+      "step": 13629
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4960162060770622e-06,
+      "loss": 0.529,
+      "step": 13630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4953494762288937e-06,
+      "loss": 0.5745,
+      "step": 13631
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4946831834007704e-06,
+      "loss": 0.6017,
+      "step": 13632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4940173276233307e-06,
+      "loss": 0.5988,
+      "step": 13633
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4933519089271895e-06,
+      "loss": 0.5871,
+      "step": 13634
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4926869273429446e-06,
+      "loss": 0.5316,
+      "step": 13635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4920223829011703e-06,
+      "loss": 0.5382,
+      "step": 13636
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.491358275632421e-06,
+      "loss": 0.513,
+      "step": 13637
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4906946055672333e-06,
+      "loss": 0.585,
+      "step": 13638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.490031372736124e-06,
+      "loss": 0.5463,
+      "step": 13639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4893685771695854e-06,
+      "loss": 0.5491,
+      "step": 13640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.488706218898093e-06,
+      "loss": 0.5849,
+      "step": 13641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4880442979521025e-06,
+      "loss": 0.5749,
+      "step": 13642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4873828143620505e-06,
+      "loss": 0.5492,
+      "step": 13643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.486721768158349e-06,
+      "loss": 0.5732,
+      "step": 13644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4860611593713926e-06,
+      "loss": 0.5626,
+      "step": 13645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.485400988031559e-06,
+      "loss": 0.5637,
+      "step": 13646
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4847412541691987e-06,
+      "loss": 0.5699,
+      "step": 13647
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.484081957814648e-06,
+      "loss": 0.5332,
+      "step": 13648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.483423098998221e-06,
+      "loss": 0.5565,
+      "step": 13649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4827646777502114e-06,
+      "loss": 0.542,
+      "step": 13650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4821066941008933e-06,
+      "loss": 0.502,
+      "step": 13651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4814491480805216e-06,
+      "loss": 0.5317,
+      "step": 13652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4807920397193275e-06,
+      "loss": 0.6108,
+      "step": 13653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4801353690475284e-06,
+      "loss": 0.527,
+      "step": 13654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.479479136095313e-06,
+      "loss": 0.6005,
+      "step": 13655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4788233408928585e-06,
+      "loss": 0.5766,
+      "step": 13656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4781679834703166e-06,
+      "loss": 0.5139,
+      "step": 13657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.477513063857821e-06,
+      "loss": 0.5425,
+      "step": 13658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4768585820854833e-06,
+      "loss": 0.5979,
+      "step": 13659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.476204538183397e-06,
+      "loss": 0.6409,
+      "step": 13660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4755509321816343e-06,
+      "loss": 0.5994,
+      "step": 13661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4748977641102504e-06,
+      "loss": 0.5361,
+      "step": 13662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4742450339992733e-06,
+      "loss": 0.6126,
+      "step": 13663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.473592741878719e-06,
+      "loss": 0.572,
+      "step": 13664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4729408877785777e-06,
+      "loss": 0.562,
+      "step": 13665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4722894717288228e-06,
+      "loss": 0.5493,
+      "step": 13666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4716384937594044e-06,
+      "loss": 0.5379,
+      "step": 13667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.470987953900255e-06,
+      "loss": 0.5303,
+      "step": 13668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.470337852181286e-06,
+      "loss": 0.4853,
+      "step": 13669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.469688188632389e-06,
+      "loss": 0.5449,
+      "step": 13670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.469038963283436e-06,
+      "loss": 0.5179,
+      "step": 13671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.468390176164276e-06,
+      "loss": 0.5897,
+      "step": 13672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.467741827304744e-06,
+      "loss": 0.6042,
+      "step": 13673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4670939167346478e-06,
+      "loss": 0.5484,
+      "step": 13674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4664464444837767e-06,
+      "loss": 0.5843,
+      "step": 13675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.465799410581904e-06,
+      "loss": 0.5318,
+      "step": 13676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.46515281505878e-06,
+      "loss": 0.6481,
+      "step": 13677
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4645066579441336e-06,
+      "loss": 0.5611,
+      "step": 13678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.463860939267674e-06,
+      "loss": 0.5847,
+      "step": 13679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.463215659059095e-06,
+      "loss": 0.605,
+      "step": 13680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4625708173480614e-06,
+      "loss": 0.6149,
+      "step": 13681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4619264141642253e-06,
+      "loss": 0.6121,
+      "step": 13682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4612824495372156e-06,
+      "loss": 0.5401,
+      "step": 13683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.460638923496642e-06,
+      "loss": 0.5477,
+      "step": 13684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4599958360720934e-06,
+      "loss": 0.5733,
+      "step": 13685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.459353187293139e-06,
+      "loss": 0.4801,
+      "step": 13686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4587109771893273e-06,
+      "loss": 0.5467,
+      "step": 13687
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.458069205790187e-06,
+      "loss": 0.5372,
+      "step": 13688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4574278731252255e-06,
+      "loss": 0.5839,
+      "step": 13689
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.456786979223932e-06,
+      "loss": 0.5448,
+      "step": 13690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.456146524115775e-06,
+      "loss": 0.5632,
+      "step": 13691
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.455506507830201e-06,
+      "loss": 0.5585,
+      "step": 13692
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.454866930396639e-06,
+      "loss": 0.5771,
+      "step": 13693
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4542277918444947e-06,
+      "loss": 0.5689,
+      "step": 13694
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.453589092203157e-06,
+      "loss": 0.5252,
+      "step": 13695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4529508315019938e-06,
+      "loss": 0.5429,
+      "step": 13696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4523130097703497e-06,
+      "loss": 0.6015,
+      "step": 13697
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.451675627037553e-06,
+      "loss": 0.648,
+      "step": 13698
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4510386833329104e-06,
+      "loss": 0.5761,
+      "step": 13699
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.450402178685707e-06,
+      "loss": 0.5253,
+      "step": 13700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4497661131252106e-06,
+      "loss": 0.6444,
+      "step": 13701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4491304866806666e-06,
+      "loss": 0.5436,
+      "step": 13702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.448495299381303e-06,
+      "loss": 0.5392,
+      "step": 13703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4478605512563203e-06,
+      "loss": 0.543,
+      "step": 13704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.447226242334908e-06,
+      "loss": 0.5242,
+      "step": 13705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4465923726462305e-06,
+      "loss": 0.5701,
+      "step": 13706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4459589422194344e-06,
+      "loss": 0.6052,
+      "step": 13707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4453259510836415e-06,
+      "loss": 0.455,
+      "step": 13708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.444693399267958e-06,
+      "loss": 0.554,
+      "step": 13709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4440612868014672e-06,
+      "loss": 0.5481,
+      "step": 13710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4434296137132364e-06,
+      "loss": 0.5289,
+      "step": 13711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4427983800323065e-06,
+      "loss": 0.5028,
+      "step": 13712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4421675857877027e-06,
+      "loss": 0.5678,
+      "step": 13713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.441537231008431e-06,
+      "loss": 0.575,
+      "step": 13714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.44090731572347e-06,
+      "loss": 0.6202,
+      "step": 13715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4402778399617847e-06,
+      "loss": 0.6328,
+      "step": 13716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4396488037523207e-06,
+      "loss": 0.5843,
+      "step": 13717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.439020207123998e-06,
+      "loss": 0.6118,
+      "step": 13718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.438392050105719e-06,
+      "loss": 0.5877,
+      "step": 13719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4377643327263693e-06,
+      "loss": 0.5778,
+      "step": 13720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.437137055014808e-06,
+      "loss": 0.527,
+      "step": 13721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.43651021699988e-06,
+      "loss": 0.554,
+      "step": 13722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.435883818710404e-06,
+      "loss": 0.544,
+      "step": 13723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.435257860175183e-06,
+      "loss": 0.574,
+      "step": 13724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.434632341422998e-06,
+      "loss": 0.5437,
+      "step": 13725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4340072624826108e-06,
+      "loss": 0.5309,
+      "step": 13726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4333826233827607e-06,
+      "loss": 0.5908,
+      "step": 13727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4327584241521697e-06,
+      "loss": 0.519,
+      "step": 13728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.43213466481954e-06,
+      "loss": 0.5615,
+      "step": 13729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4315113454135476e-06,
+      "loss": 0.4888,
+      "step": 13730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4308884659628544e-06,
+      "loss": 0.5602,
+      "step": 13731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4302660264961e-06,
+      "loss": 0.5701,
+      "step": 13732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4296440270419054e-06,
+      "loss": 0.5244,
+      "step": 13733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.429022467628868e-06,
+      "loss": 0.5725,
+      "step": 13734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4284013482855676e-06,
+      "loss": 0.575,
+      "step": 13735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.427780669040564e-06,
+      "loss": 0.5299,
+      "step": 13736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4271604299223953e-06,
+      "loss": 0.6048,
+      "step": 13737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4265406309595776e-06,
+      "loss": 0.5672,
+      "step": 13738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4259212721806116e-06,
+      "loss": 0.505,
+      "step": 13739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4253023536139734e-06,
+      "loss": 0.589,
+      "step": 13740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4246838752881232e-06,
+      "loss": 0.5091,
+      "step": 13741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4240658372314965e-06,
+      "loss": 0.666,
+      "step": 13742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.423448239472509e-06,
+      "loss": 0.5814,
+      "step": 13743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.422831082039561e-06,
+      "loss": 0.6002,
+      "step": 13744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.422214364961028e-06,
+      "loss": 0.6152,
+      "step": 13745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.421598088265265e-06,
+      "loss": 0.6027,
+      "step": 13746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4209822519806082e-06,
+      "loss": 0.5859,
+      "step": 13747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4203668561353755e-06,
+      "loss": 0.5472,
+      "step": 13748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.419751900757861e-06,
+      "loss": 0.5308,
+      "step": 13749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.41913738587634e-06,
+      "loss": 0.561,
+      "step": 13750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.418523311519068e-06,
+      "loss": 0.6654,
+      "step": 13751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4179096777142796e-06,
+      "loss": 0.5627,
+      "step": 13752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4172964844901916e-06,
+      "loss": 0.5545,
+      "step": 13753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.416683731874995e-06,
+      "loss": 0.5719,
+      "step": 13754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4160714198968662e-06,
+      "loss": 0.5301,
+      "step": 13755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4154595485839596e-06,
+      "loss": 0.5073,
+      "step": 13756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.414848117964406e-06,
+      "loss": 0.5042,
+      "step": 13757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4142371280663203e-06,
+      "loss": 0.5218,
+      "step": 13758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4136265789177963e-06,
+      "loss": 0.6207,
+      "step": 13759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4130164705469074e-06,
+      "loss": 0.5008,
+      "step": 13760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.412406802981703e-06,
+      "loss": 0.5333,
+      "step": 13761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.411797576250218e-06,
+      "loss": 0.6179,
+      "step": 13762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4111887903804644e-06,
+      "loss": 0.5718,
+      "step": 13763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4105804454004327e-06,
+      "loss": 0.5682,
+      "step": 13764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.409972541338094e-06,
+      "loss": 0.5653,
+      "step": 13765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4093650782214e-06,
+      "loss": 0.598,
+      "step": 13766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.408758056078282e-06,
+      "loss": 0.6073,
+      "step": 13767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.408151474936652e-06,
+      "loss": 0.5662,
+      "step": 13768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4075453348243974e-06,
+      "loss": 0.4896,
+      "step": 13769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.406939635769391e-06,
+      "loss": 0.5838,
+      "step": 13770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.406334377799482e-06,
+      "loss": 0.5495,
+      "step": 13771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.405729560942498e-06,
+      "loss": 0.5309,
+      "step": 13772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4051251852262495e-06,
+      "loss": 0.6284,
+      "step": 13773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4045212506785257e-06,
+      "loss": 0.4652,
+      "step": 13774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.403917757327097e-06,
+      "loss": 0.5245,
+      "step": 13775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4033147051997077e-06,
+      "loss": 0.5546,
+      "step": 13776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.402712094324089e-06,
+      "loss": 0.5792,
+      "step": 13777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4021099247279467e-06,
+      "loss": 0.4692,
+      "step": 13778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.401508196438972e-06,
+      "loss": 0.5283,
+      "step": 13779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.400906909484827e-06,
+      "loss": 0.5724,
+      "step": 13780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.400306063893162e-06,
+      "loss": 0.5786,
+      "step": 13781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3997056596916036e-06,
+      "loss": 0.5535,
+      "step": 13782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3991056969077567e-06,
+      "loss": 0.5609,
+      "step": 13783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3985061755692075e-06,
+      "loss": 0.5706,
+      "step": 13784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3979070957035222e-06,
+      "loss": 0.5668,
+      "step": 13785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.397308457338247e-06,
+      "loss": 0.5622,
+      "step": 13786
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3967102605009054e-06,
+      "loss": 0.534,
+      "step": 13787
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.396112505219004e-06,
+      "loss": 0.5474,
+      "step": 13788
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.395515191520026e-06,
+      "loss": 0.5203,
+      "step": 13789
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3949183194314386e-06,
+      "loss": 0.6462,
+      "step": 13790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.394321888980681e-06,
+      "loss": 0.6101,
+      "step": 13791
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3937259001951797e-06,
+      "loss": 0.5829,
+      "step": 13792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3931303531023376e-06,
+      "loss": 0.5756,
+      "step": 13793
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3925352477295385e-06,
+      "loss": 0.5581,
+      "step": 13794
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3919405841041433e-06,
+      "loss": 0.6073,
+      "step": 13795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.391346362253495e-06,
+      "loss": 0.6171,
+      "step": 13796
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3907525822049174e-06,
+      "loss": 0.5042,
+      "step": 13797
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3901592439857093e-06,
+      "loss": 0.5494,
+      "step": 13798
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.389566347623155e-06,
+      "loss": 0.633,
+      "step": 13799
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.388973893144513e-06,
+      "loss": 0.5396,
+      "step": 13800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3883818805770266e-06,
+      "loss": 0.5439,
+      "step": 13801
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.387790309947915e-06,
+      "loss": 0.5671,
+      "step": 13802
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.387199181284378e-06,
+      "loss": 0.589,
+      "step": 13803
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.386608494613596e-06,
+      "loss": 0.5374,
+      "step": 13804
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3860182499627306e-06,
+      "loss": 0.5686,
+      "step": 13805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3854284473589175e-06,
+      "loss": 0.5888,
+      "step": 13806
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.384839086829277e-06,
+      "loss": 0.6285,
+      "step": 13807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.384250168400908e-06,
+      "loss": 0.5574,
+      "step": 13808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3836616921008893e-06,
+      "loss": 0.5986,
+      "step": 13809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3830736579562775e-06,
+      "loss": 0.5168,
+      "step": 13810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.38248606599411e-06,
+      "loss": 0.5545,
+      "step": 13811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3818989162414043e-06,
+      "loss": 0.5739,
+      "step": 13812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3813122087251595e-06,
+      "loss": 0.5539,
+      "step": 13813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3807259434723497e-06,
+      "loss": 0.5942,
+      "step": 13814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3801401205099318e-06,
+      "loss": 0.587,
+      "step": 13815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3795547398648425e-06,
+      "loss": 0.5535,
+      "step": 13816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3789698015639952e-06,
+      "loss": 0.4947,
+      "step": 13817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.378385305634288e-06,
+      "loss": 0.5278,
+      "step": 13818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.377801252102593e-06,
+      "loss": 0.5752,
+      "step": 13819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.377217640995766e-06,
+      "loss": 0.5569,
+      "step": 13820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3766344723406423e-06,
+      "loss": 0.5466,
+      "step": 13821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.376051746164033e-06,
+      "loss": 0.5249,
+      "step": 13822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3754694624927355e-06,
+      "loss": 0.5431,
+      "step": 13823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3748876213535213e-06,
+      "loss": 0.5613,
+      "step": 13824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.374306222773141e-06,
+      "loss": 0.5294,
+      "step": 13825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.37372526677833e-06,
+      "loss": 0.5685,
+      "step": 13826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.373144753395799e-06,
+      "loss": 0.5119,
+      "step": 13827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.37256468265224e-06,
+      "loss": 0.6531,
+      "step": 13828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.371985054574324e-06,
+      "loss": 0.5398,
+      "step": 13829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3714058691887036e-06,
+      "loss": 0.6024,
+      "step": 13830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.370827126522009e-06,
+      "loss": 0.598,
+      "step": 13831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3702488266008475e-06,
+      "loss": 0.5694,
+      "step": 13832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.369670969451813e-06,
+      "loss": 0.5924,
+      "step": 13833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.369093555101474e-06,
+      "loss": 0.5784,
+      "step": 13834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.368516583576378e-06,
+      "loss": 0.5909,
+      "step": 13835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3679400549030575e-06,
+      "loss": 0.5154,
+      "step": 13836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3673639691080175e-06,
+      "loss": 0.5722,
+      "step": 13837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3667883262177486e-06,
+      "loss": 0.5498,
+      "step": 13838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3662131262587185e-06,
+      "loss": 0.5776,
+      "step": 13839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3656383692573727e-06,
+      "loss": 0.59,
+      "step": 13840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3650640552401395e-06,
+      "loss": 0.5273,
+      "step": 13841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3644901842334257e-06,
+      "loss": 0.607,
+      "step": 13842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3639167562636193e-06,
+      "loss": 0.5674,
+      "step": 13843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.363343771357083e-06,
+      "loss": 0.5725,
+      "step": 13844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.362771229540164e-06,
+      "loss": 0.5473,
+      "step": 13845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.362199130839188e-06,
+      "loss": 0.5688,
+      "step": 13846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3616274752804604e-06,
+      "loss": 0.5341,
+      "step": 13847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3610562628902635e-06,
+      "loss": 0.5685,
+      "step": 13848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.360485493694862e-06,
+      "loss": 0.4704,
+      "step": 13849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3599151677205024e-06,
+      "loss": 0.5475,
+      "step": 13850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.359345284993404e-06,
+      "loss": 0.5567,
+      "step": 13851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3587758455397727e-06,
+      "loss": 0.5763,
+      "step": 13852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.358206849385789e-06,
+      "loss": 0.5105,
+      "step": 13853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3576382965576163e-06,
+      "loss": 0.5206,
+      "step": 13854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3570701870813975e-06,
+      "loss": 0.5658,
+      "step": 13855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3565025209832516e-06,
+      "loss": 0.608,
+      "step": 13856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3559352982892826e-06,
+      "loss": 0.5621,
+      "step": 13857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3553685190255696e-06,
+      "loss": 0.5938,
+      "step": 13858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3548021832181714e-06,
+      "loss": 0.5018,
+      "step": 13859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3542362908931306e-06,
+      "loss": 0.5621,
+      "step": 13860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3536708420764645e-06,
+      "loss": 0.5877,
+      "step": 13861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3531058367941742e-06,
+      "loss": 0.5167,
+      "step": 13862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3525412750722365e-06,
+      "loss": 0.529,
+      "step": 13863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3519771569366103e-06,
+      "loss": 0.5753,
+      "step": 13864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3514134824132352e-06,
+      "loss": 0.5898,
+      "step": 13865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3508502515280266e-06,
+      "loss": 0.5282,
+      "step": 13866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.350287464306882e-06,
+      "loss": 0.5697,
+      "step": 13867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.349725120775679e-06,
+      "loss": 0.5682,
+      "step": 13868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3491632209602722e-06,
+      "loss": 0.541,
+      "step": 13869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3486017648865003e-06,
+      "loss": 0.5424,
+      "step": 13870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3480407525801763e-06,
+      "loss": 0.57,
+      "step": 13871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.347480184067096e-06,
+      "loss": 0.572,
+      "step": 13872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.346920059373037e-06,
+      "loss": 0.6103,
+      "step": 13873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.346360378523749e-06,
+      "loss": 0.602,
+      "step": 13874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3458011415449677e-06,
+      "loss": 0.5777,
+      "step": 13875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.345242348462408e-06,
+      "loss": 0.619,
+      "step": 13876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3446839993017613e-06,
+      "loss": 0.5386,
+      "step": 13877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3441260940887014e-06,
+      "loss": 0.507,
+      "step": 13878
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3435686328488782e-06,
+      "loss": 0.5521,
+      "step": 13879
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3430116156079276e-06,
+      "loss": 0.6016,
+      "step": 13880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3424550423914573e-06,
+      "loss": 0.5385,
+      "step": 13881
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.34189891322506e-06,
+      "loss": 0.5467,
+      "step": 13882
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3413432281343064e-06,
+      "loss": 0.6278,
+      "step": 13883
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.340787987144746e-06,
+      "loss": 0.611,
+      "step": 13884
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.340233190281909e-06,
+      "loss": 0.5414,
+      "step": 13885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3396788375713043e-06,
+      "loss": 0.6138,
+      "step": 13886
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3391249290384213e-06,
+      "loss": 0.5555,
+      "step": 13887
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.338571464708729e-06,
+      "loss": 0.6073,
+      "step": 13888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.338018444607674e-06,
+      "loss": 0.5461,
+      "step": 13889
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3374658687606846e-06,
+      "loss": 0.5665,
+      "step": 13890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3369137371931686e-06,
+      "loss": 0.5101,
+      "step": 13891
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.336362049930513e-06,
+      "loss": 0.5844,
+      "step": 13892
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3358108069980837e-06,
+      "loss": 0.5585,
+      "step": 13893
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3352600084212253e-06,
+      "loss": 0.5491,
+      "step": 13894
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3347096542252645e-06,
+      "loss": 0.5573,
+      "step": 13895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.334159744435508e-06,
+      "loss": 0.5097,
+      "step": 13896
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3336102790772364e-06,
+      "loss": 0.591,
+      "step": 13897
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3330612581757174e-06,
+      "loss": 0.5326,
+      "step": 13898
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3325126817561953e-06,
+      "loss": 0.5398,
+      "step": 13899
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3319645498438895e-06,
+      "loss": 0.6432,
+      "step": 13900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3314168624640044e-06,
+      "loss": 0.611,
+      "step": 13901
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.330869619641725e-06,
+      "loss": 0.5541,
+      "step": 13902
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3303228214022102e-06,
+      "loss": 0.5469,
+      "step": 13903
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.329776467770604e-06,
+      "loss": 0.5807,
+      "step": 13904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3292305587720254e-06,
+      "loss": 0.5712,
+      "step": 13905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3286850944315766e-06,
+      "loss": 0.6057,
+      "step": 13906
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3281400747743377e-06,
+      "loss": 0.547,
+      "step": 13907
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.327595499825367e-06,
+      "loss": 0.5491,
+      "step": 13908
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3270513696097056e-06,
+      "loss": 0.5708,
+      "step": 13909
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.326507684152372e-06,
+      "loss": 0.479,
+      "step": 13910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.325964443478365e-06,
+      "loss": 0.5242,
+      "step": 13911
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3254216476126603e-06,
+      "loss": 0.4923,
+      "step": 13912
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.324879296580217e-06,
+      "loss": 0.5107,
+      "step": 13913
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3243373904059742e-06,
+      "loss": 0.488,
+      "step": 13914
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.323795929114845e-06,
+      "loss": 0.5294,
+      "step": 13915
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3232549127317274e-06,
+      "loss": 0.536,
+      "step": 13916
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.322714341281497e-06,
+      "loss": 0.545,
+      "step": 13917
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.322174214789008e-06,
+      "loss": 0.5986,
+      "step": 13918
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.321634533279097e-06,
+      "loss": 0.554,
+      "step": 13919
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.321095296776578e-06,
+      "loss": 0.5569,
+      "step": 13920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.320556505306243e-06,
+      "loss": 0.5321,
+      "step": 13921
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3200181588928683e-06,
+      "loss": 0.5111,
+      "step": 13922
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3194802575612053e-06,
+      "loss": 0.5775,
+      "step": 13923
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.318942801335985e-06,
+      "loss": 0.5339,
+      "step": 13924
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3184057902419225e-06,
+      "loss": 0.5291,
+      "step": 13925
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3178692243037075e-06,
+      "loss": 0.6005,
+      "step": 13926
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.317333103546011e-06,
+      "loss": 0.5355,
+      "step": 13927
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3167974279934838e-06,
+      "loss": 0.56,
+      "step": 13928
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.316262197670756e-06,
+      "loss": 0.6199,
+      "step": 13929
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.315727412602439e-06,
+      "loss": 0.5622,
+      "step": 13930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.315193072813119e-06,
+      "loss": 0.5433,
+      "step": 13931
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3146591783273668e-06,
+      "loss": 0.6045,
+      "step": 13932
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.314125729169732e-06,
+      "loss": 0.5855,
+      "step": 13933
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3135927253647374e-06,
+      "loss": 0.5737,
+      "step": 13934
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3130601669368947e-06,
+      "loss": 0.5832,
+      "step": 13935
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.31252805391069e-06,
+      "loss": 0.5504,
+      "step": 13936
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3119963863105878e-06,
+      "loss": 0.5783,
+      "step": 13937
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3114651641610354e-06,
+      "loss": 0.5398,
+      "step": 13938
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3109343874864587e-06,
+      "loss": 0.545,
+      "step": 13939
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.310404056311262e-06,
+      "loss": 0.5019,
+      "step": 13940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3098741706598296e-06,
+      "loss": 0.5864,
+      "step": 13941
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3093447305565255e-06,
+      "loss": 0.5613,
+      "step": 13942
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3088157360256928e-06,
+      "loss": 0.5506,
+      "step": 13943
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3082871870916547e-06,
+      "loss": 0.5287,
+      "step": 13944
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3077590837787147e-06,
+      "loss": 0.5799,
+      "step": 13945
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.307231426111153e-06,
+      "loss": 0.4911,
+      "step": 13946
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3067042141132315e-06,
+      "loss": 0.5924,
+      "step": 13947
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3061774478091927e-06,
+      "loss": 0.547,
+      "step": 13948
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.305651127223255e-06,
+      "loss": 0.5745,
+      "step": 13949
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3051252523796197e-06,
+      "loss": 0.5692,
+      "step": 13950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.304599823302466e-06,
+      "loss": 0.5567,
+      "step": 13951
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.304074840015953e-06,
+      "loss": 0.5716,
+      "step": 13952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3035503025442197e-06,
+      "loss": 0.5531,
+      "step": 13953
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3030262109113823e-06,
+      "loss": 0.5659,
+      "step": 13954
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.302502565141541e-06,
+      "loss": 0.5852,
+      "step": 13955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.301979365258772e-06,
+      "loss": 0.566,
+      "step": 13956
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3014566112871295e-06,
+      "loss": 0.5487,
+      "step": 13957
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3009343032506526e-06,
+      "loss": 0.5675,
+      "step": 13958
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3004124411733543e-06,
+      "loss": 0.5098,
+      "step": 13959
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.299891025079233e-06,
+      "loss": 0.5559,
+      "step": 13960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2993700549922597e-06,
+      "loss": 0.5546,
+      "step": 13961
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2988495309363892e-06,
+      "loss": 0.5971,
+      "step": 13962
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.298329452935555e-06,
+      "loss": 0.5591,
+      "step": 13963
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2978098210136726e-06,
+      "loss": 0.5561,
+      "step": 13964
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2972906351946313e-06,
+      "loss": 0.549,
+      "step": 13965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2967718955023045e-06,
+      "loss": 0.5245,
+      "step": 13966
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2962536019605443e-06,
+      "loss": 0.5603,
+      "step": 13967
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2957357545931795e-06,
+      "loss": 0.6091,
+      "step": 13968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2952183534240213e-06,
+      "loss": 0.5668,
+      "step": 13969
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2947013984768597e-06,
+      "loss": 0.5436,
+      "step": 13970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.294184889775466e-06,
+      "loss": 0.5352,
+      "step": 13971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.293668827343586e-06,
+      "loss": 0.5638,
+      "step": 13972
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2931532112049508e-06,
+      "loss": 0.5387,
+      "step": 13973
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.292638041383265e-06,
+      "loss": 0.5563,
+      "step": 13974
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2921233179022208e-06,
+      "loss": 0.5599,
+      "step": 13975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.29160904078548e-06,
+      "loss": 0.558,
+      "step": 13976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.291095210056691e-06,
+      "loss": 0.5565,
+      "step": 13977
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.29058182573948e-06,
+      "loss": 0.5778,
+      "step": 13978
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.290068887857452e-06,
+      "loss": 0.6063,
+      "step": 13979
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2895563964341894e-06,
+      "loss": 0.5491,
+      "step": 13980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2890443514932595e-06,
+      "loss": 0.4616,
+      "step": 13981
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.288532753058206e-06,
+      "loss": 0.447,
+      "step": 13982
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.288021601152549e-06,
+      "loss": 0.5615,
+      "step": 13983
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.287510895799793e-06,
+      "loss": 0.5465,
+      "step": 13984
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.287000637023419e-06,
+      "loss": 0.5761,
+      "step": 13985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.28649082484689e-06,
+      "loss": 0.5422,
+      "step": 13986
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.285981459293646e-06,
+      "loss": 0.5645,
+      "step": 13987
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.285472540387107e-06,
+      "loss": 0.547,
+      "step": 13988
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2849640681506747e-06,
+      "loss": 0.5286,
+      "step": 13989
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.284456042607728e-06,
+      "loss": 0.577,
+      "step": 13990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.283948463781624e-06,
+      "loss": 0.5234,
+      "step": 13991
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2834413316957005e-06,
+      "loss": 0.6257,
+      "step": 13992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2829346463732784e-06,
+      "loss": 0.5554,
+      "step": 13993
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2824284078376535e-06,
+      "loss": 0.5472,
+      "step": 13994
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2819226161121014e-06,
+      "loss": 0.5679,
+      "step": 13995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2814172712198783e-06,
+      "loss": 0.5567,
+      "step": 13996
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.280912373184221e-06,
+      "loss": 0.5195,
+      "step": 13997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.280407922028345e-06,
+      "loss": 0.5508,
+      "step": 13998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2799039177754422e-06,
+      "loss": 0.529,
+      "step": 13999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2794003604486884e-06,
+      "loss": 0.6213,
+      "step": 14000
+    },
+    {
+      "epoch": 0.93,
+      "eval_text_loss": 0.6506561636924744,
+      "eval_text_runtime": 1.8169,
+      "eval_text_samples_per_second": 40.729,
+      "eval_text_steps_per_second": 1.101,
+      "step": 14000
+    },
+    {
+      "epoch": 0.93,
+      "eval_svit_loss": 0.751133918762207,
+      "eval_svit_runtime": 6.5659,
+      "eval_svit_samples_per_second": 124.279,
+      "eval_svit_steps_per_second": 1.98,
+      "step": 14000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.278897250071237e-06,
+      "loss": 0.5845,
+      "step": 14001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2783945866662195e-06,
+      "loss": 0.5217,
+      "step": 14002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.277892370256749e-06,
+      "loss": 0.5569,
+      "step": 14003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2773906008659178e-06,
+      "loss": 0.5737,
+      "step": 14004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.276889278516795e-06,
+      "loss": 0.5752,
+      "step": 14005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2763884032324346e-06,
+      "loss": 0.5266,
+      "step": 14006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.275887975035863e-06,
+      "loss": 0.522,
+      "step": 14007
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2753879939500934e-06,
+      "loss": 0.5789,
+      "step": 14008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.274888459998113e-06,
+      "loss": 0.5402,
+      "step": 14009
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2743893732028878e-06,
+      "loss": 0.5687,
+      "step": 14010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2738907335873688e-06,
+      "loss": 0.552,
+      "step": 14011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.273392541174482e-06,
+      "loss": 0.574,
+      "step": 14012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2728947959871353e-06,
+      "loss": 0.5518,
+      "step": 14013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2723974980482123e-06,
+      "loss": 0.5664,
+      "step": 14014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2719006473805805e-06,
+      "loss": 0.5928,
+      "step": 14015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.271404244007085e-06,
+      "loss": 0.5587,
+      "step": 14016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2709082879505487e-06,
+      "loss": 0.4894,
+      "step": 14017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2704127792337775e-06,
+      "loss": 0.5104,
+      "step": 14018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.269917717879553e-06,
+      "loss": 0.5648,
+      "step": 14019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2694231039106386e-06,
+      "loss": 0.5905,
+      "step": 14020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.268928937349776e-06,
+      "loss": 0.5332,
+      "step": 14021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.268435218219687e-06,
+      "loss": 0.5818,
+      "step": 14022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2679419465430733e-06,
+      "loss": 0.5225,
+      "step": 14023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2674491223426166e-06,
+      "loss": 0.544,
+      "step": 14024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2669567456409725e-06,
+      "loss": 0.51,
+      "step": 14025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.266464816460784e-06,
+      "loss": 0.5575,
+      "step": 14026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2659733348246683e-06,
+      "loss": 0.4874,
+      "step": 14027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2654823007552242e-06,
+      "loss": 0.5646,
+      "step": 14028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.264991714275029e-06,
+      "loss": 0.5651,
+      "step": 14029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2645015754066393e-06,
+      "loss": 0.5264,
+      "step": 14030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.264011884172591e-06,
+      "loss": 0.594,
+      "step": 14031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2635226405954015e-06,
+      "loss": 0.5786,
+      "step": 14032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2630338446975652e-06,
+      "loss": 0.5602,
+      "step": 14033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.262545496501556e-06,
+      "loss": 0.525,
+      "step": 14034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2620575960298298e-06,
+      "loss": 0.6044,
+      "step": 14035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.261570143304818e-06,
+      "loss": 0.527,
+      "step": 14036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2610831383489342e-06,
+      "loss": 0.5161,
+      "step": 14037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2605965811845712e-06,
+      "loss": 0.5447,
+      "step": 14038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2601104718341006e-06,
+      "loss": 0.5656,
+      "step": 14039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2596248103198726e-06,
+      "loss": 0.5332,
+      "step": 14040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2591395966642197e-06,
+      "loss": 0.6091,
+      "step": 14041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2586548308894497e-06,
+      "loss": 0.6395,
+      "step": 14042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2581705130178545e-06,
+      "loss": 0.5586,
+      "step": 14043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2576866430716996e-06,
+      "loss": 0.485,
+      "step": 14044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2572032210732344e-06,
+      "loss": 0.5866,
+      "step": 14045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.256720247044687e-06,
+      "loss": 0.5554,
+      "step": 14046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2562377210082654e-06,
+      "loss": 0.5359,
+      "step": 14047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.255755642986154e-06,
+      "loss": 0.5808,
+      "step": 14048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2552740130005183e-06,
+      "loss": 0.6341,
+      "step": 14049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.254792831073506e-06,
+      "loss": 0.5303,
+      "step": 14050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2543120972272385e-06,
+      "loss": 0.5756,
+      "step": 14051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.253831811483822e-06,
+      "loss": 0.5022,
+      "step": 14052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2533519738653387e-06,
+      "loss": 0.5315,
+      "step": 14053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.252872584393852e-06,
+      "loss": 0.5714,
+      "step": 14054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2523936430914043e-06,
+      "loss": 0.5663,
+      "step": 14055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.251915149980015e-06,
+      "loss": 0.5298,
+      "step": 14056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2514371050816887e-06,
+      "loss": 0.5854,
+      "step": 14057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2509595084184036e-06,
+      "loss": 0.5197,
+      "step": 14058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2504823600121188e-06,
+      "loss": 0.5608,
+      "step": 14059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2500056598847733e-06,
+      "loss": 0.5141,
+      "step": 14060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2495294080582875e-06,
+      "loss": 0.5751,
+      "step": 14061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.249053604554559e-06,
+      "loss": 0.5579,
+      "step": 14062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2485782493954625e-06,
+      "loss": 0.5466,
+      "step": 14063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.248103342602857e-06,
+      "loss": 0.565,
+      "step": 14064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.247628884198577e-06,
+      "loss": 0.5611,
+      "step": 14065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2471548742044406e-06,
+      "loss": 0.5663,
+      "step": 14066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2466813126422386e-06,
+      "loss": 0.5329,
+      "step": 14067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2462081995337485e-06,
+      "loss": 0.5379,
+      "step": 14068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.245735534900723e-06,
+      "loss": 0.5329,
+      "step": 14069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2452633187648932e-06,
+      "loss": 0.538,
+      "step": 14070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2447915511479737e-06,
+      "loss": 0.5388,
+      "step": 14071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.244320232071655e-06,
+      "loss": 0.5794,
+      "step": 14072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.243849361557611e-06,
+      "loss": 0.5241,
+      "step": 14073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2433789396274874e-06,
+      "loss": 0.5875,
+      "step": 14074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.242908966302917e-06,
+      "loss": 0.5465,
+      "step": 14075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2424394416055076e-06,
+      "loss": 0.5367,
+      "step": 14076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2419703655568503e-06,
+      "loss": 0.5822,
+      "step": 14077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.24150173817851e-06,
+      "loss": 0.5424,
+      "step": 14078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2410335594920343e-06,
+      "loss": 0.48,
+      "step": 14079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2405658295189515e-06,
+      "loss": 0.5761,
+      "step": 14080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2400985482807683e-06,
+      "loss": 0.5506,
+      "step": 14081
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.239631715798967e-06,
+      "loss": 0.5661,
+      "step": 14082
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2391653320950148e-06,
+      "loss": 0.5418,
+      "step": 14083
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.238699397190355e-06,
+      "loss": 0.5394,
+      "step": 14084
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2382339111064114e-06,
+      "loss": 0.5197,
+      "step": 14085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2377688738645856e-06,
+      "loss": 0.5022,
+      "step": 14086
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2373042854862616e-06,
+      "loss": 0.5658,
+      "step": 14087
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2368401459928003e-06,
+      "loss": 0.595,
+      "step": 14088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2363764554055426e-06,
+      "loss": 0.5105,
+      "step": 14089
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2359132137458087e-06,
+      "loss": 0.5469,
+      "step": 14090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2354504210348982e-06,
+      "loss": 0.5942,
+      "step": 14091
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2349880772940924e-06,
+      "loss": 0.5109,
+      "step": 14092
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2345261825446454e-06,
+      "loss": 0.5081,
+      "step": 14093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2340647368077987e-06,
+      "loss": 0.5702,
+      "step": 14094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2336037401047662e-06,
+      "loss": 0.5231,
+      "step": 14095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2331431924567484e-06,
+      "loss": 0.4992,
+      "step": 14096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2326830938849167e-06,
+      "loss": 0.5034,
+      "step": 14097
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.232223444410429e-06,
+      "loss": 0.5839,
+      "step": 14098
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2317642440544188e-06,
+      "loss": 0.5856,
+      "step": 14099
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.231305492838e-06,
+      "loss": 0.521,
+      "step": 14100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.230847190782266e-06,
+      "loss": 0.5801,
+      "step": 14101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.23038933790829e-06,
+      "loss": 0.61,
+      "step": 14102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.229931934237122e-06,
+      "loss": 0.5535,
+      "step": 14103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2294749797897955e-06,
+      "loss": 0.557,
+      "step": 14104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.229018474587319e-06,
+      "loss": 0.5093,
+      "step": 14105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.228562418650684e-06,
+      "loss": 0.6091,
+      "step": 14106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.228106812000861e-06,
+      "loss": 0.5522,
+      "step": 14107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2276516546587947e-06,
+      "loss": 0.5987,
+      "step": 14108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.227196946645416e-06,
+      "loss": 0.4711,
+      "step": 14109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2267426879816313e-06,
+      "loss": 0.5608,
+      "step": 14110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.226288878688329e-06,
+      "loss": 0.496,
+      "step": 14111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2258355187863712e-06,
+      "loss": 0.5437,
+      "step": 14112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.225382608296606e-06,
+      "loss": 0.5943,
+      "step": 14113
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.224930147239858e-06,
+      "loss": 0.5245,
+      "step": 14114
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2244781356369307e-06,
+      "loss": 0.5559,
+      "step": 14115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.224026573508607e-06,
+      "loss": 0.574,
+      "step": 14116
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.223575460875649e-06,
+      "loss": 0.5463,
+      "step": 14117
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2231247977588013e-06,
+      "loss": 0.5998,
+      "step": 14118
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2226745841787824e-06,
+      "loss": 0.5033,
+      "step": 14119
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.222224820156295e-06,
+      "loss": 0.5441,
+      "step": 14120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2217755057120164e-06,
+      "loss": 0.5084,
+      "step": 14121
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.221326640866608e-06,
+      "loss": 0.5612,
+      "step": 14122
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.220878225640709e-06,
+      "loss": 0.5643,
+      "step": 14123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2204302600549356e-06,
+      "loss": 0.5225,
+      "step": 14124
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.219982744129886e-06,
+      "loss": 0.5509,
+      "step": 14125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2195356778861377e-06,
+      "loss": 0.493,
+      "step": 14126
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2190890613442453e-06,
+      "loss": 0.563,
+      "step": 14127
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2186428945247443e-06,
+      "loss": 0.5371,
+      "step": 14128
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2181971774481482e-06,
+      "loss": 0.5599,
+      "step": 14129
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2177519101349544e-06,
+      "loss": 0.5484,
+      "step": 14130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2173070926056318e-06,
+      "loss": 0.5686,
+      "step": 14131
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2168627248806354e-06,
+      "loss": 0.6182,
+      "step": 14132
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.216418806980396e-06,
+      "loss": 0.5107,
+      "step": 14133
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.215975338925326e-06,
+      "loss": 0.5723,
+      "step": 14134
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.215532320735815e-06,
+      "loss": 0.5831,
+      "step": 14135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2150897524322327e-06,
+      "loss": 0.5151,
+      "step": 14136
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.214647634034928e-06,
+      "loss": 0.5676,
+      "step": 14137
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.21420596556423e-06,
+      "loss": 0.5164,
+      "step": 14138
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2137647470404467e-06,
+      "loss": 0.4957,
+      "step": 14139
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2133239784838646e-06,
+      "loss": 0.5524,
+      "step": 14140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.212883659914751e-06,
+      "loss": 0.6153,
+      "step": 14141
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2124437913533503e-06,
+      "loss": 0.6212,
+      "step": 14142
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.212004372819887e-06,
+      "loss": 0.5397,
+      "step": 14143
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.211565404334567e-06,
+      "loss": 0.5986,
+      "step": 14144
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.211126885917574e-06,
+      "loss": 0.6251,
+      "step": 14145
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2106888175890698e-06,
+      "loss": 0.5714,
+      "step": 14146
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2102511993691958e-06,
+      "loss": 0.5013,
+      "step": 14147
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2098140312780747e-06,
+      "loss": 0.5656,
+      "step": 14148
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.209377313335809e-06,
+      "loss": 0.5426,
+      "step": 14149
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2089410455624763e-06,
+      "loss": 0.5543,
+      "step": 14150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2085052279781364e-06,
+      "loss": 0.5573,
+      "step": 14151
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2080698606028287e-06,
+      "loss": 0.5246,
+      "step": 14152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2076349434565703e-06,
+      "loss": 0.4653,
+      "step": 14153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2072004765593592e-06,
+      "loss": 0.4856,
+      "step": 14154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2067664599311723e-06,
+      "loss": 0.5804,
+      "step": 14155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.206332893591965e-06,
+      "loss": 0.5659,
+      "step": 14156
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.205899777561672e-06,
+      "loss": 0.5301,
+      "step": 14157
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.205467111860209e-06,
+      "loss": 0.606,
+      "step": 14158
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2050348965074697e-06,
+      "loss": 0.5628,
+      "step": 14159
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2046031315233275e-06,
+      "loss": 0.5474,
+      "step": 14160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.204171816927633e-06,
+      "loss": 0.505,
+      "step": 14161
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2037409527402188e-06,
+      "loss": 0.4889,
+      "step": 14162
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2033105389808964e-06,
+      "loss": 0.573,
+      "step": 14163
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2028805756694564e-06,
+      "loss": 0.5439,
+      "step": 14164
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.202451062825666e-06,
+      "loss": 0.5789,
+      "step": 14165
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.202022000469276e-06,
+      "loss": 0.5214,
+      "step": 14166
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.201593388620015e-06,
+      "loss": 0.5751,
+      "step": 14167
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2011652272975893e-06,
+      "loss": 0.5807,
+      "step": 14168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.200737516521685e-06,
+      "loss": 0.54,
+      "step": 14169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2003102563119697e-06,
+      "loss": 0.5241,
+      "step": 14170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.199883446688087e-06,
+      "loss": 0.6014,
+      "step": 14171
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1994570876696623e-06,
+      "loss": 0.5988,
+      "step": 14172
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1990311792762998e-06,
+      "loss": 0.5521,
+      "step": 14173
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1986057215275816e-06,
+      "loss": 0.5757,
+      "step": 14174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1981807144430725e-06,
+      "loss": 0.6008,
+      "step": 14175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1977561580423105e-06,
+      "loss": 0.5878,
+      "step": 14176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1973320523448194e-06,
+      "loss": 0.5462,
+      "step": 14177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.196908397370098e-06,
+      "loss": 0.5529,
+      "step": 14178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1964851931376264e-06,
+      "loss": 0.5058,
+      "step": 14179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.196062439666863e-06,
+      "loss": 0.5497,
+      "step": 14180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1956401369772453e-06,
+      "loss": 0.5382,
+      "step": 14181
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.195218285088192e-06,
+      "loss": 0.5669,
+      "step": 14182
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.194796884019099e-06,
+      "loss": 0.5092,
+      "step": 14183
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1943759337893415e-06,
+      "loss": 0.5893,
+      "step": 14184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.193955434418276e-06,
+      "loss": 0.5669,
+      "step": 14185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1935353859252363e-06,
+      "loss": 0.5245,
+      "step": 14186
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1931157883295343e-06,
+      "loss": 0.5506,
+      "step": 14187
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.192696641650466e-06,
+      "loss": 0.5516,
+      "step": 14188
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1922779459073007e-06,
+      "loss": 0.5508,
+      "step": 14189
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1918597011192923e-06,
+      "loss": 0.5922,
+      "step": 14190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1914419073056704e-06,
+      "loss": 0.546,
+      "step": 14191
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1910245644856442e-06,
+      "loss": 0.5382,
+      "step": 14192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1906076726784053e-06,
+      "loss": 0.5628,
+      "step": 14193
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1901912319031208e-06,
+      "loss": 0.5098,
+      "step": 14194
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1897752421789378e-06,
+      "loss": 0.5141,
+      "step": 14195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.189359703524984e-06,
+      "loss": 0.5805,
+      "step": 14196
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1889446159603648e-06,
+      "loss": 0.5713,
+      "step": 14197
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1885299795041683e-06,
+      "loss": 0.5327,
+      "step": 14198
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1881157941754576e-06,
+      "loss": 0.5431,
+      "step": 14199
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1877020599932755e-06,
+      "loss": 0.5353,
+      "step": 14200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.187288776976648e-06,
+      "loss": 0.5749,
+      "step": 14201
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1868759451445755e-06,
+      "loss": 0.5321,
+      "step": 14202
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.186463564516042e-06,
+      "loss": 0.5744,
+      "step": 14203
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.186051635110005e-06,
+      "loss": 0.5851,
+      "step": 14204
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.185640156945409e-06,
+      "loss": 0.5459,
+      "step": 14205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.185229130041171e-06,
+      "loss": 0.5764,
+      "step": 14206
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1848185544161905e-06,
+      "loss": 0.5601,
+      "step": 14207
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1844084300893455e-06,
+      "loss": 0.5387,
+      "step": 14208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1839987570794955e-06,
+      "loss": 0.6011,
+      "step": 14209
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.183589535405474e-06,
+      "loss": 0.5593,
+      "step": 14210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.183180765086097e-06,
+      "loss": 0.5831,
+      "step": 14211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1827724461401614e-06,
+      "loss": 0.5469,
+      "step": 14212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1823645785864425e-06,
+      "loss": 0.5757,
+      "step": 14213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.18195716244369e-06,
+      "loss": 0.534,
+      "step": 14214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.181550197730639e-06,
+      "loss": 0.5199,
+      "step": 14215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.181143684466003e-06,
+      "loss": 0.5174,
+      "step": 14216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1807376226684695e-06,
+      "loss": 0.5694,
+      "step": 14217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.180332012356713e-06,
+      "loss": 0.5842,
+      "step": 14218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1799268535493813e-06,
+      "loss": 0.6436,
+      "step": 14219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.179522146265104e-06,
+      "loss": 0.5226,
+      "step": 14220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.179117890522489e-06,
+      "loss": 0.5682,
+      "step": 14221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1787140863401236e-06,
+      "loss": 0.4884,
+      "step": 14222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.178310733736575e-06,
+      "loss": 0.5103,
+      "step": 14223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1779078327303893e-06,
+      "loss": 0.5608,
+      "step": 14224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1775053833400906e-06,
+      "loss": 0.55,
+      "step": 14225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1771033855841853e-06,
+      "loss": 0.5414,
+      "step": 14226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1767018394811556e-06,
+      "loss": 0.4783,
+      "step": 14227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.176300745049467e-06,
+      "loss": 0.4971,
+      "step": 14228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.175900102307558e-06,
+      "loss": 0.5737,
+      "step": 14229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.175499911273851e-06,
+      "loss": 0.5229,
+      "step": 14230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1751001719667475e-06,
+      "loss": 0.559,
+      "step": 14231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1747008844046284e-06,
+      "loss": 0.5502,
+      "step": 14232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.174302048605851e-06,
+      "loss": 0.5113,
+      "step": 14233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1739036645887525e-06,
+      "loss": 0.6004,
+      "step": 14234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.173505732371655e-06,
+      "loss": 0.6067,
+      "step": 14235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.17310825197285e-06,
+      "loss": 0.6023,
+      "step": 14236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1727112234106154e-06,
+      "loss": 0.4733,
+      "step": 14237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1723146467032075e-06,
+      "loss": 0.5355,
+      "step": 14238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.17191852186886e-06,
+      "loss": 0.4991,
+      "step": 14239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.171522848925787e-06,
+      "loss": 0.5837,
+      "step": 14240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.17112762789218e-06,
+      "loss": 0.536,
+      "step": 14241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1707328587862124e-06,
+      "loss": 0.6084,
+      "step": 14242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.170338541626036e-06,
+      "loss": 0.5518,
+      "step": 14243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1699446764297807e-06,
+      "loss": 0.5675,
+      "step": 14244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.169551263215555e-06,
+      "loss": 0.5569,
+      "step": 14245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1691583020014498e-06,
+      "loss": 0.5304,
+      "step": 14246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.168765792805532e-06,
+      "loss": 0.5072,
+      "step": 14247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1683737356458503e-06,
+      "loss": 0.5617,
+      "step": 14248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1679821305404295e-06,
+      "loss": 0.5377,
+      "step": 14249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.167590977507278e-06,
+      "loss": 0.4909,
+      "step": 14250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1672002765643775e-06,
+      "loss": 0.5199,
+      "step": 14251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1668100277296945e-06,
+      "loss": 0.5989,
+      "step": 14252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.166420231021173e-06,
+      "loss": 0.5077,
+      "step": 14253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.166030886456735e-06,
+      "loss": 0.4927,
+      "step": 14254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1656419940542806e-06,
+      "loss": 0.525,
+      "step": 14255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1652535538316934e-06,
+      "loss": 0.5389,
+      "step": 14256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1648655658068333e-06,
+      "loss": 0.5763,
+      "step": 14257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.164478029997539e-06,
+      "loss": 0.5222,
+      "step": 14258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.16409094642163e-06,
+      "loss": 0.5523,
+      "step": 14259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.163704315096903e-06,
+      "loss": 0.5406,
+      "step": 14260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1633181360411373e-06,
+      "loss": 0.6113,
+      "step": 14261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1629324092720875e-06,
+      "loss": 0.5774,
+      "step": 14262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.162547134807489e-06,
+      "loss": 0.612,
+      "step": 14263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1621623126650584e-06,
+      "loss": 0.5211,
+      "step": 14264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1617779428624875e-06,
+      "loss": 0.5275,
+      "step": 14265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1613940254174517e-06,
+      "loss": 0.5633,
+      "step": 14266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1610105603476006e-06,
+      "loss": 0.512,
+      "step": 14267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1606275476705674e-06,
+      "loss": 0.5509,
+      "step": 14268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1602449874039647e-06,
+      "loss": 0.5553,
+      "step": 14269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.159862879565379e-06,
+      "loss": 0.5939,
+      "step": 14270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.159481224172381e-06,
+      "loss": 0.5657,
+      "step": 14271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1591000212425197e-06,
+      "loss": 0.5444,
+      "step": 14272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1587192707933217e-06,
+      "loss": 0.6081,
+      "step": 14273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1583389728422934e-06,
+      "loss": 0.5882,
+      "step": 14274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1579591274069222e-06,
+      "loss": 0.5275,
+      "step": 14275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1575797345046723e-06,
+      "loss": 0.5394,
+      "step": 14276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1572007941529894e-06,
+      "loss": 0.4931,
+      "step": 14277
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.156822306369294e-06,
+      "loss": 0.5495,
+      "step": 14278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.156444271170991e-06,
+      "loss": 0.5575,
+      "step": 14279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.156066688575462e-06,
+      "loss": 0.5521,
+      "step": 14280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1556895586000703e-06,
+      "loss": 0.6299,
+      "step": 14281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1553128812621516e-06,
+      "loss": 0.5251,
+      "step": 14282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1549366565790286e-06,
+      "loss": 0.5532,
+      "step": 14283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.154560884567999e-06,
+      "loss": 0.5365,
+      "step": 14284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1541855652463406e-06,
+      "loss": 0.5684,
+      "step": 14285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1538106986313103e-06,
+      "loss": 0.556,
+      "step": 14286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.153436284740145e-06,
+      "loss": 0.5863,
+      "step": 14287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1530623235900597e-06,
+      "loss": 0.4636,
+      "step": 14288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1526888151982484e-06,
+      "loss": 0.5436,
+      "step": 14289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1523157595818856e-06,
+      "loss": 0.5675,
+      "step": 14290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.151943156758124e-06,
+      "loss": 0.5731,
+      "step": 14291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1515710067440968e-06,
+      "loss": 0.5377,
+      "step": 14292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.151199309556913e-06,
+      "loss": 0.5814,
+      "step": 14293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.150828065213665e-06,
+      "loss": 0.5726,
+      "step": 14294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1504572737314226e-06,
+      "loss": 0.5019,
+      "step": 14295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1500869351272338e-06,
+      "loss": 0.5018,
+      "step": 14296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1497170494181266e-06,
+      "loss": 0.5463,
+      "step": 14297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.149347616621107e-06,
+      "loss": 0.601,
+      "step": 14298
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.148978636753164e-06,
+      "loss": 0.5437,
+      "step": 14299
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.148610109831262e-06,
+      "loss": 0.5614,
+      "step": 14300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.148242035872346e-06,
+      "loss": 0.5596,
+      "step": 14301
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1478744148933376e-06,
+      "loss": 0.585,
+      "step": 14302
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1475072469111436e-06,
+      "loss": 0.5112,
+      "step": 14303
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1471405319426442e-06,
+      "loss": 0.6191,
+      "step": 14304
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1467742700047e-06,
+      "loss": 0.5534,
+      "step": 14305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1464084611141525e-06,
+      "loss": 0.5384,
+      "step": 14306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.146043105287822e-06,
+      "loss": 0.5361,
+      "step": 14307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.145678202542507e-06,
+      "loss": 0.5163,
+      "step": 14308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1453137528949855e-06,
+      "loss": 0.5643,
+      "step": 14309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1449497563620147e-06,
+      "loss": 0.5546,
+      "step": 14310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1445862129603326e-06,
+      "loss": 0.5351,
+      "step": 14311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1442231227066516e-06,
+      "loss": 0.5527,
+      "step": 14312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1438604856176683e-06,
+      "loss": 0.5392,
+      "step": 14313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1434983017100554e-06,
+      "loss": 0.5663,
+      "step": 14314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1431365710004693e-06,
+      "loss": 0.5979,
+      "step": 14315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.142775293505538e-06,
+      "loss": 0.511,
+      "step": 14316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.142414469241876e-06,
+      "loss": 0.5037,
+      "step": 14317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.142054098226073e-06,
+      "loss": 0.5638,
+      "step": 14318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1416941804746974e-06,
+      "loss": 0.5059,
+      "step": 14319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1413347160042986e-06,
+      "loss": 0.6022,
+      "step": 14320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1409757048314063e-06,
+      "loss": 0.5907,
+      "step": 14321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.140617146972526e-06,
+      "loss": 0.5198,
+      "step": 14322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.140259042444144e-06,
+      "loss": 0.5587,
+      "step": 14323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1399013912627277e-06,
+      "loss": 0.5277,
+      "step": 14324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1395441934447195e-06,
+      "loss": 0.5924,
+      "step": 14325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1391874490065445e-06,
+      "loss": 0.5767,
+      "step": 14326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.138831157964605e-06,
+      "loss": 0.529,
+      "step": 14327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.138475320335283e-06,
+      "loss": 0.575,
+      "step": 14328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1381199361349406e-06,
+      "loss": 0.4892,
+      "step": 14329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1377650053799185e-06,
+      "loss": 0.5304,
+      "step": 14330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1374105280865343e-06,
+      "loss": 0.5731,
+      "step": 14331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.137056504271089e-06,
+      "loss": 0.521,
+      "step": 14332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.136702933949859e-06,
+      "loss": 0.5276,
+      "step": 14333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.136349817139103e-06,
+      "loss": 0.4895,
+      "step": 14334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.135997153855054e-06,
+      "loss": 0.5471,
+      "step": 14335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1356449441139305e-06,
+      "loss": 0.5515,
+      "step": 14336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1352931879319265e-06,
+      "loss": 0.4973,
+      "step": 14337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1349418853252127e-06,
+      "loss": 0.5427,
+      "step": 14338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1345910363099454e-06,
+      "loss": 0.5099,
+      "step": 14339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1342406409022543e-06,
+      "loss": 0.5615,
+      "step": 14340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.133890699118252e-06,
+      "loss": 0.5884,
+      "step": 14341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1335412109740275e-06,
+      "loss": 0.5538,
+      "step": 14342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.133192176485651e-06,
+      "loss": 0.5244,
+      "step": 14343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1328435956691696e-06,
+      "loss": 0.5993,
+      "step": 14344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1324954685406133e-06,
+      "loss": 0.501,
+      "step": 14345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1321477951159865e-06,
+      "loss": 0.5633,
+      "step": 14346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.131800575411276e-06,
+      "loss": 0.5967,
+      "step": 14347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1314538094424467e-06,
+      "loss": 0.5136,
+      "step": 14348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1311074972254444e-06,
+      "loss": 0.5634,
+      "step": 14349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1307616387761888e-06,
+      "loss": 0.5148,
+      "step": 14350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.130416234110585e-06,
+      "loss": 0.5485,
+      "step": 14351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.130071283244515e-06,
+      "loss": 0.4931,
+      "step": 14352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1297267861938377e-06,
+      "loss": 0.5379,
+      "step": 14353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.129382742974393e-06,
+      "loss": 0.5866,
+      "step": 14354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1290391536020006e-06,
+      "loss": 0.5391,
+      "step": 14355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.128696018092459e-06,
+      "loss": 0.5471,
+      "step": 14356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1283533364615452e-06,
+      "loss": 0.5332,
+      "step": 14357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1280111087250156e-06,
+      "loss": 0.5566,
+      "step": 14358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1276693348986045e-06,
+      "loss": 0.6145,
+      "step": 14359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1273280149980294e-06,
+      "loss": 0.5148,
+      "step": 14360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1269871490389805e-06,
+      "loss": 0.5554,
+      "step": 14361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1266467370371328e-06,
+      "loss": 0.5253,
+      "step": 14362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1263067790081374e-06,
+      "loss": 0.5504,
+      "step": 14363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1259672749676264e-06,
+      "loss": 0.4843,
+      "step": 14364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.125628224931209e-06,
+      "loss": 0.5376,
+      "step": 14365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1252896289144757e-06,
+      "loss": 0.5696,
+      "step": 14366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.124951486932995e-06,
+      "loss": 0.5495,
+      "step": 14367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1246137990023134e-06,
+      "loss": 0.5422,
+      "step": 14368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1242765651379573e-06,
+      "loss": 0.5576,
+      "step": 14369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1239397853554336e-06,
+      "loss": 0.5025,
+      "step": 14370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.123603459670228e-06,
+      "loss": 0.5123,
+      "step": 14371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1232675880978037e-06,
+      "loss": 0.4775,
+      "step": 14372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1229321706536033e-06,
+      "loss": 0.5019,
+      "step": 14373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1225972073530513e-06,
+      "loss": 0.5891,
+      "step": 14374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.122262698211547e-06,
+      "loss": 0.5164,
+      "step": 14375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.121928643244471e-06,
+      "loss": 0.504,
+      "step": 14376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.121595042467185e-06,
+      "loss": 0.4936,
+      "step": 14377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.121261895895026e-06,
+      "loss": 0.5824,
+      "step": 14378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.120929203543313e-06,
+      "loss": 0.5781,
+      "step": 14379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.120596965427343e-06,
+      "loss": 0.601,
+      "step": 14380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.120265181562391e-06,
+      "loss": 0.5425,
+      "step": 14381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1199338519637125e-06,
+      "loss": 0.6066,
+      "step": 14382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1196029766465443e-06,
+      "loss": 0.555,
+      "step": 14383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1192725556260974e-06,
+      "loss": 0.5795,
+      "step": 14384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1189425889175654e-06,
+      "loss": 0.5174,
+      "step": 14385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1186130765361193e-06,
+      "loss": 0.5338,
+      "step": 14386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.11828401849691e-06,
+      "loss": 0.5637,
+      "step": 14387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1179554148150686e-06,
+      "loss": 0.4944,
+      "step": 14388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.117627265505703e-06,
+      "loss": 0.5545,
+      "step": 14389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.117299570583903e-06,
+      "loss": 0.5423,
+      "step": 14390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.116972330064734e-06,
+      "loss": 0.5704,
+      "step": 14391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.116645543963242e-06,
+      "loss": 0.5749,
+      "step": 14392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1163192122944543e-06,
+      "loss": 0.5245,
+      "step": 14393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1159933350733766e-06,
+      "loss": 0.5492,
+      "step": 14394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1156679123149882e-06,
+      "loss": 0.5419,
+      "step": 14395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.115342944034257e-06,
+      "loss": 0.5148,
+      "step": 14396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1150184302461196e-06,
+      "loss": 0.5245,
+      "step": 14397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1146943709655023e-06,
+      "loss": 0.5418,
+      "step": 14398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.114370766207302e-06,
+      "loss": 0.5045,
+      "step": 14399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.114047615986398e-06,
+      "loss": 0.5227,
+      "step": 14400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1137249203176504e-06,
+      "loss": 0.4875,
+      "step": 14401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1134026792158948e-06,
+      "loss": 0.5463,
+      "step": 14402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1130808926959492e-06,
+      "loss": 0.5087,
+      "step": 14403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1127595607726073e-06,
+      "loss": 0.4863,
+      "step": 14404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.112438683460646e-06,
+      "loss": 0.5305,
+      "step": 14405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1121182607748173e-06,
+      "loss": 0.5617,
+      "step": 14406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1117982927298553e-06,
+      "loss": 0.5185,
+      "step": 14407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1114787793404715e-06,
+      "loss": 0.5527,
+      "step": 14408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1111597206213565e-06,
+      "loss": 0.5859,
+      "step": 14409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1108411165871813e-06,
+      "loss": 0.5203,
+      "step": 14410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1105229672525953e-06,
+      "loss": 0.5418,
+      "step": 14411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1102052726322264e-06,
+      "loss": 0.5138,
+      "step": 14412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1098880327406834e-06,
+      "loss": 0.5386,
+      "step": 14413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.10957124759255e-06,
+      "loss": 0.5356,
+      "step": 14414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.109254917202395e-06,
+      "loss": 0.5062,
+      "step": 14415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.10893904158476e-06,
+      "loss": 0.5664,
+      "step": 14416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1086236207541725e-06,
+      "loss": 0.5234,
+      "step": 14417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1083086547251323e-06,
+      "loss": 0.5168,
+      "step": 14418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.107994143512122e-06,
+      "loss": 0.5936,
+      "step": 14419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1076800871296045e-06,
+      "loss": 0.5677,
+      "step": 14420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1073664855920187e-06,
+      "loss": 0.4964,
+      "step": 14421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.107053338913783e-06,
+      "loss": 0.5319,
+      "step": 14422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1067406471092967e-06,
+      "loss": 0.4825,
+      "step": 14423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.106428410192937e-06,
+      "loss": 0.526,
+      "step": 14424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1061166281790613e-06,
+      "loss": 0.5277,
+      "step": 14425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1058053010820045e-06,
+      "loss": 0.5562,
+      "step": 14426
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1054944289160806e-06,
+      "loss": 0.4887,
+      "step": 14427
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.105184011695586e-06,
+      "loss": 0.563,
+      "step": 14428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.10487404943479e-06,
+      "loss": 0.5858,
+      "step": 14429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1045645421479466e-06,
+      "loss": 0.5369,
+      "step": 14430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1042554898492874e-06,
+      "loss": 0.5132,
+      "step": 14431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1039468925530208e-06,
+      "loss": 0.5211,
+      "step": 14432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.103638750273337e-06,
+      "loss": 0.5818,
+      "step": 14433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1033310630244034e-06,
+      "loss": 0.5487,
+      "step": 14434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.103023830820369e-06,
+      "loss": 0.5247,
+      "step": 14435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1027170536753586e-06,
+      "loss": 0.5686,
+      "step": 14436
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1024107316034784e-06,
+      "loss": 0.4776,
+      "step": 14437
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1021048646188114e-06,
+      "loss": 0.5929,
+      "step": 14438
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.101799452735425e-06,
+      "loss": 0.5819,
+      "step": 14439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.101494495967358e-06,
+      "loss": 0.5588,
+      "step": 14440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.101189994328633e-06,
+      "loss": 0.5436,
+      "step": 14441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1008859478332534e-06,
+      "loss": 0.4609,
+      "step": 14442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.100582356495197e-06,
+      "loss": 0.557,
+      "step": 14443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.100279220328423e-06,
+      "loss": 0.549,
+      "step": 14444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.099976539346869e-06,
+      "loss": 0.5943,
+      "step": 14445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0996743135644537e-06,
+      "loss": 0.5309,
+      "step": 14446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0993725429950717e-06,
+      "loss": 0.5636,
+      "step": 14447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0990712276525994e-06,
+      "loss": 0.56,
+      "step": 14448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.098770367550891e-06,
+      "loss": 0.5096,
+      "step": 14449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0984699627037798e-06,
+      "loss": 0.5313,
+      "step": 14450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0981700131250784e-06,
+      "loss": 0.6171,
+      "step": 14451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0978705188285776e-06,
+      "loss": 0.5303,
+      "step": 14452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.097571479828048e-06,
+      "loss": 0.5283,
+      "step": 14453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0972728961372415e-06,
+      "loss": 0.513,
+      "step": 14454
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.096974767769884e-06,
+      "loss": 0.5057,
+      "step": 14455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0966770947396845e-06,
+      "loss": 0.578,
+      "step": 14456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.096379877060331e-06,
+      "loss": 0.4968,
+      "step": 14457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0960831147454878e-06,
+      "loss": 0.5124,
+      "step": 14458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0957868078087996e-06,
+      "loss": 0.5564,
+      "step": 14459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0954909562638926e-06,
+      "loss": 0.5534,
+      "step": 14460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.095195560124368e-06,
+      "loss": 0.5414,
+      "step": 14461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0949006194038093e-06,
+      "loss": 0.5071,
+      "step": 14462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0946061341157765e-06,
+      "loss": 0.5026,
+      "step": 14463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0943121042738113e-06,
+      "loss": 0.5299,
+      "step": 14464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.094018529891432e-06,
+      "loss": 0.5037,
+      "step": 14465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.093725410982138e-06,
+      "loss": 0.4484,
+      "step": 14466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0934327475594064e-06,
+      "loss": 0.5834,
+      "step": 14467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0931405396366923e-06,
+      "loss": 0.47,
+      "step": 14468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.092848787227434e-06,
+      "loss": 0.5539,
+      "step": 14469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.092557490345043e-06,
+      "loss": 0.5311,
+      "step": 14470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0922666490029167e-06,
+      "loss": 0.5223,
+      "step": 14471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.091976263214424e-06,
+      "loss": 0.5469,
+      "step": 14472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0916863329929193e-06,
+      "loss": 0.5654,
+      "step": 14473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0913968583517332e-06,
+      "loss": 0.5454,
+      "step": 14474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.091107839304174e-06,
+      "loss": 0.5321,
+      "step": 14475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.090819275863533e-06,
+      "loss": 0.5914,
+      "step": 14476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0905311680430777e-06,
+      "loss": 0.5118,
+      "step": 14477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.090243515856054e-06,
+      "loss": 0.5321,
+      "step": 14478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0899563193156883e-06,
+      "loss": 0.5776,
+      "step": 14479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0896695784351865e-06,
+      "loss": 0.5683,
+      "step": 14480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0893832932277326e-06,
+      "loss": 0.5648,
+      "step": 14481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.089097463706489e-06,
+      "loss": 0.5252,
+      "step": 14482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0888120898846e-06,
+      "loss": 0.5501,
+      "step": 14483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0885271717751846e-06,
+      "loss": 0.5012,
+      "step": 14484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.088242709391345e-06,
+      "loss": 0.496,
+      "step": 14485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0879587027461603e-06,
+      "loss": 0.4734,
+      "step": 14486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0876751518526874e-06,
+      "loss": 0.5099,
+      "step": 14487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0873920567239672e-06,
+      "loss": 0.5585,
+      "step": 14488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.087109417373012e-06,
+      "loss": 0.5204,
+      "step": 14489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.086827233812822e-06,
+      "loss": 0.5215,
+      "step": 14490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.086545506056368e-06,
+      "loss": 0.5138,
+      "step": 14491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0862642341166063e-06,
+      "loss": 0.4938,
+      "step": 14492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0859834180064677e-06,
+      "loss": 0.5573,
+      "step": 14493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.085703057738866e-06,
+      "loss": 0.448,
+      "step": 14494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.08542315332669e-06,
+      "loss": 0.5634,
+      "step": 14495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.085143704782813e-06,
+      "loss": 0.4946,
+      "step": 14496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0848647121200797e-06,
+      "loss": 0.5363,
+      "step": 14497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.08458617535132e-06,
+      "loss": 0.5271,
+      "step": 14498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0843080944893415e-06,
+      "loss": 0.4762,
+      "step": 14499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0840304695469306e-06,
+      "loss": 0.5329,
+      "step": 14500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0837533005368493e-06,
+      "loss": 0.4584,
+      "step": 14501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0834765874718456e-06,
+      "loss": 0.4772,
+      "step": 14502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.083200330364641e-06,
+      "loss": 0.5215,
+      "step": 14503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.082924529227937e-06,
+      "loss": 0.5012,
+      "step": 14504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0826491840744147e-06,
+      "loss": 0.4968,
+      "step": 14505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.082374294916736e-06,
+      "loss": 0.5206,
+      "step": 14506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0820998617675395e-06,
+      "loss": 0.4936,
+      "step": 14507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0818258846394425e-06,
+      "loss": 0.4874,
+      "step": 14508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.081552363545043e-06,
+      "loss": 0.5611,
+      "step": 14509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0812792984969178e-06,
+      "loss": 0.5094,
+      "step": 14510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.081006689507623e-06,
+      "loss": 0.5323,
+      "step": 14511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0807345365896914e-06,
+      "loss": 0.5583,
+      "step": 14512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0804628397556383e-06,
+      "loss": 0.5103,
+      "step": 14513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0801915990179534e-06,
+      "loss": 0.564,
+      "step": 14514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.079920814389112e-06,
+      "loss": 0.5651,
+      "step": 14515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0796504858815614e-06,
+      "loss": 0.4791,
+      "step": 14516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.079380613507732e-06,
+      "loss": 0.5183,
+      "step": 14517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0791111972800337e-06,
+      "loss": 0.5079,
+      "step": 14518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0788422372108533e-06,
+      "loss": 0.5176,
+      "step": 14519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.078573733312558e-06,
+      "loss": 0.5472,
+      "step": 14520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.078305685597492e-06,
+      "loss": 0.5861,
+      "step": 14521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.078038094077981e-06,
+      "loss": 0.5085,
+      "step": 14522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.07777095876633e-06,
+      "loss": 0.4855,
+      "step": 14523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0775042796748193e-06,
+      "loss": 0.4664,
+      "step": 14524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0772380568157115e-06,
+      "loss": 0.5189,
+      "step": 14525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0769722902012494e-06,
+      "loss": 0.5626,
+      "step": 14526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0767069798436503e-06,
+      "loss": 0.5105,
+      "step": 14527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0764421257551137e-06,
+      "loss": 0.4868,
+      "step": 14528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.076177727947819e-06,
+      "loss": 0.5212,
+      "step": 14529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.075913786433922e-06,
+      "loss": 0.5179,
+      "step": 14530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.075650301225558e-06,
+      "loss": 0.5357,
+      "step": 14531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0753872723348436e-06,
+      "loss": 0.4704,
+      "step": 14532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0751246997738707e-06,
+      "loss": 0.4859,
+      "step": 14533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.074862583554713e-06,
+      "loss": 0.525,
+      "step": 14534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.074600923689424e-06,
+      "loss": 0.5337,
+      "step": 14535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0743397201900317e-06,
+      "loss": 0.487,
+      "step": 14536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0740789730685496e-06,
+      "loss": 0.5037,
+      "step": 14537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0738186823369647e-06,
+      "loss": 0.5046,
+      "step": 14538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0735588480072457e-06,
+      "loss": 0.5412,
+      "step": 14539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.073299470091339e-06,
+      "loss": 0.5469,
+      "step": 14540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.073040548601171e-06,
+      "loss": 0.5651,
+      "step": 14541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0727820835486467e-06,
+      "loss": 0.5414,
+      "step": 14542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0725240749456512e-06,
+      "loss": 0.5009,
+      "step": 14543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0722665228040464e-06,
+      "loss": 0.5579,
+      "step": 14544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.072009427135675e-06,
+      "loss": 0.5369,
+      "step": 14545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0717527879523583e-06,
+      "loss": 0.5403,
+      "step": 14546
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.071496605265896e-06,
+      "loss": 0.502,
+      "step": 14547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.071240879088068e-06,
+      "loss": 0.4896,
+      "step": 14548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0709856094306315e-06,
+      "loss": 0.5151,
+      "step": 14549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0707307963053247e-06,
+      "loss": 0.54,
+      "step": 14550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.070476439723863e-06,
+      "loss": 0.5297,
+      "step": 14551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0702225396979427e-06,
+      "loss": 0.5583,
+      "step": 14552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.069969096239236e-06,
+      "loss": 0.4545,
+      "step": 14553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.069716109359398e-06,
+      "loss": 0.438,
+      "step": 14554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.06946357907006e-06,
+      "loss": 0.5577,
+      "step": 14555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0692115053828345e-06,
+      "loss": 0.5257,
+      "step": 14556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.06895988830931e-06,
+      "loss": 0.5174,
+      "step": 14557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0687087278610566e-06,
+      "loss": 0.5355,
+      "step": 14558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0684580240496225e-06,
+      "loss": 0.4984,
+      "step": 14559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.068207776886535e-06,
+      "loss": 0.5262,
+      "step": 14560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0679579863833e-06,
+      "loss": 0.5593,
+      "step": 14561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0677086525514046e-06,
+      "loss": 0.5549,
+      "step": 14562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0674597754023097e-06,
+      "loss": 0.5093,
+      "step": 14563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.067211354947463e-06,
+      "loss": 0.5851,
+      "step": 14564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0669633911982825e-06,
+      "loss": 0.5016,
+      "step": 14565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0667158841661706e-06,
+      "loss": 0.4599,
+      "step": 14566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0664688338625095e-06,
+      "loss": 0.4883,
+      "step": 14567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0662222402986566e-06,
+      "loss": 0.4911,
+      "step": 14568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0659761034859505e-06,
+      "loss": 0.5431,
+      "step": 14569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.065730423435709e-06,
+      "loss": 0.4966,
+      "step": 14570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0654852001592287e-06,
+      "loss": 0.5264,
+      "step": 14571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0652404336677836e-06,
+      "loss": 0.5479,
+      "step": 14572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.064996123972628e-06,
+      "loss": 0.543,
+      "step": 14573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0647522710849967e-06,
+      "loss": 0.536,
+      "step": 14574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0645088750161004e-06,
+      "loss": 0.534,
+      "step": 14575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.064265935777131e-06,
+      "loss": 0.5508,
+      "step": 14576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0640234533792593e-06,
+      "loss": 0.492,
+      "step": 14577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0637814278336347e-06,
+      "loss": 0.5109,
+      "step": 14578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0635398591513835e-06,
+      "loss": 0.5452,
+      "step": 14579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.063298747343614e-06,
+      "loss": 0.5545,
+      "step": 14580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.063058092421413e-06,
+      "loss": 0.4978,
+      "step": 14581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0628178943958455e-06,
+      "loss": 0.5072,
+      "step": 14582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0625781532779558e-06,
+      "loss": 0.4632,
+      "step": 14583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.062338869078766e-06,
+      "loss": 0.5831,
+      "step": 14584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0621000418092796e-06,
+      "loss": 0.5758,
+      "step": 14585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0618616714804774e-06,
+      "loss": 0.4976,
+      "step": 14586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0616237581033197e-06,
+      "loss": 0.503,
+      "step": 14587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.061386301688745e-06,
+      "loss": 0.4499,
+      "step": 14588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0611493022476715e-06,
+      "loss": 0.5158,
+      "step": 14589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0609127597909973e-06,
+      "loss": 0.554,
+      "step": 14590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.060676674329598e-06,
+      "loss": 0.5099,
+      "step": 14591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0604410458743277e-06,
+      "loss": 0.5473,
+      "step": 14592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.060205874436022e-06,
+      "loss": 0.4935,
+      "step": 14593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0599711600254937e-06,
+      "loss": 0.4422,
+      "step": 14594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0597369026535344e-06,
+      "loss": 0.5036,
+      "step": 14595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0595031023309154e-06,
+      "loss": 0.5158,
+      "step": 14596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0592697590683865e-06,
+      "loss": 0.5159,
+      "step": 14597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0590368728766774e-06,
+      "loss": 0.4984,
+      "step": 14598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0588044437664956e-06,
+      "loss": 0.4701,
+      "step": 14599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.058572471748528e-06,
+      "loss": 0.5034,
+      "step": 14600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0583409568334412e-06,
+      "loss": 0.531,
+      "step": 14601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0581098990318796e-06,
+      "loss": 0.5704,
+      "step": 14602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.057879298354467e-06,
+      "loss": 0.4868,
+      "step": 14603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0576491548118063e-06,
+      "loss": 0.5314,
+      "step": 14604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.05741946841448e-06,
+      "loss": 0.5725,
+      "step": 14605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0571902391730495e-06,
+      "loss": 0.5272,
+      "step": 14606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.056961467098052e-06,
+      "loss": 0.5345,
+      "step": 14607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.056733152200009e-06,
+      "loss": 0.4704,
+      "step": 14608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0565052944894182e-06,
+      "loss": 0.5351,
+      "step": 14609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.056277893976755e-06,
+      "loss": 0.527,
+      "step": 14610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.056050950672476e-06,
+      "loss": 0.5297,
+      "step": 14611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0558244645870164e-06,
+      "loss": 0.4842,
+      "step": 14612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0555984357307893e-06,
+      "loss": 0.5527,
+      "step": 14613
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.055372864114188e-06,
+      "loss": 0.5116,
+      "step": 14614
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0551477497475827e-06,
+      "loss": 0.5159,
+      "step": 14615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054923092641326e-06,
+      "loss": 0.5394,
+      "step": 14616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054698892805746e-06,
+      "loss": 0.5333,
+      "step": 14617
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054475150251152e-06,
+      "loss": 0.5274,
+      "step": 14618
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054251864987832e-06,
+      "loss": 0.561,
+      "step": 14619
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054029037026052e-06,
+      "loss": 0.5265,
+      "step": 14620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.053806666376057e-06,
+      "loss": 0.4869,
+      "step": 14621
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0535847530480735e-06,
+      "loss": 0.4862,
+      "step": 14622
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0533632970523024e-06,
+      "loss": 0.4969,
+      "step": 14623
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.053142298398928e-06,
+      "loss": 0.4831,
+      "step": 14624
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.052921757098111e-06,
+      "loss": 0.5618,
+      "step": 14625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0527016731599925e-06,
+      "loss": 0.5668,
+      "step": 14626
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0524820465946905e-06,
+      "loss": 0.5364,
+      "step": 14627
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0522628774123044e-06,
+      "loss": 0.5198,
+      "step": 14628
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.052044165622912e-06,
+      "loss": 0.4956,
+      "step": 14629
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0518259112365683e-06,
+      "loss": 0.4912,
+      "step": 14630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0516081142633082e-06,
+      "loss": 0.4826,
+      "step": 14631
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.051390774713148e-06,
+      "loss": 0.5743,
+      "step": 14632
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.05117389259608e-06,
+      "loss": 0.5177,
+      "step": 14633
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0509574679220757e-06,
+      "loss": 0.4531,
+      "step": 14634
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0507415007010857e-06,
+      "loss": 0.4752,
+      "step": 14635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.050525990943042e-06,
+      "loss": 0.4736,
+      "step": 14636
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0503109386578525e-06,
+      "loss": 0.4743,
+      "step": 14637
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.050096343855406e-06,
+      "loss": 0.5529,
+      "step": 14638
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0498822065455685e-06,
+      "loss": 0.5429,
+      "step": 14639
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.049668526738186e-06,
+      "loss": 0.5073,
+      "step": 14640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0494553044430835e-06,
+      "loss": 0.578,
+      "step": 14641
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0492425396700645e-06,
+      "loss": 0.5757,
+      "step": 14642
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0490302324289127e-06,
+      "loss": 0.5331,
+      "step": 14643
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0488183827293896e-06,
+      "loss": 0.5427,
+      "step": 14644
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.048606990581237e-06,
+      "loss": 0.5347,
+      "step": 14645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.048396055994173e-06,
+      "loss": 0.526,
+      "step": 14646
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.048185578977897e-06,
+      "loss": 0.506,
+      "step": 14647
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.047975559542086e-06,
+      "loss": 0.5064,
+      "step": 14648
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.047765997696398e-06,
+      "loss": 0.5503,
+      "step": 14649
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0475568934504664e-06,
+      "loss": 0.5424,
+      "step": 14650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0473482468139086e-06,
+      "loss": 0.5227,
+      "step": 14651
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0471400577963166e-06,
+      "loss": 0.5019,
+      "step": 14652
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.046932326407262e-06,
+      "loss": 0.5106,
+      "step": 14653
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.046725052656297e-06,
+      "loss": 0.5201,
+      "step": 14654
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0465182365529527e-06,
+      "loss": 0.5329,
+      "step": 14655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.046311878106736e-06,
+      "loss": 0.5478,
+      "step": 14656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0461059773271387e-06,
+      "loss": 0.513,
+      "step": 14657
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.045900534223625e-06,
+      "loss": 0.5229,
+      "step": 14658
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0456955488056432e-06,
+      "loss": 0.5181,
+      "step": 14659
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0454910210826163e-06,
+      "loss": 0.5321,
+      "step": 14660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.045286951063951e-06,
+      "loss": 0.5236,
+      "step": 14661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.045083338759029e-06,
+      "loss": 0.4969,
+      "step": 14662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0448801841772105e-06,
+      "loss": 0.5452,
+      "step": 14663
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0446774873278397e-06,
+      "loss": 0.5327,
+      "step": 14664
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0444752482202346e-06,
+      "loss": 0.4564,
+      "step": 14665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0442734668636955e-06,
+      "loss": 0.5275,
+      "step": 14666
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0440721432674993e-06,
+      "loss": 0.523,
+      "step": 14667
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.043871277440902e-06,
+      "loss": 0.5433,
+      "step": 14668
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0436708693931407e-06,
+      "loss": 0.4994,
+      "step": 14669
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0434709191334294e-06,
+      "loss": 0.4961,
+      "step": 14670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.043271426670961e-06,
+      "loss": 0.485,
+      "step": 14671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.043072392014909e-06,
+      "loss": 0.5179,
+      "step": 14672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0428738151744256e-06,
+      "loss": 0.5262,
+      "step": 14673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0426756961586402e-06,
+      "loss": 0.4669,
+      "step": 14674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.042478034976663e-06,
+      "loss": 0.5548,
+      "step": 14675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.042280831637581e-06,
+      "loss": 0.5311,
+      "step": 14676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0420840861504635e-06,
+      "loss": 0.5596,
+      "step": 14677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0418877985243548e-06,
+      "loss": 0.5037,
+      "step": 14678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0416919687682813e-06,
+      "loss": 0.4989,
+      "step": 14679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.041496596891247e-06,
+      "loss": 0.4841,
+      "step": 14680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.041301682902236e-06,
+      "loss": 0.4723,
+      "step": 14681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0411072268102086e-06,
+      "loss": 0.4614,
+      "step": 14682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0409132286241065e-06,
+      "loss": 0.5757,
+      "step": 14683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0407196883528494e-06,
+      "loss": 0.5407,
+      "step": 14684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0405266060053375e-06,
+      "loss": 0.479,
+      "step": 14685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.040333981590447e-06,
+      "loss": 0.5008,
+      "step": 14686
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0401418151170353e-06,
+      "loss": 0.4837,
+      "step": 14687
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0399501065939387e-06,
+      "loss": 0.5305,
+      "step": 14688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.039758856029971e-06,
+      "loss": 0.468,
+      "step": 14689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0395680634339267e-06,
+      "loss": 0.4524,
+      "step": 14690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0393777288145768e-06,
+      "loss": 0.5571,
+      "step": 14691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.039187852180674e-06,
+      "loss": 0.5217,
+      "step": 14692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.03899843354095e-06,
+      "loss": 0.5056,
+      "step": 14693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0388094729041123e-06,
+      "loss": 0.4837,
+      "step": 14694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0386209702788486e-06,
+      "loss": 0.5341,
+      "step": 14695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0384329256738293e-06,
+      "loss": 0.4311,
+      "step": 14696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0382453390976964e-06,
+      "loss": 0.5341,
+      "step": 14697
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.038058210559079e-06,
+      "loss": 0.5118,
+      "step": 14698
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0378715400665786e-06,
+      "loss": 0.4684,
+      "step": 14699
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0376853276287803e-06,
+      "loss": 0.5273,
+      "step": 14700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0374995732542437e-06,
+      "loss": 0.4868,
+      "step": 14701
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.037314276951512e-06,
+      "loss": 0.5087,
+      "step": 14702
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0371294387291033e-06,
+      "loss": 0.5086,
+      "step": 14703
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0369450585955176e-06,
+      "loss": 0.4884,
+      "step": 14704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0367611365592312e-06,
+      "loss": 0.5055,
+      "step": 14705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0365776726287026e-06,
+      "loss": 0.5087,
+      "step": 14706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036394666812366e-06,
+      "loss": 0.5667,
+      "step": 14707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036212119118636e-06,
+      "loss": 0.5838,
+      "step": 14708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036030029555907e-06,
+      "loss": 0.5154,
+      "step": 14709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0358483981325506e-06,
+      "loss": 0.5327,
+      "step": 14710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.035667224856919e-06,
+      "loss": 0.4486,
+      "step": 14711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0354865097373412e-06,
+      "loss": 0.4591,
+      "step": 14712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0353062527821265e-06,
+      "loss": 0.5362,
+      "step": 14713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0351264539995653e-06,
+      "loss": 0.4895,
+      "step": 14714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.034947113397923e-06,
+      "loss": 0.4912,
+      "step": 14715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0347682309854446e-06,
+      "loss": 0.4424,
+      "step": 14716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0345898067703553e-06,
+      "loss": 0.4831,
+      "step": 14717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0344118407608606e-06,
+      "loss": 0.5018,
+      "step": 14718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0342343329651428e-06,
+      "loss": 0.5145,
+      "step": 14719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.034057283391363e-06,
+      "loss": 0.5166,
+      "step": 14720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.033880692047663e-06,
+      "loss": 0.5059,
+      "step": 14721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0337045589421614e-06,
+      "loss": 0.4139,
+      "step": 14722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.033528884082956e-06,
+      "loss": 0.5246,
+      "step": 14723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.033353667478125e-06,
+      "loss": 0.5291,
+      "step": 14724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0331789091357247e-06,
+      "loss": 0.5355,
+      "step": 14725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0330046090637913e-06,
+      "loss": 0.4888,
+      "step": 14726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0328307672703388e-06,
+      "loss": 0.5319,
+      "step": 14727
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0326573837633594e-06,
+      "loss": 0.5213,
+      "step": 14728
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0324844585508266e-06,
+      "loss": 0.5348,
+      "step": 14729
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0323119916406898e-06,
+      "loss": 0.475,
+      "step": 14730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.03213998304088e-06,
+      "loss": 0.4803,
+      "step": 14731
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.031968432759305e-06,
+      "loss": 0.4905,
+      "step": 14732
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0317973408038553e-06,
+      "loss": 0.5519,
+      "step": 14733
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0316267071823947e-06,
+      "loss": 0.4817,
+      "step": 14734
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.03145653190277e-06,
+      "loss": 0.5092,
+      "step": 14735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0312868149728052e-06,
+      "loss": 0.4933,
+      "step": 14736
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.031117556400306e-06,
+      "loss": 0.4608,
+      "step": 14737
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.030948756193052e-06,
+      "loss": 0.5046,
+      "step": 14738
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.030780414358806e-06,
+      "loss": 0.5081,
+      "step": 14739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0306125309053084e-06,
+      "loss": 0.5542,
+      "step": 14740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0304451058402775e-06,
+      "loss": 0.5079,
+      "step": 14741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0302781391714122e-06,
+      "loss": 0.48,
+      "step": 14742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0301116309063895e-06,
+      "loss": 0.5187,
+      "step": 14743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0299455810528658e-06,
+      "loss": 0.5516,
+      "step": 14744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.029779989618475e-06,
+      "loss": 0.483,
+      "step": 14745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.029614856610831e-06,
+      "loss": 0.4945,
+      "step": 14746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.029450182037528e-06,
+      "loss": 0.4784,
+      "step": 14747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.029285965906136e-06,
+      "loss": 0.5161,
+      "step": 14748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.029122208224207e-06,
+      "loss": 0.4852,
+      "step": 14749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0289589089992693e-06,
+      "loss": 0.4805,
+      "step": 14750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.028796068238831e-06,
+      "loss": 0.4834,
+      "step": 14751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.028633685950381e-06,
+      "loss": 0.4553,
+      "step": 14752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.028471762141384e-06,
+      "loss": 0.4961,
+      "step": 14753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0283102968192873e-06,
+      "loss": 0.4824,
+      "step": 14754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0281492899915123e-06,
+      "loss": 0.4981,
+      "step": 14755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0279887416654643e-06,
+      "loss": 0.5036,
+      "step": 14756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0278286518485247e-06,
+      "loss": 0.4844,
+      "step": 14757
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.027669020548053e-06,
+      "loss": 0.5196,
+      "step": 14758
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0275098477713907e-06,
+      "loss": 0.5097,
+      "step": 14759
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0273511335258558e-06,
+      "loss": 0.4593,
+      "step": 14760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0271928778187456e-06,
+      "loss": 0.5102,
+      "step": 14761
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0270350806573382e-06,
+      "loss": 0.5517,
+      "step": 14762
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0268777420488865e-06,
+      "loss": 0.5689,
+      "step": 14763
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0267208620006265e-06,
+      "loss": 0.4816,
+      "step": 14764
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0265644405197716e-06,
+      "loss": 0.5191,
+      "step": 14765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.026408477613513e-06,
+      "loss": 0.5278,
+      "step": 14766
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0262529732890224e-06,
+      "loss": 0.553,
+      "step": 14767
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.026097927553451e-06,
+      "loss": 0.531,
+      "step": 14768
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0259433404139264e-06,
+      "loss": 0.5024,
+      "step": 14769
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.025789211877555e-06,
+      "loss": 0.488,
+      "step": 14770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0256355419514266e-06,
+      "loss": 0.4985,
+      "step": 14771
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0254823306426053e-06,
+      "loss": 0.5211,
+      "step": 14772
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0253295779581352e-06,
+      "loss": 0.5253,
+      "step": 14773
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.025177283905041e-06,
+      "loss": 0.4971,
+      "step": 14774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0250254484903246e-06,
+      "loss": 0.5052,
+      "step": 14775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0248740717209673e-06,
+      "loss": 0.5041,
+      "step": 14776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0247231536039293e-06,
+      "loss": 0.4634,
+      "step": 14777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.02457269414615e-06,
+      "loss": 0.4884,
+      "step": 14778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0244226933545465e-06,
+      "loss": 0.5072,
+      "step": 14779
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.024273151236017e-06,
+      "loss": 0.5104,
+      "step": 14780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0241240677974375e-06,
+      "loss": 0.5036,
+      "step": 14781
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.023975443045662e-06,
+      "loss": 0.5771,
+      "step": 14782
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0238272769875246e-06,
+      "loss": 0.4872,
+      "step": 14783
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0236795696298373e-06,
+      "loss": 0.482,
+      "step": 14784
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0235323209793926e-06,
+      "loss": 0.5195,
+      "step": 14785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.02338553104296e-06,
+      "loss": 0.4822,
+      "step": 14786
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0232391998272894e-06,
+      "loss": 0.4969,
+      "step": 14787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0230933273391086e-06,
+      "loss": 0.4917,
+      "step": 14788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.022947913585125e-06,
+      "loss": 0.5153,
+      "step": 14789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0228029585720258e-06,
+      "loss": 0.5286,
+      "step": 14790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.022658462306474e-06,
+      "loss": 0.4839,
+      "step": 14791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.022514424795115e-06,
+      "loss": 0.4865,
+      "step": 14792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0223708460445703e-06,
+      "loss": 0.5067,
+      "step": 14793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0222277260614432e-06,
+      "loss": 0.5105,
+      "step": 14794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.022085064852313e-06,
+      "loss": 0.5695,
+      "step": 14795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.021942862423739e-06,
+      "loss": 0.4866,
+      "step": 14796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0218011187822606e-06,
+      "loss": 0.5052,
+      "step": 14797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0216598339343954e-06,
+      "loss": 0.4462,
+      "step": 14798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0215190078866377e-06,
+      "loss": 0.5214,
+      "step": 14799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.021378640645465e-06,
+      "loss": 0.4867,
+      "step": 14800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.02123873221733e-06,
+      "loss": 0.5404,
+      "step": 14801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0210992826086653e-06,
+      "loss": 0.4339,
+      "step": 14802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0209602918258832e-06,
+      "loss": 0.4554,
+      "step": 14803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0208217598753756e-06,
+      "loss": 0.472,
+      "step": 14804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0206836867635107e-06,
+      "loss": 0.5334,
+      "step": 14805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0205460724966364e-06,
+      "loss": 0.5079,
+      "step": 14806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0204089170810816e-06,
+      "loss": 0.474,
+      "step": 14807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.020272220523153e-06,
+      "loss": 0.4768,
+      "step": 14808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0201359828291345e-06,
+      "loss": 0.5162,
+      "step": 14809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.02000020400529e-06,
+      "loss": 0.4994,
+      "step": 14810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.019864884057865e-06,
+      "loss": 0.474,
+      "step": 14811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0197300229930785e-06,
+      "loss": 0.5176,
+      "step": 14812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.019595620817133e-06,
+      "loss": 0.4873,
+      "step": 14813
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.019461677536209e-06,
+      "loss": 0.5721,
+      "step": 14814
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0193281931564626e-06,
+      "loss": 0.4807,
+      "step": 14815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.019195167684033e-06,
+      "loss": 0.5379,
+      "step": 14816
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.019062601125037e-06,
+      "loss": 0.4905,
+      "step": 14817
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0189304934855703e-06,
+      "loss": 0.4763,
+      "step": 14818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0187988447717047e-06,
+      "loss": 0.5045,
+      "step": 14819
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018667654989496e-06,
+      "loss": 0.5018,
+      "step": 14820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018536924144975e-06,
+      "loss": 0.488,
+      "step": 14821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0184066522441526e-06,
+      "loss": 0.4884,
+      "step": 14822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0182768392930198e-06,
+      "loss": 0.5302,
+      "step": 14823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0181474852975436e-06,
+      "loss": 0.5128,
+      "step": 14824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018018590263673e-06,
+      "loss": 0.5154,
+      "step": 14825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017890154197334e-06,
+      "loss": 0.4413,
+      "step": 14826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0177621771044323e-06,
+      "loss": 0.4802,
+      "step": 14827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0176346589908516e-06,
+      "loss": 0.4421,
+      "step": 14828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017507599862456e-06,
+      "loss": 0.4856,
+      "step": 14829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017380999725087e-06,
+      "loss": 0.4766,
+      "step": 14830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0172548585845655e-06,
+      "loss": 0.4904,
+      "step": 14831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017129176446693e-06,
+      "loss": 0.4966,
+      "step": 14832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017003953317246e-06,
+      "loss": 0.4901,
+      "step": 14833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.016879189201984e-06,
+      "loss": 0.4347,
+      "step": 14834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.016754884106642e-06,
+      "loss": 0.4936,
+      "step": 14835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.016631038036937e-06,
+      "loss": 0.48,
+      "step": 14836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0165076509985617e-06,
+      "loss": 0.465,
+      "step": 14837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0163847229971913e-06,
+      "loss": 0.5015,
+      "step": 14838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0162622540384783e-06,
+      "loss": 0.5173,
+      "step": 14839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.016140244128051e-06,
+      "loss": 0.4824,
+      "step": 14840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0160186932715213e-06,
+      "loss": 0.4609,
+      "step": 14841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0158976014744783e-06,
+      "loss": 0.5072,
+      "step": 14842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015776968742488e-06,
+      "loss": 0.5165,
+      "step": 14843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0156567950810993e-06,
+      "loss": 0.4774,
+      "step": 14844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0155370804958364e-06,
+      "loss": 0.4858,
+      "step": 14845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0154178249922043e-06,
+      "loss": 0.4616,
+      "step": 14846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015299028575686e-06,
+      "loss": 0.5405,
+      "step": 14847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015180691251744e-06,
+      "loss": 0.5506,
+      "step": 14848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0150628130258176e-06,
+      "loss": 0.4597,
+      "step": 14849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0149453939033295e-06,
+      "loss": 0.4649,
+      "step": 14850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0148284338896775e-06,
+      "loss": 0.5243,
+      "step": 14851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0147119329902387e-06,
+      "loss": 0.5583,
+      "step": 14852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0145958912103714e-06,
+      "loss": 0.4713,
+      "step": 14853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01448030855541e-06,
+      "loss": 0.5325,
+      "step": 14854
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014365185030668e-06,
+      "loss": 0.532,
+      "step": 14855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014250520641441e-06,
+      "loss": 0.545,
+      "step": 14856
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0141363153929996e-06,
+      "loss": 0.4822,
+      "step": 14857
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0140225692905947e-06,
+      "loss": 0.4603,
+      "step": 14858
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.013909282339457e-06,
+      "loss": 0.4572,
+      "step": 14859
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0137964545447965e-06,
+      "loss": 0.5103,
+      "step": 14860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.013684085911798e-06,
+      "loss": 0.5279,
+      "step": 14861
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.013572176445632e-06,
+      "loss": 0.5061,
+      "step": 14862
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0134607261514405e-06,
+      "loss": 0.4716,
+      "step": 14863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01334973503435e-06,
+      "loss": 0.5037,
+      "step": 14864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0132392030994633e-06,
+      "loss": 0.504,
+      "step": 14865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.013129130351862e-06,
+      "loss": 0.5231,
+      "step": 14866
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0130195167966083e-06,
+      "loss": 0.4901,
+      "step": 14867
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0129103624387414e-06,
+      "loss": 0.5405,
+      "step": 14868
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.012801667283282e-06,
+      "loss": 0.4674,
+      "step": 14869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0126934313352245e-06,
+      "loss": 0.4513,
+      "step": 14870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0125856545995484e-06,
+      "loss": 0.4985,
+      "step": 14871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0124783370812075e-06,
+      "loss": 0.5166,
+      "step": 14872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.012371478785137e-06,
+      "loss": 0.4864,
+      "step": 14873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0122650797162505e-06,
+      "loss": 0.5263,
+      "step": 14874
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0121591398794397e-06,
+      "loss": 0.4668,
+      "step": 14875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0120536592795763e-06,
+      "loss": 0.463,
+      "step": 14876
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.011948637921508e-06,
+      "loss": 0.4906,
+      "step": 14877
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0118440758100666e-06,
+      "loss": 0.4815,
+      "step": 14878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0117399729500587e-06,
+      "loss": 0.4878,
+      "step": 14879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0116363293462708e-06,
+      "loss": 0.4435,
+      "step": 14880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.011533145003468e-06,
+      "loss": 0.4814,
+      "step": 14881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.011430419926396e-06,
+      "loss": 0.5004,
+      "step": 14882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0113281541197764e-06,
+      "loss": 0.4815,
+      "step": 14883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0112263475883113e-06,
+      "loss": 0.4792,
+      "step": 14884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0111250003366835e-06,
+      "loss": 0.4647,
+      "step": 14885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.011024112369552e-06,
+      "loss": 0.4574,
+      "step": 14886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0109236836915547e-06,
+      "loss": 0.446,
+      "step": 14887
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010823714307311e-06,
+      "loss": 0.4479,
+      "step": 14888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010724204221416e-06,
+      "loss": 0.5049,
+      "step": 14889
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010625153438445e-06,
+      "loss": 0.4898,
+      "step": 14890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0105265619629534e-06,
+      "loss": 0.4405,
+      "step": 14891
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010428429799475e-06,
+      "loss": 0.4435,
+      "step": 14892
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.01033075695252e-06,
+      "loss": 0.4525,
+      "step": 14893
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.01023354342658e-06,
+      "loss": 0.5096,
+      "step": 14894
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010136789226125e-06,
+      "loss": 0.5132,
+      "step": 14895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0100404943556043e-06,
+      "loss": 0.4764,
+      "step": 14896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009944658819444e-06,
+      "loss": 0.4438,
+      "step": 14897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009849282622053e-06,
+      "loss": 0.4848,
+      "step": 14898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0097543657678144e-06,
+      "loss": 0.4389,
+      "step": 14899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0096599082610927e-06,
+      "loss": 0.4805,
+      "step": 14900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0095659101062315e-06,
+      "loss": 0.4481,
+      "step": 14901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0094723713075537e-06,
+      "loss": 0.478,
+      "step": 14902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009379291869359e-06,
+      "loss": 0.4704,
+      "step": 14903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009286671795928e-06,
+      "loss": 0.4789,
+      "step": 14904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009194511091518e-06,
+      "loss": 0.4791,
+      "step": 14905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0091028097603682e-06,
+      "loss": 0.4633,
+      "step": 14906
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0090115678066937e-06,
+      "loss": 0.4648,
+      "step": 14907
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0089207852346894e-06,
+      "loss": 0.4248,
+      "step": 14908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008830462048531e-06,
+      "loss": 0.4297,
+      "step": 14909
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0087405982523714e-06,
+      "loss": 0.4817,
+      "step": 14910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008651193850341e-06,
+      "loss": 0.4927,
+      "step": 14911
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008562248846551e-06,
+      "loss": 0.4748,
+      "step": 14912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0084737632450925e-06,
+      "loss": 0.4374,
+      "step": 14913
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008385737050033e-06,
+      "loss": 0.4948,
+      "step": 14914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0082981702654193e-06,
+      "loss": 0.4575,
+      "step": 14915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008211062895278e-06,
+      "loss": 0.4506,
+      "step": 14916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0081244149436156e-06,
+      "loss": 0.4887,
+      "step": 14917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008038226414414e-06,
+      "loss": 0.4724,
+      "step": 14918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0079524973116383e-06,
+      "loss": 0.4797,
+      "step": 14919
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0078672276392288e-06,
+      "loss": 0.4709,
+      "step": 14920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0077824174011062e-06,
+      "loss": 0.4792,
+      "step": 14921
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0076980666011717e-06,
+      "loss": 0.457,
+      "step": 14922
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0076141752433012e-06,
+      "loss": 0.4468,
+      "step": 14923
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007530743331353e-06,
+      "loss": 0.5194,
+      "step": 14924
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0074477708691645e-06,
+      "loss": 0.4637,
+      "step": 14925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0073652578605496e-06,
+      "loss": 0.4642,
+      "step": 14926
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0072832043093016e-06,
+      "loss": 0.4716,
+      "step": 14927
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007201610219195e-06,
+      "loss": 0.5058,
+      "step": 14928
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0071204755939802e-06,
+      "loss": 0.4547,
+      "step": 14929
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0070398004373883e-06,
+      "loss": 0.5081,
+      "step": 14930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006959584753128e-06,
+      "loss": 0.5238,
+      "step": 14931
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0068798285448877e-06,
+      "loss": 0.4852,
+      "step": 14932
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0068005318163355e-06,
+      "loss": 0.4489,
+      "step": 14933
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0067216945711165e-06,
+      "loss": 0.4784,
+      "step": 14934
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0066433168128564e-06,
+      "loss": 0.5001,
+      "step": 14935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0065653985451585e-06,
+      "loss": 0.459,
+      "step": 14936
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0064879397716057e-06,
+      "loss": 0.4339,
+      "step": 14937
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006410940495759e-06,
+      "loss": 0.5038,
+      "step": 14938
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0063344007211584e-06,
+      "loss": 0.5021,
+      "step": 14939
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0062583204513243e-06,
+      "loss": 0.418,
+      "step": 14940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006182699689755e-06,
+      "loss": 0.4728,
+      "step": 14941
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0061075384399266e-06,
+      "loss": 0.4936,
+      "step": 14942
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006032836705295e-06,
+      "loss": 0.4648,
+      "step": 14943
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0059585944892957e-06,
+      "loss": 0.4238,
+      "step": 14944
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0058848117953413e-06,
+      "loss": 0.4494,
+      "step": 14945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0058114886268256e-06,
+      "loss": 0.4658,
+      "step": 14946
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0057386249871183e-06,
+      "loss": 0.502,
+      "step": 14947
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005666220879571e-06,
+      "loss": 0.5023,
+      "step": 14948
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0055942763075135e-06,
+      "loss": 0.4625,
+      "step": 14949
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0055227912742518e-06,
+      "loss": 0.4527,
+      "step": 14950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005451765783073e-06,
+      "loss": 0.4481,
+      "step": 14951
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005381199837244e-06,
+      "loss": 0.4366,
+      "step": 14952
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005311093440009e-06,
+      "loss": 0.4462,
+      "step": 14953
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0052414465945914e-06,
+      "loss": 0.4303,
+      "step": 14954
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0051722593041926e-06,
+      "loss": 0.4904,
+      "step": 14955
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005103531571996e-06,
+      "loss": 0.4045,
+      "step": 14956
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0050352634011594e-06,
+      "loss": 0.4253,
+      "step": 14957
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004967454794823e-06,
+      "loss": 0.4592,
+      "step": 14958
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0049001057561047e-06,
+      "loss": 0.4984,
+      "step": 14959
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0048332162881e-06,
+      "loss": 0.4455,
+      "step": 14960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004766786393886e-06,
+      "loss": 0.4938,
+      "step": 14961
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004700816076516e-06,
+      "loss": 0.4459,
+      "step": 14962
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0046353053390236e-06,
+      "loss": 0.5086,
+      "step": 14963
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0045702541844204e-06,
+      "loss": 0.4478,
+      "step": 14964
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004505662615699e-06,
+      "loss": 0.4503,
+      "step": 14965
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004441530635827e-06,
+      "loss": 0.4411,
+      "step": 14966
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004377858247756e-06,
+      "loss": 0.4484,
+      "step": 14967
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0043146454544117e-06,
+      "loss": 0.4526,
+      "step": 14968
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004251892258701e-06,
+      "loss": 0.468,
+      "step": 14969
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004189598663509e-06,
+      "loss": 0.4886,
+      "step": 14970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0041277646716996e-06,
+      "loss": 0.4115,
+      "step": 14971
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004066390286117e-06,
+      "loss": 0.4324,
+      "step": 14972
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004005475509583e-06,
+      "loss": 0.4401,
+      "step": 14973
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0039450203448985e-06,
+      "loss": 0.4114,
+      "step": 14974
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0038850247948416e-06,
+      "loss": 0.4443,
+      "step": 14975
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0038254888621736e-06,
+      "loss": 0.4656,
+      "step": 14976
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0037664125496294e-06,
+      "loss": 0.4307,
+      "step": 14977
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003707795859926e-06,
+      "loss": 0.4465,
+      "step": 14978
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0036496387957595e-06,
+      "loss": 0.4324,
+      "step": 14979
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003591941359804e-06,
+      "loss": 0.4329,
+      "step": 14980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0035347035547106e-06,
+      "loss": 0.4689,
+      "step": 14981
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003477925383113e-06,
+      "loss": 0.4394,
+      "step": 14982
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003421606847621e-06,
+      "loss": 0.466,
+      "step": 14983
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003365747950823e-06,
+      "loss": 0.4592,
+      "step": 14984
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00331034869529e-06,
+      "loss": 0.4433,
+      "step": 14985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003255409083567e-06,
+      "loss": 0.4129,
+      "step": 14986
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003200929118181e-06,
+      "loss": 0.5091,
+      "step": 14987
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0031469088016377e-06,
+      "loss": 0.4553,
+      "step": 14988
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0030933481364193e-06,
+      "loss": 0.4397,
+      "step": 14989
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003040247124989e-06,
+      "loss": 0.459,
+      "step": 14990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0029876057697886e-06,
+      "loss": 0.4437,
+      "step": 14991
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0029354240732392e-06,
+      "loss": 0.4963,
+      "step": 14992
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002883702037739e-06,
+      "loss": 0.4433,
+      "step": 14993
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0028324396656677e-06,
+      "loss": 0.4419,
+      "step": 14994
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002781636959379e-06,
+      "loss": 0.4635,
+      "step": 14995
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002731293921213e-06,
+      "loss": 0.4917,
+      "step": 14996
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0026814105534823e-06,
+      "loss": 0.4189,
+      "step": 14997
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00263198685848e-06,
+      "loss": 0.4026,
+      "step": 14998
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0025830228384793e-06,
+      "loss": 0.3855,
+      "step": 14999
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0025345184957315e-06,
+      "loss": 0.4474,
+      "step": 15000
+    },
+    {
+      "epoch": 0.99,
+      "eval_text_loss": 0.6559115648269653,
+      "eval_text_runtime": 1.8307,
+      "eval_text_samples_per_second": 40.421,
+      "eval_text_steps_per_second": 1.092,
+      "step": 15000
+    },
+    {
+      "epoch": 0.99,
+      "eval_svit_loss": 0.7470967769622803,
+      "eval_svit_runtime": 6.6225,
+      "eval_svit_samples_per_second": 123.217,
+      "eval_svit_steps_per_second": 1.963,
+      "step": 15000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0024864738324668e-06,
+      "loss": 0.4559,
+      "step": 15001
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002438888850895e-06,
+      "loss": 0.4138,
+      "step": 15002
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002391763553203e-06,
+      "loss": 0.4422,
+      "step": 15003
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0023450979415575e-06,
+      "loss": 0.4565,
+      "step": 15004
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0022988920181054e-06,
+      "loss": 0.457,
+      "step": 15005
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002253145784969e-06,
+      "loss": 0.431,
+      "step": 15006
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0022078592442554e-06,
+      "loss": 0.4596,
+      "step": 15007
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0021630323980426e-06,
+      "loss": 0.4457,
+      "step": 15008
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002118665248395e-06,
+      "loss": 0.4397,
+      "step": 15009
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002074757797351e-06,
+      "loss": 0.4374,
+      "step": 15010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0020313100469294e-06,
+      "loss": 0.4572,
+      "step": 15011
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001988321999128e-06,
+      "loss": 0.4864,
+      "step": 15012
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001945793655924e-06,
+      "loss": 0.4409,
+      "step": 15013
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001903725019273e-06,
+      "loss": 0.4756,
+      "step": 15014
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001862116091108e-06,
+      "loss": 0.4457,
+      "step": 15015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0018209668733427e-06,
+      "loss": 0.4248,
+      "step": 15016
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0017802773678698e-06,
+      "loss": 0.4263,
+      "step": 15017
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0017400475765604e-06,
+      "loss": 0.432,
+      "step": 15018
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001700277501263e-06,
+      "loss": 0.4209,
+      "step": 15019
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0016609671438074e-06,
+      "loss": 0.4324,
+      "step": 15020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001622116505999e-06,
+      "loss": 0.4597,
+      "step": 15021
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0015837255896266e-06,
+      "loss": 0.4329,
+      "step": 15022
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0015457943964547e-06,
+      "loss": 0.4211,
+      "step": 15023
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0015083229282263e-06,
+      "loss": 0.3841,
+      "step": 15024
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0014713111866658e-06,
+      "loss": 0.4431,
+      "step": 15025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001434759173474e-06,
+      "loss": 0.4499,
+      "step": 15026
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001398666890332e-06,
+      "loss": 0.4319,
+      "step": 15027
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0013630343388983e-06,
+      "loss": 0.4601,
+      "step": 15028
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0013278615208134e-06,
+      "loss": 0.4395,
+      "step": 15029
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.001293148437692e-06,
+      "loss": 0.4599,
+      "step": 15030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0012588950911317e-06,
+      "loss": 0.4733,
+      "step": 15031
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.001225101482708e-06,
+      "loss": 0.4577,
+      "step": 15032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.001191767613974e-06,
+      "loss": 0.4002,
+      "step": 15033
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0011588934864615e-06,
+      "loss": 0.4395,
+      "step": 15034
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0011264791016827e-06,
+      "loss": 0.4249,
+      "step": 15035
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0010945244611292e-06,
+      "loss": 0.4339,
+      "step": 15036
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0010630295662684e-06,
+      "loss": 0.3991,
+      "step": 15037
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.001031994418549e-06,
+      "loss": 0.4044,
+      "step": 15038
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0010014190193985e-06,
+      "loss": 0.4289,
+      "step": 15039
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0009713033702218e-06,
+      "loss": 0.4389,
+      "step": 15040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0009416474724053e-06,
+      "loss": 0.4062,
+      "step": 15041
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0009124513273106e-06,
+      "loss": 0.4154,
+      "step": 15042
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0008837149362816e-06,
+      "loss": 0.4408,
+      "step": 15043
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0008554383006396e-06,
+      "loss": 0.4437,
+      "step": 15044
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0008276214216828e-06,
+      "loss": 0.3934,
+      "step": 15045
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000800264300693e-06,
+      "loss": 0.4522,
+      "step": 15046
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000773366938925e-06,
+      "loss": 0.4366,
+      "step": 15047
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0007469293376193e-06,
+      "loss": 0.4019,
+      "step": 15048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0007209514979877e-06,
+      "loss": 0.4015,
+      "step": 15049
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000695433421227e-06,
+      "loss": 0.4178,
+      "step": 15050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.00067037510851e-06,
+      "loss": 0.431,
+      "step": 15051
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000645776560988e-06,
+      "loss": 0.4406,
+      "step": 15052
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000621637779793e-06,
+      "loss": 0.4128,
+      "step": 15053
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005979587660354e-06,
+      "loss": 0.3849,
+      "step": 15054
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000574739520802e-06,
+      "loss": 0.4358,
+      "step": 15055
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005519800451624e-06,
+      "loss": 0.425,
+      "step": 15056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005296803401623e-06,
+      "loss": 0.4404,
+      "step": 15057
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000507840406826e-06,
+      "loss": 0.4304,
+      "step": 15058
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0004864602461594e-06,
+      "loss": 0.3761,
+      "step": 15059
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0004655398591444e-06,
+      "loss": 0.4348,
+      "step": 15060
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0004450792467446e-06,
+      "loss": 0.4323,
+      "step": 15061
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000425078409898e-06,
+      "loss": 0.4197,
+      "step": 15062
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000405537349527e-06,
+      "loss": 0.461,
+      "step": 15063
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003864560665276e-06,
+      "loss": 0.4242,
+      "step": 15064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003678345617797e-06,
+      "loss": 0.447,
+      "step": 15065
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003496728361375e-06,
+      "loss": 0.3874,
+      "step": 15066
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000331970890437e-06,
+      "loss": 0.3886,
+      "step": 15067
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000314728725491e-06,
+      "loss": 0.4183,
+      "step": 15068
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002979463420947e-06,
+      "loss": 0.3888,
+      "step": 15069
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002816237410175e-06,
+      "loss": 0.397,
+      "step": 15070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002657609230106e-06,
+      "loss": 0.4337,
+      "step": 15071
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002503578888035e-06,
+      "loss": 0.4477,
+      "step": 15072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002354146391036e-06,
+      "loss": 0.448,
+      "step": 15073
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000220931174599e-06,
+      "loss": 0.4268,
+      "step": 15074
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002069074959565e-06,
+      "loss": 0.3942,
+      "step": 15075
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001933436038193e-06,
+      "loss": 0.4031,
+      "step": 15076
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001802394988104e-06,
+      "loss": 0.3821,
+      "step": 15077
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000167595181535e-06,
+      "loss": 0.4255,
+      "step": 15078
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001554106525713e-06,
+      "loss": 0.4022,
+      "step": 15079
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001436859124824e-06,
+      "loss": 0.3879,
+      "step": 15080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000132420961806e-06,
+      "loss": 0.4226,
+      "step": 15081
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001216158010597e-06,
+      "loss": 0.4048,
+      "step": 15082
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001112704307414e-06,
+      "loss": 0.3949,
+      "step": 15083
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000101384851326e-06,
+      "loss": 0.3952,
+      "step": 15084
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000919590632686e-06,
+      "loss": 0.381,
+      "step": 15085
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000082993067002e-06,
+      "loss": 0.4508,
+      "step": 15086
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000744868629386e-06,
+      "loss": 0.404,
+      "step": 15087
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000664404514693e-06,
+      "loss": 0.391,
+      "step": 15088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000058853832965e-06,
+      "loss": 0.365,
+      "step": 15089
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000517270077737e-06,
+      "loss": 0.4056,
+      "step": 15090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000450599762224e-06,
+      "loss": 0.4019,
+      "step": 15091
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000388527386197e-06,
+      "loss": 0.4185,
+      "step": 15092
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000331052952495e-06,
+      "loss": 0.4225,
+      "step": 15093
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000278176463763e-06,
+      "loss": 0.3763,
+      "step": 15094
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000229897922432e-06,
+      "loss": 0.404,
+      "step": 15095
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000018621733073e-06,
+      "loss": 0.4054,
+      "step": 15096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000014713469065e-06,
+      "loss": 0.4629,
+      "step": 15097
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000112650004002e-06,
+      "loss": 0.3848,
+      "step": 15098
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000082763272366e-06,
+      "loss": 0.4161,
+      "step": 15099
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000057474497128e-06,
+      "loss": 0.3959,
+      "step": 15100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000036783679426e-06,
+      "loss": 0.3962,
+      "step": 15101
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000020690820234e-06,
+      "loss": 0.3956,
+      "step": 15102
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000000919592028e-06,
+      "loss": 0.3627,
+      "step": 15103
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000002298980093e-06,
+      "loss": 0.383,
+      "step": 15104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-06,
+      "loss": 0.3869,
+      "step": 15105
+    },
+    {
+      "epoch": 1.0,
+      "step": 15105,
+      "total_flos": 3.6153900577898e+20,
+      "train_loss": 0.6609694663064049,
+      "train_runtime": 96807.734,
+      "train_samples_per_second": 19.972,
+      "train_steps_per_second": 0.156
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 15105,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100000,
+  "total_flos": 3.6153900577898e+20,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}