diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,54616 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.7196140970556448,
+  "global_step": 9100,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9972321075523924e-05,
+      "loss": 4.767,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996836694345591e-05,
+      "loss": 4.3596,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99644128113879e-05,
+      "loss": 4.5169,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996045867931989e-05,
+      "loss": 4.3695,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995650454725188e-05,
+      "loss": 4.1322,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.995255041518387e-05,
+      "loss": 3.8682,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994859628311586e-05,
+      "loss": 4.0494,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9944642151047846e-05,
+      "loss": 4.0975,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9940688018979835e-05,
+      "loss": 3.7939,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9936733886911824e-05,
+      "loss": 3.6444,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.993277975484381e-05,
+      "loss": 3.5763,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99288256227758e-05,
+      "loss": 3.5716,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992487149070779e-05,
+      "loss": 3.4343,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992091735863978e-05,
+      "loss": 3.6212,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9916963226571775e-05,
+      "loss": 3.6875,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9913009094503757e-05,
+      "loss": 3.6689,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9909054962435745e-05,
+      "loss": 3.5561,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9905100830367734e-05,
+      "loss": 3.5279,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.990114669829973e-05,
+      "loss": 3.3662,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989719256623171e-05,
+      "loss": 3.4393,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.98932384341637e-05,
+      "loss": 3.4712,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.988928430209569e-05,
+      "loss": 3.4906,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9885330170027685e-05,
+      "loss": 3.0898,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9881376037959674e-05,
+      "loss": 3.095,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9877421905891656e-05,
+      "loss": 3.2337,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9873467773823645e-05,
+      "loss": 3.3593,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.986951364175564e-05,
+      "loss": 3.2735,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.986555950968763e-05,
+      "loss": 2.7868,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.986160537761961e-05,
+      "loss": 3.1718,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.985765124555161e-05,
+      "loss": 3.0401,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9853697113483596e-05,
+      "loss": 3.0889,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9849742981415585e-05,
+      "loss": 2.9358,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.984578884934757e-05,
+      "loss": 2.7485,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.984183471727956e-05,
+      "loss": 2.7861,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.983788058521155e-05,
+      "loss": 2.9983,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.983392645314354e-05,
+      "loss": 2.9298,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.982997232107552e-05,
+      "loss": 2.7851,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.982601818900752e-05,
+      "loss": 3.2828,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9822064056939506e-05,
+      "loss": 3.0774,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9818109924871495e-05,
+      "loss": 2.7081,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.981415579280348e-05,
+      "loss": 3.1135,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.981020166073547e-05,
+      "loss": 2.8273,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.980624752866746e-05,
+      "loss": 2.8364,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.980229339659945e-05,
+      "loss": 2.8584,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.979833926453143e-05,
+      "loss": 2.7399,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.979438513246343e-05,
+      "loss": 2.654,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.979043100039542e-05,
+      "loss": 2.9236,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9786476868327406e-05,
+      "loss": 2.8366,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9782522736259395e-05,
+      "loss": 3.0304,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9778568604191383e-05,
+      "loss": 2.9207,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.977461447212337e-05,
+      "loss": 2.6993,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.977066034005536e-05,
+      "loss": 2.6265,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.976670620798735e-05,
+      "loss": 2.7448,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.976275207591934e-05,
+      "loss": 2.7378,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.975879794385133e-05,
+      "loss": 2.7803,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9754843811783316e-05,
+      "loss": 2.7649,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9750889679715305e-05,
+      "loss": 3.0021,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9746935547647294e-05,
+      "loss": 2.8791,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.974298141557928e-05,
+      "loss": 2.7324,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.973902728351127e-05,
+      "loss": 2.7443,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.973507315144326e-05,
+      "loss": 2.8848,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.973111901937525e-05,
+      "loss": 2.7237,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.972716488730724e-05,
+      "loss": 2.6204,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.972321075523923e-05,
+      "loss": 2.776,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9719256623171216e-05,
+      "loss": 2.5551,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9715302491103205e-05,
+      "loss": 2.806,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9711348359035194e-05,
+      "loss": 2.9198,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.970739422696718e-05,
+      "loss": 2.7398,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.970344009489917e-05,
+      "loss": 2.545,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.969948596283116e-05,
+      "loss": 2.6266,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.969553183076315e-05,
+      "loss": 2.6437,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.969157769869514e-05,
+      "loss": 2.8542,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9687623566627127e-05,
+      "loss": 2.8376,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9683669434559115e-05,
+      "loss": 2.661,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9679715302491104e-05,
+      "loss": 2.3818,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.967576117042309e-05,
+      "loss": 2.4862,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.967180703835508e-05,
+      "loss": 2.4458,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.966785290628707e-05,
+      "loss": 2.5405,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.966389877421906e-05,
+      "loss": 2.6886,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9659944642151055e-05,
+      "loss": 2.6014,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.965599051008304e-05,
+      "loss": 2.4923,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9652036378015026e-05,
+      "loss": 2.617,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.964808224594702e-05,
+      "loss": 2.2373,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.964412811387901e-05,
+      "loss": 2.5651,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.964017398181099e-05,
+      "loss": 2.6257,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.963621984974298e-05,
+      "loss": 2.7719,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.963226571767498e-05,
+      "loss": 2.6988,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9628311585606966e-05,
+      "loss": 2.4753,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.962435745353895e-05,
+      "loss": 2.3971,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9620403321470937e-05,
+      "loss": 2.798,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.961644918940293e-05,
+      "loss": 2.6542,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.961249505733492e-05,
+      "loss": 2.7599,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96085409252669e-05,
+      "loss": 2.5064,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.960458679319889e-05,
+      "loss": 2.352,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.960063266113089e-05,
+      "loss": 2.5799,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9596678529062876e-05,
+      "loss": 2.7832,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.959272439699486e-05,
+      "loss": 2.5654,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958877026492685e-05,
+      "loss": 2.7591,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958481613285884e-05,
+      "loss": 2.5776,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958086200079083e-05,
+      "loss": 2.488,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9576907868722814e-05,
+      "loss": 2.7375,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.957295373665481e-05,
+      "loss": 2.5838,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.95689996045868e-05,
+      "loss": 2.4541,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.956504547251879e-05,
+      "loss": 2.6201,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.956109134045077e-05,
+      "loss": 2.8214,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9557137208382765e-05,
+      "loss": 2.5232,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9553183076314753e-05,
+      "loss": 2.5802,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.954922894424674e-05,
+      "loss": 2.6958,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9545274812178724e-05,
+      "loss": 2.6539,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.954132068011072e-05,
+      "loss": 2.5941,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.953736654804271e-05,
+      "loss": 2.5255,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.95334124159747e-05,
+      "loss": 2.6247,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.952945828390668e-05,
+      "loss": 2.7515,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9525504151838675e-05,
+      "loss": 2.4326,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9521550019770664e-05,
+      "loss": 2.4463,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.951759588770265e-05,
+      "loss": 2.705,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.951364175563464e-05,
+      "loss": 2.7498,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.950968762356663e-05,
+      "loss": 2.4851,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.950573349149862e-05,
+      "loss": 2.5479,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.950177935943061e-05,
+      "loss": 2.5636,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.94978252273626e-05,
+      "loss": 2.5321,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9493871095294586e-05,
+      "loss": 2.6615,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9489916963226575e-05,
+      "loss": 2.3996,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9485962831158563e-05,
+      "loss": 2.5744,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948200869909055e-05,
+      "loss": 2.6405,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.947805456702254e-05,
+      "loss": 2.4037,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.947410043495453e-05,
+      "loss": 2.6336,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.947014630288652e-05,
+      "loss": 2.5762,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.946619217081851e-05,
+      "loss": 2.7728,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9462238038750496e-05,
+      "loss": 2.6854,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9458283906682485e-05,
+      "loss": 2.2578,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9454329774614474e-05,
+      "loss": 2.7428,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.945037564254646e-05,
+      "loss": 2.4607,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.944642151047845e-05,
+      "loss": 2.2105,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.944246737841044e-05,
+      "loss": 2.6256,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.943851324634243e-05,
+      "loss": 2.4635,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.943455911427442e-05,
+      "loss": 2.5211,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.943060498220641e-05,
+      "loss": 2.3961,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9426650850138396e-05,
+      "loss": 2.646,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.942269671807039e-05,
+      "loss": 2.321,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9418742586002374e-05,
+      "loss": 2.4965,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.941478845393436e-05,
+      "loss": 2.577,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.941083432186635e-05,
+      "loss": 2.6231,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.940688018979835e-05,
+      "loss": 2.7244,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.940292605773033e-05,
+      "loss": 2.676,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.939897192566232e-05,
+      "loss": 2.8092,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9395017793594306e-05,
+      "loss": 2.6117,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.93910636615263e-05,
+      "loss": 2.5062,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9387109529458284e-05,
+      "loss": 2.7818,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.938315539739027e-05,
+      "loss": 2.4878,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.937920126532227e-05,
+      "loss": 2.6651,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.937524713325426e-05,
+      "loss": 2.4966,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.937129300118624e-05,
+      "loss": 2.7688,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.936733886911823e-05,
+      "loss": 2.5197,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9363384737050224e-05,
+      "loss": 2.206,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.935943060498221e-05,
+      "loss": 2.6327,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9355476472914195e-05,
+      "loss": 2.778,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9351522340846184e-05,
+      "loss": 2.5067,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.934756820877818e-05,
+      "loss": 2.9623,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.934361407671017e-05,
+      "loss": 2.3995,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.933965994464215e-05,
+      "loss": 2.5268,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.933570581257414e-05,
+      "loss": 2.6173,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9331751680506134e-05,
+      "loss": 2.4863,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.932779754843812e-05,
+      "loss": 2.5146,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9323843416370105e-05,
+      "loss": 2.5179,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9319889284302094e-05,
+      "loss": 2.4488,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.931593515223409e-05,
+      "loss": 2.3088,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.931198102016608e-05,
+      "loss": 2.7716,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.930802688809806e-05,
+      "loss": 2.64,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9304072756030056e-05,
+      "loss": 2.4598,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9300118623962045e-05,
+      "loss": 2.6048,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9296164491894034e-05,
+      "loss": 2.7289,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9292210359826016e-05,
+      "loss": 2.6761,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.928825622775801e-05,
+      "loss": 2.6837,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.928430209569e-05,
+      "loss": 2.2509,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.928034796362199e-05,
+      "loss": 2.4422,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.927639383155397e-05,
+      "loss": 2.4846,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.927243969948597e-05,
+      "loss": 2.3434,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9268485567417956e-05,
+      "loss": 2.4963,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9264531435349945e-05,
+      "loss": 2.3841,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9260577303281927e-05,
+      "loss": 2.2648,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.925662317121392e-05,
+      "loss": 2.4891,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.925266903914591e-05,
+      "loss": 2.4982,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.92487149070779e-05,
+      "loss": 2.1587,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.924476077500988e-05,
+      "loss": 2.3477,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.924080664294188e-05,
+      "loss": 2.4221,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9236852510873866e-05,
+      "loss": 2.3882,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9232898378805855e-05,
+      "loss": 2.6948,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9228944246737844e-05,
+      "loss": 2.4098,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.922499011466983e-05,
+      "loss": 2.5074,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.922103598260182e-05,
+      "loss": 2.3186,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.921708185053381e-05,
+      "loss": 2.1822,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.92131277184658e-05,
+      "loss": 2.3294,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.920917358639779e-05,
+      "loss": 2.294,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.920521945432978e-05,
+      "loss": 2.4551,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9201265322261766e-05,
+      "loss": 2.3843,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9197311190193755e-05,
+      "loss": 2.214,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9193357058125743e-05,
+      "loss": 2.0175,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.918940292605773e-05,
+      "loss": 2.8738,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.918544879398972e-05,
+      "loss": 2.2447,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.918149466192171e-05,
+      "loss": 2.7389,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.91775405298537e-05,
+      "loss": 2.4373,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.917358639778569e-05,
+      "loss": 2.3596,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.916963226571768e-05,
+      "loss": 2.4204,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9165678133649665e-05,
+      "loss": 2.5408,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9161724001581654e-05,
+      "loss": 2.5297,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.915776986951364e-05,
+      "loss": 2.56,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.915381573744564e-05,
+      "loss": 2.3533,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.914986160537762e-05,
+      "loss": 2.4923,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.914590747330961e-05,
+      "loss": 2.2952,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.91419533412416e-05,
+      "loss": 2.3479,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9137999209173594e-05,
+      "loss": 2.2571,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9134045077105576e-05,
+      "loss": 2.3025,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9130090945037565e-05,
+      "loss": 2.5348,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9126136812969553e-05,
+      "loss": 2.16,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.912218268090155e-05,
+      "loss": 2.2274,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.911822854883353e-05,
+      "loss": 2.1935,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.911427441676552e-05,
+      "loss": 2.6529,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.911032028469751e-05,
+      "loss": 2.3852,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9106366152629504e-05,
+      "loss": 2.7637,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9102412020561486e-05,
+      "loss": 2.2891,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9098457888493475e-05,
+      "loss": 2.382,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.909450375642547e-05,
+      "loss": 2.5936,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.909054962435746e-05,
+      "loss": 2.5766,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.908659549228944e-05,
+      "loss": 2.4206,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.908264136022143e-05,
+      "loss": 2.4334,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9078687228153426e-05,
+      "loss": 2.3427,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9074733096085415e-05,
+      "loss": 2.2512,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90707789640174e-05,
+      "loss": 2.2593,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9066824831949386e-05,
+      "loss": 2.4839,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906287069988138e-05,
+      "loss": 2.2702,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.905891656781337e-05,
+      "loss": 2.4496,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.905496243574535e-05,
+      "loss": 2.318,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.905100830367734e-05,
+      "loss": 2.1865,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.904705417160934e-05,
+      "loss": 2.4578,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9043100039541326e-05,
+      "loss": 2.4489,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.903914590747331e-05,
+      "loss": 2.4091,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.90351917754053e-05,
+      "loss": 2.3573,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.903123764333729e-05,
+      "loss": 2.3951,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.902728351126928e-05,
+      "loss": 2.597,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.902332937920126e-05,
+      "loss": 2.3305,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.901937524713326e-05,
+      "loss": 2.6198,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.901542111506525e-05,
+      "loss": 2.4753,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9011466982997236e-05,
+      "loss": 2.4833,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.900751285092922e-05,
+      "loss": 2.3666,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9003558718861214e-05,
+      "loss": 2.4592,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.89996045867932e-05,
+      "loss": 2.4247,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.899565045472519e-05,
+      "loss": 2.4126,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8991696322657174e-05,
+      "loss": 2.5492,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.898774219058917e-05,
+      "loss": 2.3567,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.898378805852116e-05,
+      "loss": 2.4577,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.897983392645315e-05,
+      "loss": 2.4993,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.897587979438513e-05,
+      "loss": 2.779,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8971925662317125e-05,
+      "loss": 2.3065,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.896797153024911e-05,
+      "loss": 2.2111,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.89640173981811e-05,
+      "loss": 2.5347,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.896006326611309e-05,
+      "loss": 2.4284,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.895610913404508e-05,
+      "loss": 2.2601,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.895215500197707e-05,
+      "loss": 2.2584,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.894820086990906e-05,
+      "loss": 2.2438,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8944246737841046e-05,
+      "loss": 2.6547,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8940292605773035e-05,
+      "loss": 2.5384,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8936338473705024e-05,
+      "loss": 2.2919,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.893238434163701e-05,
+      "loss": 2.2069,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8928430209569e-05,
+      "loss": 2.3713,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.892447607750099e-05,
+      "loss": 2.3639,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.892052194543298e-05,
+      "loss": 2.2645,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.891656781336497e-05,
+      "loss": 2.2864,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.891261368129696e-05,
+      "loss": 2.4459,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8908659549228946e-05,
+      "loss": 2.5416,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8904705417160935e-05,
+      "loss": 2.5047,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8900751285092923e-05,
+      "loss": 2.5546,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.889679715302491e-05,
+      "loss": 2.5979,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88928430209569e-05,
+      "loss": 2.4378,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 2.2762,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8884934756820885e-05,
+      "loss": 2.4382,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.888098062475287e-05,
+      "loss": 2.3463,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8877026492684856e-05,
+      "loss": 2.3383,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8873072360616845e-05,
+      "loss": 2.3337,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.886911822854884e-05,
+      "loss": 2.6359,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.886516409648082e-05,
+      "loss": 2.27,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.886120996441281e-05,
+      "loss": 2.2323,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88572558323448e-05,
+      "loss": 2.5137,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8853301700276796e-05,
+      "loss": 2.3147,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.884934756820878e-05,
+      "loss": 2.4897,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.884539343614077e-05,
+      "loss": 2.4153,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8841439304072756e-05,
+      "loss": 2.1367,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.883748517200475e-05,
+      "loss": 2.4361,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8833531039936733e-05,
+      "loss": 2.3649,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.882957690786872e-05,
+      "loss": 2.3277,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.882562277580072e-05,
+      "loss": 2.3274,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.882166864373271e-05,
+      "loss": 2.4558,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.881771451166469e-05,
+      "loss": 2.3032,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.881376037959668e-05,
+      "loss": 2.4684,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.880980624752867e-05,
+      "loss": 2.3098,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.880585211546066e-05,
+      "loss": 2.433,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8801897983392644e-05,
+      "loss": 2.4381,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.879794385132463e-05,
+      "loss": 2.5101,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.879398971925663e-05,
+      "loss": 2.2444,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.879003558718862e-05,
+      "loss": 2.4598,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.87860814551206e-05,
+      "loss": 2.4847,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.878212732305259e-05,
+      "loss": 2.4224,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8778173190984584e-05,
+      "loss": 2.7444,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.877421905891657e-05,
+      "loss": 2.2763,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8770264926848555e-05,
+      "loss": 2.3771,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8766310794780544e-05,
+      "loss": 2.5266,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.876235666271254e-05,
+      "loss": 2.3809,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.875840253064453e-05,
+      "loss": 2.4226,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.875444839857651e-05,
+      "loss": 2.6713,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8750494266508506e-05,
+      "loss": 2.4574,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8746540134440494e-05,
+      "loss": 2.4468,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.874258600237248e-05,
+      "loss": 2.4543,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8738631870304465e-05,
+      "loss": 2.47,
+      "step": 313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.873467773823646e-05,
+      "loss": 2.2606,
+      "step": 314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.873072360616845e-05,
+      "loss": 2.2641,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.872676947410044e-05,
+      "loss": 2.3247,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.872281534203243e-05,
+      "loss": 2.5016,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8718861209964416e-05,
+      "loss": 2.3194,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8714907077896405e-05,
+      "loss": 2.2846,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8710952945828394e-05,
+      "loss": 2.6164,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.870699881376038e-05,
+      "loss": 2.4861,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.870304468169237e-05,
+      "loss": 2.5447,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.869909054962436e-05,
+      "loss": 2.2217,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.869513641755635e-05,
+      "loss": 2.4419,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.869118228548834e-05,
+      "loss": 2.1938,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.868722815342033e-05,
+      "loss": 2.1865,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8683274021352316e-05,
+      "loss": 2.4815,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8679319889284304e-05,
+      "loss": 2.4057,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.867536575721629e-05,
+      "loss": 2.3431,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.867141162514828e-05,
+      "loss": 2.5853,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.866745749308027e-05,
+      "loss": 2.2617,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.866350336101226e-05,
+      "loss": 2.55,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.865954922894425e-05,
+      "loss": 2.3492,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.865559509687624e-05,
+      "loss": 2.3192,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8651640964808226e-05,
+      "loss": 2.1685,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8647686832740215e-05,
+      "loss": 2.1465,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8643732700672204e-05,
+      "loss": 2.5162,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.863977856860419e-05,
+      "loss": 2.4048,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.863582443653618e-05,
+      "loss": 2.3069,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.863187030446817e-05,
+      "loss": 2.5456,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.862791617240016e-05,
+      "loss": 2.3712,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.862396204033215e-05,
+      "loss": 2.347,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.862000790826414e-05,
+      "loss": 2.3399,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.861605377619613e-05,
+      "loss": 2.2335,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8612099644128115e-05,
+      "loss": 2.4074,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86081455120601e-05,
+      "loss": 2.2523,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.860419137999209e-05,
+      "loss": 2.5205,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.860023724792409e-05,
+      "loss": 2.5625,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.859628311585607e-05,
+      "loss": 2.7774,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.859232898378806e-05,
+      "loss": 2.2352,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.858837485172005e-05,
+      "loss": 2.3736,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.858442071965204e-05,
+      "loss": 2.445,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8580466587584025e-05,
+      "loss": 2.1757,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8576512455516014e-05,
+      "loss": 2.2711,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8572558323448e-05,
+      "loss": 2.8052,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.856860419138e-05,
+      "loss": 2.4265,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.856465005931198e-05,
+      "loss": 2.3376,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.856069592724397e-05,
+      "loss": 2.2165,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.855674179517596e-05,
+      "loss": 2.4041,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8552787663107954e-05,
+      "loss": 2.2693,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8548833531039936e-05,
+      "loss": 2.5233,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8544879398971925e-05,
+      "loss": 2.5309,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.854092526690392e-05,
+      "loss": 2.4554,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.853697113483591e-05,
+      "loss": 2.2398,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.853301700276789e-05,
+      "loss": 2.1243,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.852906287069988e-05,
+      "loss": 2.1369,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8525108738631876e-05,
+      "loss": 2.3576,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8521154606563864e-05,
+      "loss": 2.252,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8517200474495846e-05,
+      "loss": 2.4865,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8513246342427835e-05,
+      "loss": 2.3622,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.850929221035983e-05,
+      "loss": 2.4578,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.850533807829182e-05,
+      "loss": 2.2419,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.850138394622381e-05,
+      "loss": 2.3595,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.849742981415579e-05,
+      "loss": 2.3863,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8493475682087786e-05,
+      "loss": 2.3873,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8489521550019775e-05,
+      "loss": 2.4688,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8485567417951764e-05,
+      "loss": 2.2024,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.848161328588375e-05,
+      "loss": 2.3859,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.847765915381574e-05,
+      "loss": 2.3521,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.847370502174773e-05,
+      "loss": 2.1218,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.846975088967972e-05,
+      "loss": 2.2669,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.846579675761171e-05,
+      "loss": 2.5356,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.84618426255437e-05,
+      "loss": 2.6155,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8457888493475686e-05,
+      "loss": 2.4716,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8453934361407674e-05,
+      "loss": 2.4091,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.844998022933966e-05,
+      "loss": 2.6779,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.844602609727165e-05,
+      "loss": 2.448,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.844207196520364e-05,
+      "loss": 2.1958,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.843811783313563e-05,
+      "loss": 2.6181,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.843416370106762e-05,
+      "loss": 2.2527,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.843020956899961e-05,
+      "loss": 2.5835,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8426255436931596e-05,
+      "loss": 2.2824,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8422301304863585e-05,
+      "loss": 2.4161,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8418347172795574e-05,
+      "loss": 2.7243,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.841439304072756e-05,
+      "loss": 2.362,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.841043890865955e-05,
+      "loss": 2.3806,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.840648477659154e-05,
+      "loss": 2.2457,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.840253064452353e-05,
+      "loss": 2.5105,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.839857651245552e-05,
+      "loss": 2.274,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.839462238038751e-05,
+      "loss": 2.4724,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8390668248319496e-05,
+      "loss": 2.2172,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8386714116251484e-05,
+      "loss": 2.5117,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.838275998418347e-05,
+      "loss": 2.4435,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.837880585211546e-05,
+      "loss": 2.408,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.837485172004745e-05,
+      "loss": 2.408,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.837089758797944e-05,
+      "loss": 2.3648,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.836694345591143e-05,
+      "loss": 2.2385,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.836298932384342e-05,
+      "loss": 2.5754,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8359035191775406e-05,
+      "loss": 2.2799,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8355081059707395e-05,
+      "loss": 2.5605,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8351126927639384e-05,
+      "loss": 2.5141,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.834717279557138e-05,
+      "loss": 2.4916,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.834321866350336e-05,
+      "loss": 2.4921,
+      "step": 413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.833926453143535e-05,
+      "loss": 2.4425,
+      "step": 414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.833531039936734e-05,
+      "loss": 2.2882,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8331356267299335e-05,
+      "loss": 2.7359,
+      "step": 416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.832740213523132e-05,
+      "loss": 2.4507,
+      "step": 417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8323448003163306e-05,
+      "loss": 2.3708,
+      "step": 418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8319493871095295e-05,
+      "loss": 2.5657,
+      "step": 419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.831553973902729e-05,
+      "loss": 2.5326,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.831158560695927e-05,
+      "loss": 2.1654,
+      "step": 421
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.830763147489126e-05,
+      "loss": 2.3933,
+      "step": 422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.830367734282325e-05,
+      "loss": 2.1338,
+      "step": 423
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8299723210755245e-05,
+      "loss": 2.4947,
+      "step": 424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.829576907868723e-05,
+      "loss": 2.2321,
+      "step": 425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8291814946619216e-05,
+      "loss": 2.7057,
+      "step": 426
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8287860814551205e-05,
+      "loss": 2.2711,
+      "step": 427
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.82839066824832e-05,
+      "loss": 2.3861,
+      "step": 428
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.827995255041518e-05,
+      "loss": 2.1532,
+      "step": 429
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.827599841834717e-05,
+      "loss": 2.3942,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.827204428627917e-05,
+      "loss": 2.3375,
+      "step": 431
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8268090154211156e-05,
+      "loss": 2.3975,
+      "step": 432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8264136022143145e-05,
+      "loss": 2.0476,
+      "step": 433
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.826018189007513e-05,
+      "loss": 2.2775,
+      "step": 434
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.825622775800712e-05,
+      "loss": 2.4367,
+      "step": 435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.825227362593911e-05,
+      "loss": 2.2874,
+      "step": 436
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.82483194938711e-05,
+      "loss": 2.2442,
+      "step": 437
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.824436536180308e-05,
+      "loss": 2.2624,
+      "step": 438
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.824041122973508e-05,
+      "loss": 2.2286,
+      "step": 439
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.823645709766707e-05,
+      "loss": 2.468,
+      "step": 440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8232502965599055e-05,
+      "loss": 2.0985,
+      "step": 441
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.822854883353104e-05,
+      "loss": 2.2316,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.822459470146303e-05,
+      "loss": 2.2536,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.822064056939502e-05,
+      "loss": 2.3628,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.821668643732701e-05,
+      "loss": 2.251,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.821273230525899e-05,
+      "loss": 2.3903,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.820877817319099e-05,
+      "loss": 2.101,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.820482404112298e-05,
+      "loss": 2.4964,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8200869909054966e-05,
+      "loss": 2.0388,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8196915776986955e-05,
+      "loss": 2.3784,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8192961644918944e-05,
+      "loss": 2.3148,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.818900751285093e-05,
+      "loss": 2.2593,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.818505338078292e-05,
+      "loss": 2.2049,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.818109924871491e-05,
+      "loss": 2.2943,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.81771451166469e-05,
+      "loss": 2.1187,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.817319098457889e-05,
+      "loss": 2.5383,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.816923685251088e-05,
+      "loss": 2.1414,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8165282720442866e-05,
+      "loss": 2.0724,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8161328588374854e-05,
+      "loss": 2.1782,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.815737445630684e-05,
+      "loss": 2.2245,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.815342032423883e-05,
+      "loss": 2.3469,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.814946619217082e-05,
+      "loss": 2.3125,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.814551206010281e-05,
+      "loss": 2.5177,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.81415579280348e-05,
+      "loss": 2.3273,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.813760379596679e-05,
+      "loss": 2.3367,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8133649663898776e-05,
+      "loss": 2.3516,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8129695531830765e-05,
+      "loss": 2.3921,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8125741399762754e-05,
+      "loss": 2.3521,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.812178726769474e-05,
+      "loss": 2.5433,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.811783313562673e-05,
+      "loss": 2.4045,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.811387900355872e-05,
+      "loss": 2.2905,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.810992487149071e-05,
+      "loss": 2.2085,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.81059707394227e-05,
+      "loss": 2.2548,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.810201660735469e-05,
+      "loss": 2.2672,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8098062475286676e-05,
+      "loss": 2.491,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8094108343218664e-05,
+      "loss": 2.4696,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.809015421115065e-05,
+      "loss": 2.3047,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.808620007908264e-05,
+      "loss": 1.8936,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.808224594701463e-05,
+      "loss": 2.4372,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.807829181494662e-05,
+      "loss": 2.3451,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.807433768287861e-05,
+      "loss": 2.2427,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.80703835508106e-05,
+      "loss": 2.5037,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8066429418742586e-05,
+      "loss": 2.2146,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.806247528667458e-05,
+      "loss": 2.364,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8058521154606564e-05,
+      "loss": 2.3199,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.805456702253855e-05,
+      "loss": 2.3413,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.805061289047054e-05,
+      "loss": 2.1827,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.804665875840254e-05,
+      "loss": 2.1797,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8042704626334526e-05,
+      "loss": 2.2559,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.803875049426651e-05,
+      "loss": 2.3819,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.80347963621985e-05,
+      "loss": 2.1204,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.803084223013049e-05,
+      "loss": 2.3256,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.802688809806248e-05,
+      "loss": 2.2683,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.802293396599446e-05,
+      "loss": 2.0614,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.801897983392645e-05,
+      "loss": 2.2131,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.801502570185845e-05,
+      "loss": 2.3713,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8011071569790437e-05,
+      "loss": 2.5495,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.800711743772242e-05,
+      "loss": 2.3671,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8003163305654414e-05,
+      "loss": 2.3088,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.79992091735864e-05,
+      "loss": 2.2768,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.799525504151839e-05,
+      "loss": 2.3115,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7991300909450374e-05,
+      "loss": 2.221,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.798734677738237e-05,
+      "loss": 2.4884,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.798339264531436e-05,
+      "loss": 2.4079,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.797943851324635e-05,
+      "loss": 2.3576,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.797548438117833e-05,
+      "loss": 2.1922,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7971530249110325e-05,
+      "loss": 2.4386,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7967576117042314e-05,
+      "loss": 2.1326,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.79636219849743e-05,
+      "loss": 2.2995,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7959667852906285e-05,
+      "loss": 2.5608,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.795571372083828e-05,
+      "loss": 2.4407,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.795175958877027e-05,
+      "loss": 2.1049,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.794780545670226e-05,
+      "loss": 2.0915,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.794385132463424e-05,
+      "loss": 2.5368,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7939897192566235e-05,
+      "loss": 2.1789,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7935943060498224e-05,
+      "loss": 2.2517,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.793198892843021e-05,
+      "loss": 2.4371,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.79280347963622e-05,
+      "loss": 2.5458,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.792408066429419e-05,
+      "loss": 2.529,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.792012653222618e-05,
+      "loss": 2.2936,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.791617240015817e-05,
+      "loss": 2.0149,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.791221826809016e-05,
+      "loss": 2.5432,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7908264136022146e-05,
+      "loss": 2.1795,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7904310003954135e-05,
+      "loss": 2.2337,
+      "step": 524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7900355871886124e-05,
+      "loss": 2.4108,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.789640173981811e-05,
+      "loss": 2.2044,
+      "step": 526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.78924476077501e-05,
+      "loss": 2.2059,
+      "step": 527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.788849347568209e-05,
+      "loss": 2.1275,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.788453934361408e-05,
+      "loss": 2.102,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.788058521154607e-05,
+      "loss": 2.4161,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.787663107947806e-05,
+      "loss": 2.133,
+      "step": 531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7872676947410046e-05,
+      "loss": 2.3481,
+      "step": 532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7868722815342034e-05,
+      "loss": 2.2897,
+      "step": 533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.786476868327402e-05,
+      "loss": 2.2672,
+      "step": 534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.786081455120601e-05,
+      "loss": 2.1195,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7856860419138e-05,
+      "loss": 2.1859,
+      "step": 536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.785290628706999e-05,
+      "loss": 2.2968,
+      "step": 537
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.784895215500198e-05,
+      "loss": 2.4302,
+      "step": 538
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.784499802293397e-05,
+      "loss": 2.3847,
+      "step": 539
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7841043890865956e-05,
+      "loss": 2.1471,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7837089758797945e-05,
+      "loss": 2.296,
+      "step": 541
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7833135626729934e-05,
+      "loss": 2.1693,
+      "step": 542
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.782918149466192e-05,
+      "loss": 2.3893,
+      "step": 543
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.782522736259391e-05,
+      "loss": 2.53,
+      "step": 544
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.782127323052591e-05,
+      "loss": 2.6187,
+      "step": 545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.781731909845789e-05,
+      "loss": 2.3629,
+      "step": 546
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.781336496638988e-05,
+      "loss": 2.3636,
+      "step": 547
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.780941083432187e-05,
+      "loss": 2.3521,
+      "step": 548
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.780545670225386e-05,
+      "loss": 2.3706,
+      "step": 549
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7801502570185844e-05,
+      "loss": 2.3019,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.779754843811783e-05,
+      "loss": 2.1555,
+      "step": 551
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.779359430604983e-05,
+      "loss": 2.2995,
+      "step": 552
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.778964017398182e-05,
+      "loss": 2.4989,
+      "step": 553
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.77856860419138e-05,
+      "loss": 2.2584,
+      "step": 554
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.778173190984579e-05,
+      "loss": 2.3064,
+      "step": 555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7777777777777784e-05,
+      "loss": 2.1753,
+      "step": 556
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.777382364570977e-05,
+      "loss": 2.1814,
+      "step": 557
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7769869513641755e-05,
+      "loss": 2.5681,
+      "step": 558
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7765915381573744e-05,
+      "loss": 2.2059,
+      "step": 559
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.776196124950574e-05,
+      "loss": 2.1812,
+      "step": 560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.775800711743773e-05,
+      "loss": 2.1739,
+      "step": 561
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.775405298536971e-05,
+      "loss": 2.4098,
+      "step": 562
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.77500988533017e-05,
+      "loss": 2.2866,
+      "step": 563
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7746144721233695e-05,
+      "loss": 2.4782,
+      "step": 564
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7742190589165684e-05,
+      "loss": 2.4297,
+      "step": 565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7738236457097666e-05,
+      "loss": 2.4175,
+      "step": 566
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.7734282325029654e-05,
+      "loss": 2.1168,
+      "step": 567
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.773032819296165e-05,
+      "loss": 2.2938,
+      "step": 568
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.772637406089364e-05,
+      "loss": 2.2185,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.772241992882562e-05,
+      "loss": 2.0953,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7718465796757617e-05,
+      "loss": 2.2069,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7714511664689605e-05,
+      "loss": 2.0762,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7710557532621594e-05,
+      "loss": 2.0331,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7706603400553576e-05,
+      "loss": 2.3355,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.770264926848557e-05,
+      "loss": 2.2116,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.769869513641756e-05,
+      "loss": 2.4427,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.769474100434955e-05,
+      "loss": 2.3443,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.769078687228153e-05,
+      "loss": 2.1728,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.768683274021353e-05,
+      "loss": 2.5352,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7682878608145516e-05,
+      "loss": 2.268,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7678924476077505e-05,
+      "loss": 2.0959,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.767497034400949e-05,
+      "loss": 2.0998,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.767101621194148e-05,
+      "loss": 2.4436,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.766706207987347e-05,
+      "loss": 2.1733,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.766310794780546e-05,
+      "loss": 2.5413,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.765915381573745e-05,
+      "loss": 2.1588,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.765519968366944e-05,
+      "loss": 1.9933,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7651245551601427e-05,
+      "loss": 2.3438,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7647291419533415e-05,
+      "loss": 2.3166,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7643337287465404e-05,
+      "loss": 2.3378,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.763938315539739e-05,
+      "loss": 2.4174,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.763542902332938e-05,
+      "loss": 2.303,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.763147489126137e-05,
+      "loss": 2.2067,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.762752075919336e-05,
+      "loss": 2.5826,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.762356662712535e-05,
+      "loss": 2.1842,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.761961249505734e-05,
+      "loss": 2.2318,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7615658362989326e-05,
+      "loss": 2.2705,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7611704230921315e-05,
+      "loss": 2.3468,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7607750098853304e-05,
+      "loss": 2.0514,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.760379596678529e-05,
+      "loss": 2.2341,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.759984183471728e-05,
+      "loss": 2.1751,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.759588770264927e-05,
+      "loss": 2.2858,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.759193357058126e-05,
+      "loss": 2.1223,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.758797943851325e-05,
+      "loss": 2.2344,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7584025306445243e-05,
+      "loss": 2.3855,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7580071174377225e-05,
+      "loss": 2.4153,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7576117042309214e-05,
+      "loss": 2.3842,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.75721629102412e-05,
+      "loss": 2.1153,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.75682087781732e-05,
+      "loss": 2.3779,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.756425464610518e-05,
+      "loss": 2.283,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.756030051403717e-05,
+      "loss": 2.6368,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.755634638196916e-05,
+      "loss": 2.3413,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7552392249901154e-05,
+      "loss": 2.1507,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7548438117833136e-05,
+      "loss": 2.53,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7544483985765125e-05,
+      "loss": 2.3087,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7540529853697114e-05,
+      "loss": 2.2205,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.753657572162911e-05,
+      "loss": 2.3946,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.753262158956109e-05,
+      "loss": 2.1964,
+      "step": 618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.752866745749308e-05,
+      "loss": 2.3008,
+      "step": 619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.752471332542507e-05,
+      "loss": 2.1466,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7520759193357065e-05,
+      "loss": 2.1727,
+      "step": 621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.751680506128905e-05,
+      "loss": 2.2288,
+      "step": 622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7512850929221036e-05,
+      "loss": 2.2441,
+      "step": 623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.750889679715303e-05,
+      "loss": 2.1413,
+      "step": 624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.750494266508502e-05,
+      "loss": 2.1497,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7500988533017e-05,
+      "loss": 2.2466,
+      "step": 626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.749703440094899e-05,
+      "loss": 2.1126,
+      "step": 627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7493080268880986e-05,
+      "loss": 2.223,
+      "step": 628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7489126136812975e-05,
+      "loss": 2.1147,
+      "step": 629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.748517200474496e-05,
+      "loss": 2.0963,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7481217872676946e-05,
+      "loss": 2.2381,
+      "step": 631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.747726374060894e-05,
+      "loss": 2.2906,
+      "step": 632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.747330960854093e-05,
+      "loss": 2.17,
+      "step": 633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.746935547647291e-05,
+      "loss": 2.3105,
+      "step": 634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.74654013444049e-05,
+      "loss": 2.2008,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.74614472123369e-05,
+      "loss": 2.3319,
+      "step": 636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7457493080268886e-05,
+      "loss": 2.2166,
+      "step": 637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.745353894820087e-05,
+      "loss": 2.2966,
+      "step": 638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7449584816132864e-05,
+      "loss": 2.3047,
+      "step": 639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.744563068406485e-05,
+      "loss": 2.2683,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.744167655199684e-05,
+      "loss": 2.3004,
+      "step": 641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.743772241992882e-05,
+      "loss": 2.3411,
+      "step": 642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.743376828786082e-05,
+      "loss": 2.2656,
+      "step": 643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.742981415579281e-05,
+      "loss": 2.3563,
+      "step": 644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7425860023724797e-05,
+      "loss": 2.1895,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.742190589165678e-05,
+      "loss": 2.2992,
+      "step": 646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7417951759588774e-05,
+      "loss": 2.3488,
+      "step": 647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.741399762752076e-05,
+      "loss": 2.265,
+      "step": 648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.741004349545275e-05,
+      "loss": 2.494,
+      "step": 649
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7406089363384734e-05,
+      "loss": 2.3814,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.740213523131673e-05,
+      "loss": 2.348,
+      "step": 651
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.739818109924872e-05,
+      "loss": 2.3121,
+      "step": 652
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.739422696718071e-05,
+      "loss": 2.1039,
+      "step": 653
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.739027283511269e-05,
+      "loss": 2.3803,
+      "step": 654
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7386318703044685e-05,
+      "loss": 1.9667,
+      "step": 655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7382364570976674e-05,
+      "loss": 2.2105,
+      "step": 656
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.737841043890866e-05,
+      "loss": 2.0981,
+      "step": 657
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.737445630684065e-05,
+      "loss": 2.5506,
+      "step": 658
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.737050217477264e-05,
+      "loss": 2.2128,
+      "step": 659
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.736654804270463e-05,
+      "loss": 2.2608,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.736259391063662e-05,
+      "loss": 2.3973,
+      "step": 661
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7358639778568607e-05,
+      "loss": 2.3934,
+      "step": 662
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7354685646500595e-05,
+      "loss": 2.2719,
+      "step": 663
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7350731514432584e-05,
+      "loss": 2.3033,
+      "step": 664
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.734677738236457e-05,
+      "loss": 2.327,
+      "step": 665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.734282325029656e-05,
+      "loss": 2.1913,
+      "step": 666
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.733886911822855e-05,
+      "loss": 2.1929,
+      "step": 667
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.733491498616054e-05,
+      "loss": 2.5119,
+      "step": 668
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.733096085409253e-05,
+      "loss": 2.7765,
+      "step": 669
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.732700672202452e-05,
+      "loss": 2.2982,
+      "step": 670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7323052589956506e-05,
+      "loss": 2.423,
+      "step": 671
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7319098457888495e-05,
+      "loss": 1.9505,
+      "step": 672
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.731514432582049e-05,
+      "loss": 2.2821,
+      "step": 673
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.731119019375247e-05,
+      "loss": 1.9487,
+      "step": 674
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.730723606168446e-05,
+      "loss": 2.2638,
+      "step": 675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.730328192961645e-05,
+      "loss": 2.2406,
+      "step": 676
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7299327797548446e-05,
+      "loss": 2.2516,
+      "step": 677
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.729537366548043e-05,
+      "loss": 2.1224,
+      "step": 678
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.729141953341242e-05,
+      "loss": 2.3096,
+      "step": 679
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7287465401344405e-05,
+      "loss": 2.3802,
+      "step": 680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.72835112692764e-05,
+      "loss": 2.1123,
+      "step": 681
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.727955713720838e-05,
+      "loss": 2.3172,
+      "step": 682
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.727560300514037e-05,
+      "loss": 1.9128,
+      "step": 683
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.727164887307236e-05,
+      "loss": 2.2989,
+      "step": 684
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7267694741004356e-05,
+      "loss": 2.0582,
+      "step": 685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.726374060893634e-05,
+      "loss": 2.1238,
+      "step": 686
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.725978647686833e-05,
+      "loss": 2.1495,
+      "step": 687
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7255832344800316e-05,
+      "loss": 2.3466,
+      "step": 688
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.725187821273231e-05,
+      "loss": 2.2849,
+      "step": 689
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7247924080664294e-05,
+      "loss": 2.5265,
+      "step": 690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.724396994859628e-05,
+      "loss": 2.3803,
+      "step": 691
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.724001581652828e-05,
+      "loss": 2.2295,
+      "step": 692
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.723606168446027e-05,
+      "loss": 2.0661,
+      "step": 693
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.723210755239225e-05,
+      "loss": 2.1448,
+      "step": 694
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.722815342032424e-05,
+      "loss": 2.1682,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7224199288256233e-05,
+      "loss": 2.2307,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.722024515618822e-05,
+      "loss": 2.2795,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7216291024120204e-05,
+      "loss": 2.1664,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.721233689205219e-05,
+      "loss": 2.3855,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.720838275998419e-05,
+      "loss": 2.3312,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.720442862791618e-05,
+      "loss": 2.0024,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.720047449584816e-05,
+      "loss": 2.1577,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.719652036378015e-05,
+      "loss": 2.1523,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7192566231712144e-05,
+      "loss": 2.1002,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.718861209964413e-05,
+      "loss": 2.3056,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7184657967576115e-05,
+      "loss": 2.1012,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7180703835508104e-05,
+      "loss": 2.3384,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.71767497034401e-05,
+      "loss": 2.1516,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.717279557137209e-05,
+      "loss": 2.1801,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.716884143930407e-05,
+      "loss": 2.2565,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7164887307236066e-05,
+      "loss": 2.4581,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7160933175168055e-05,
+      "loss": 2.3228,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7156979043100044e-05,
+      "loss": 2.3138,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7153024911032026e-05,
+      "loss": 2.0248,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.714907077896402e-05,
+      "loss": 2.2444,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.714511664689601e-05,
+      "loss": 2.2798,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7141162514828e-05,
+      "loss": 2.2515,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.713720838275998e-05,
+      "loss": 2.1511,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7133254250691976e-05,
+      "loss": 2.5378,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7129300118623965e-05,
+      "loss": 2.4241,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7125345986555954e-05,
+      "loss": 2.604,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.712139185448794e-05,
+      "loss": 2.4256,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.711743772241993e-05,
+      "loss": 2.1611,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.711348359035192e-05,
+      "loss": 2.2729,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.710952945828391e-05,
+      "loss": 2.1476,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.71055753262159e-05,
+      "loss": 2.2845,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.710162119414789e-05,
+      "loss": 2.2694,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7097667062079876e-05,
+      "loss": 2.4507,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7093712930011865e-05,
+      "loss": 2.1418,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7089758797943854e-05,
+      "loss": 2.3309,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.708580466587584e-05,
+      "loss": 2.029,
+      "step": 731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.708185053380783e-05,
+      "loss": 2.2431,
+      "step": 732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.707789640173982e-05,
+      "loss": 2.1959,
+      "step": 733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.707394226967181e-05,
+      "loss": 2.7845,
+      "step": 734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.70699881376038e-05,
+      "loss": 2.3315,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7066034005535787e-05,
+      "loss": 2.3687,
+      "step": 736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7062079873467775e-05,
+      "loss": 2.2977,
+      "step": 737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7058125741399764e-05,
+      "loss": 2.2497,
+      "step": 738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.705417160933175e-05,
+      "loss": 2.1106,
+      "step": 739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.705021747726374e-05,
+      "loss": 1.9944,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.704626334519573e-05,
+      "loss": 2.2248,
+      "step": 741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.704230921312772e-05,
+      "loss": 2.0261,
+      "step": 742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.703835508105971e-05,
+      "loss": 2.2571,
+      "step": 743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.70344009489917e-05,
+      "loss": 2.2983,
+      "step": 744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.703044681692369e-05,
+      "loss": 2.3618,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7026492684855675e-05,
+      "loss": 2.1231,
+      "step": 746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.7022538552787664e-05,
+      "loss": 2.1561,
+      "step": 747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.701858442071965e-05,
+      "loss": 2.1391,
+      "step": 748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.701463028865165e-05,
+      "loss": 2.1948,
+      "step": 749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.701067615658363e-05,
+      "loss": 2.4197,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.700672202451562e-05,
+      "loss": 1.9594,
+      "step": 751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.700276789244761e-05,
+      "loss": 2.1831,
+      "step": 752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.69988137603796e-05,
+      "loss": 2.7752,
+      "step": 753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6994859628311585e-05,
+      "loss": 2.0455,
+      "step": 754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6990905496243574e-05,
+      "loss": 2.2946,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.698695136417556e-05,
+      "loss": 2.1255,
+      "step": 756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.698299723210756e-05,
+      "loss": 2.1222,
+      "step": 757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.697904310003954e-05,
+      "loss": 2.1091,
+      "step": 758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.697508896797153e-05,
+      "loss": 2.2178,
+      "step": 759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6971134835903525e-05,
+      "loss": 2.1275,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6967180703835514e-05,
+      "loss": 2.3273,
+      "step": 761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6963226571767496e-05,
+      "loss": 2.1413,
+      "step": 762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6959272439699485e-05,
+      "loss": 2.3237,
+      "step": 763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.695531830763148e-05,
+      "loss": 2.1144,
+      "step": 764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.695136417556347e-05,
+      "loss": 2.4213,
+      "step": 765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.694741004349545e-05,
+      "loss": 2.2825,
+      "step": 766
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.694345591142744e-05,
+      "loss": 1.9171,
+      "step": 767
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6939501779359436e-05,
+      "loss": 2.1328,
+      "step": 768
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6935547647291425e-05,
+      "loss": 2.3494,
+      "step": 769
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.693159351522341e-05,
+      "loss": 2.2711,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6927639383155395e-05,
+      "loss": 2.1123,
+      "step": 771
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.692368525108739e-05,
+      "loss": 2.3131,
+      "step": 772
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.691973111901938e-05,
+      "loss": 2.0484,
+      "step": 773
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.691577698695136e-05,
+      "loss": 2.393,
+      "step": 774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.691182285488335e-05,
+      "loss": 2.2267,
+      "step": 775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6907868722815346e-05,
+      "loss": 2.4471,
+      "step": 776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6903914590747335e-05,
+      "loss": 2.2645,
+      "step": 777
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.689996045867932e-05,
+      "loss": 2.3063,
+      "step": 778
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.689600632661131e-05,
+      "loss": 2.4757,
+      "step": 779
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.68920521945433e-05,
+      "loss": 2.0866,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.688809806247529e-05,
+      "loss": 2.0899,
+      "step": 781
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.688414393040728e-05,
+      "loss": 2.2907,
+      "step": 782
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.688018979833927e-05,
+      "loss": 2.4644,
+      "step": 783
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.687623566627126e-05,
+      "loss": 2.4093,
+      "step": 784
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6872281534203246e-05,
+      "loss": 2.2863,
+      "step": 785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6868327402135235e-05,
+      "loss": 2.3901,
+      "step": 786
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6864373270067223e-05,
+      "loss": 2.2188,
+      "step": 787
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.686041913799921e-05,
+      "loss": 2.405,
+      "step": 788
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.68564650059312e-05,
+      "loss": 2.3098,
+      "step": 789
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.685251087386319e-05,
+      "loss": 2.1768,
+      "step": 790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.684855674179518e-05,
+      "loss": 2.3401,
+      "step": 791
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.684460260972717e-05,
+      "loss": 2.3047,
+      "step": 792
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6840648477659156e-05,
+      "loss": 2.4161,
+      "step": 793
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6836694345591145e-05,
+      "loss": 2.3927,
+      "step": 794
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6832740213523134e-05,
+      "loss": 2.1044,
+      "step": 795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.682878608145512e-05,
+      "loss": 2.4419,
+      "step": 796
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.682483194938711e-05,
+      "loss": 2.3137,
+      "step": 797
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.68208778173191e-05,
+      "loss": 2.35,
+      "step": 798
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.681692368525109e-05,
+      "loss": 2.1998,
+      "step": 799
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.681296955318308e-05,
+      "loss": 2.2295,
+      "step": 800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.680901542111507e-05,
+      "loss": 2.145,
+      "step": 801
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6805061289047056e-05,
+      "loss": 2.5704,
+      "step": 802
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6801107156979045e-05,
+      "loss": 2.0745,
+      "step": 803
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6797153024911034e-05,
+      "loss": 2.1873,
+      "step": 804
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.679319889284302e-05,
+      "loss": 2.4738,
+      "step": 805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.678924476077501e-05,
+      "loss": 2.3162,
+      "step": 806
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6785290628707e-05,
+      "loss": 2.355,
+      "step": 807
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.678133649663899e-05,
+      "loss": 2.4005,
+      "step": 808
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.677738236457098e-05,
+      "loss": 2.1737,
+      "step": 809
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6773428232502967e-05,
+      "loss": 2.194,
+      "step": 810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6769474100434955e-05,
+      "loss": 2.259,
+      "step": 811
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6765519968366944e-05,
+      "loss": 2.262,
+      "step": 812
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.676156583629894e-05,
+      "loss": 2.2335,
+      "step": 813
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.675761170423092e-05,
+      "loss": 2.1377,
+      "step": 814
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.675365757216291e-05,
+      "loss": 2.4861,
+      "step": 815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.67497034400949e-05,
+      "loss": 2.1853,
+      "step": 816
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6745749308026895e-05,
+      "loss": 2.0231,
+      "step": 817
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.674179517595888e-05,
+      "loss": 1.9939,
+      "step": 818
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6737841043890866e-05,
+      "loss": 2.3109,
+      "step": 819
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.6733886911822855e-05,
+      "loss": 2.1635,
+      "step": 820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.672993277975485e-05,
+      "loss": 2.3422,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.672597864768683e-05,
+      "loss": 2.3392,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.672202451561882e-05,
+      "loss": 2.1516,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.671807038355081e-05,
+      "loss": 2.5854,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6714116251482806e-05,
+      "loss": 2.3392,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.671016211941479e-05,
+      "loss": 2.6434,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6706207987346777e-05,
+      "loss": 2.0993,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6702253855278765e-05,
+      "loss": 2.2656,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.669829972321076e-05,
+      "loss": 2.0852,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.669434559114274e-05,
+      "loss": 2.29,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.669039145907473e-05,
+      "loss": 2.2536,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.668643732700673e-05,
+      "loss": 2.1333,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6682483194938716e-05,
+      "loss": 2.2313,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.66785290628707e-05,
+      "loss": 2.0869,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.667457493080269e-05,
+      "loss": 2.1988,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.667062079873468e-05,
+      "loss": 2.1487,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 2.4323,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.666271253459866e-05,
+      "loss": 2.2104,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.665875840253064e-05,
+      "loss": 1.8647,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.665480427046264e-05,
+      "loss": 2.1551,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.665085013839463e-05,
+      "loss": 2.2719,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6646896006326616e-05,
+      "loss": 2.4962,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.66429418742586e-05,
+      "loss": 2.3745,
+      "step": 843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6638987742190593e-05,
+      "loss": 2.358,
+      "step": 844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.663503361012258e-05,
+      "loss": 2.1237,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.663107947805457e-05,
+      "loss": 2.4798,
+      "step": 846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.662712534598656e-05,
+      "loss": 2.2504,
+      "step": 847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.662317121391855e-05,
+      "loss": 2.4276,
+      "step": 848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.661921708185054e-05,
+      "loss": 2.2353,
+      "step": 849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6615262949782526e-05,
+      "loss": 2.1431,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6611308817714515e-05,
+      "loss": 2.2116,
+      "step": 851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6607354685646504e-05,
+      "loss": 2.2215,
+      "step": 852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.660340055357849e-05,
+      "loss": 2.1569,
+      "step": 853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.659944642151048e-05,
+      "loss": 2.3176,
+      "step": 854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.659549228944247e-05,
+      "loss": 2.2684,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.659153815737446e-05,
+      "loss": 2.312,
+      "step": 856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.658758402530645e-05,
+      "loss": 2.1572,
+      "step": 857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.658362989323844e-05,
+      "loss": 2.6379,
+      "step": 858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6579675761170426e-05,
+      "loss": 2.0938,
+      "step": 859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6575721629102415e-05,
+      "loss": 2.1921,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6571767497034403e-05,
+      "loss": 2.44,
+      "step": 861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.656781336496639e-05,
+      "loss": 2.3221,
+      "step": 862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.656385923289838e-05,
+      "loss": 2.1527,
+      "step": 863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.655990510083037e-05,
+      "loss": 2.211,
+      "step": 864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.655595096876236e-05,
+      "loss": 1.9453,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.655199683669435e-05,
+      "loss": 2.5956,
+      "step": 866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6548042704626336e-05,
+      "loss": 2.2691,
+      "step": 867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6544088572558325e-05,
+      "loss": 2.1985,
+      "step": 868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6540134440490314e-05,
+      "loss": 2.4106,
+      "step": 869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.65361803084223e-05,
+      "loss": 2.221,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.653222617635429e-05,
+      "loss": 2.2575,
+      "step": 871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.652827204428628e-05,
+      "loss": 2.149,
+      "step": 872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.652431791221827e-05,
+      "loss": 2.3035,
+      "step": 873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.652036378015026e-05,
+      "loss": 2.1636,
+      "step": 874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.651640964808225e-05,
+      "loss": 2.4153,
+      "step": 875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6512455516014236e-05,
+      "loss": 2.1418,
+      "step": 876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6508501383946225e-05,
+      "loss": 2.1544,
+      "step": 877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6504547251878214e-05,
+      "loss": 2.2357,
+      "step": 878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.65005931198102e-05,
+      "loss": 2.3487,
+      "step": 879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.649663898774219e-05,
+      "loss": 2.1337,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.649268485567418e-05,
+      "loss": 2.2888,
+      "step": 881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.648873072360617e-05,
+      "loss": 2.2099,
+      "step": 882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.648477659153816e-05,
+      "loss": 1.9662,
+      "step": 883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6480822459470146e-05,
+      "loss": 2.2563,
+      "step": 884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.647686832740214e-05,
+      "loss": 2.3288,
+      "step": 885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6472914195334124e-05,
+      "loss": 2.0964,
+      "step": 886
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.646896006326611e-05,
+      "loss": 2.3547,
+      "step": 887
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.64650059311981e-05,
+      "loss": 2.1257,
+      "step": 888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.64610517991301e-05,
+      "loss": 2.2271,
+      "step": 889
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.645709766706208e-05,
+      "loss": 2.0774,
+      "step": 890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.645314353499407e-05,
+      "loss": 2.4196,
+      "step": 891
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.644918940292606e-05,
+      "loss": 2.4203,
+      "step": 892
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.644523527085805e-05,
+      "loss": 2.651,
+      "step": 893
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.644128113879004e-05,
+      "loss": 2.2526,
+      "step": 894
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6437327006722024e-05,
+      "loss": 2.3328,
+      "step": 895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.643337287465401e-05,
+      "loss": 2.2416,
+      "step": 896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.642941874258601e-05,
+      "loss": 2.2788,
+      "step": 897
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6425464610518e-05,
+      "loss": 2.1144,
+      "step": 898
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.642151047844998e-05,
+      "loss": 2.2401,
+      "step": 899
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6417556346381974e-05,
+      "loss": 2.0126,
+      "step": 900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.641360221431396e-05,
+      "loss": 2.1389,
+      "step": 901
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.640964808224595e-05,
+      "loss": 2.3421,
+      "step": 902
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6405693950177934e-05,
+      "loss": 2.3471,
+      "step": 903
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.640173981810993e-05,
+      "loss": 2.1582,
+      "step": 904
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.639778568604192e-05,
+      "loss": 2.0083,
+      "step": 905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.639383155397391e-05,
+      "loss": 2.0809,
+      "step": 906
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.638987742190589e-05,
+      "loss": 2.3343,
+      "step": 907
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6385923289837885e-05,
+      "loss": 2.1527,
+      "step": 908
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6381969157769874e-05,
+      "loss": 2.2522,
+      "step": 909
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.637801502570186e-05,
+      "loss": 2.4938,
+      "step": 910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6374060893633845e-05,
+      "loss": 2.1388,
+      "step": 911
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.637010676156584e-05,
+      "loss": 2.1309,
+      "step": 912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.636615262949783e-05,
+      "loss": 2.0934,
+      "step": 913
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.636219849742982e-05,
+      "loss": 2.2533,
+      "step": 914
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.63582443653618e-05,
+      "loss": 1.9911,
+      "step": 915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6354290233293796e-05,
+      "loss": 2.4011,
+      "step": 916
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6350336101225785e-05,
+      "loss": 1.8881,
+      "step": 917
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.634638196915777e-05,
+      "loss": 2.1734,
+      "step": 918
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.634242783708976e-05,
+      "loss": 2.0702,
+      "step": 919
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.633847370502175e-05,
+      "loss": 2.139,
+      "step": 920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.633451957295374e-05,
+      "loss": 2.3024,
+      "step": 921
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.633056544088573e-05,
+      "loss": 2.2434,
+      "step": 922
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.632661130881772e-05,
+      "loss": 2.375,
+      "step": 923
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6322657176749706e-05,
+      "loss": 2.4407,
+      "step": 924
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6318703044681695e-05,
+      "loss": 2.0921,
+      "step": 925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6314748912613684e-05,
+      "loss": 1.9379,
+      "step": 926
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.631079478054567e-05,
+      "loss": 2.1215,
+      "step": 927
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.630684064847766e-05,
+      "loss": 1.9427,
+      "step": 928
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.630288651640965e-05,
+      "loss": 2.2719,
+      "step": 929
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.629893238434164e-05,
+      "loss": 2.0316,
+      "step": 930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.629497825227363e-05,
+      "loss": 2.1333,
+      "step": 931
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.629102412020562e-05,
+      "loss": 2.0706,
+      "step": 932
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6287069988137606e-05,
+      "loss": 2.2865,
+      "step": 933
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6283115856069595e-05,
+      "loss": 2.2018,
+      "step": 934
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6279161724001583e-05,
+      "loss": 2.3598,
+      "step": 935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.627520759193357e-05,
+      "loss": 2.1145,
+      "step": 936
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.627125345986556e-05,
+      "loss": 2.4158,
+      "step": 937
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.626729932779755e-05,
+      "loss": 2.4874,
+      "step": 938
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.626334519572954e-05,
+      "loss": 2.0998,
+      "step": 939
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.625939106366153e-05,
+      "loss": 2.3658,
+      "step": 940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6255436931593516e-05,
+      "loss": 2.1531,
+      "step": 941
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6251482799525505e-05,
+      "loss": 2.2577,
+      "step": 942
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.6247528667457494e-05,
+      "loss": 2.1895,
+      "step": 943
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.624357453538948e-05,
+      "loss": 2.4013,
+      "step": 944
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.623962040332147e-05,
+      "loss": 2.3147,
+      "step": 945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.623566627125346e-05,
+      "loss": 2.2226,
+      "step": 946
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.623171213918545e-05,
+      "loss": 2.3043,
+      "step": 947
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.622775800711744e-05,
+      "loss": 2.3151,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.622380387504943e-05,
+      "loss": 2.3226,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.621984974298142e-05,
+      "loss": 2.1693,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6215895610913405e-05,
+      "loss": 2.1917,
+      "step": 951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6211941478845393e-05,
+      "loss": 2.5578,
+      "step": 952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.620798734677739e-05,
+      "loss": 1.9936,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.620403321470938e-05,
+      "loss": 2.3241,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.620007908264136e-05,
+      "loss": 2.0583,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.619612495057335e-05,
+      "loss": 2.5749,
+      "step": 956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6192170818505344e-05,
+      "loss": 2.4197,
+      "step": 957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.618821668643733e-05,
+      "loss": 2.1095,
+      "step": 958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6184262554369315e-05,
+      "loss": 2.1913,
+      "step": 959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6180308422301304e-05,
+      "loss": 2.1865,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.61763542902333e-05,
+      "loss": 2.136,
+      "step": 961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.617240015816529e-05,
+      "loss": 1.9779,
+      "step": 962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.616844602609727e-05,
+      "loss": 2.3219,
+      "step": 963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.616449189402926e-05,
+      "loss": 2.1443,
+      "step": 964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6160537761961255e-05,
+      "loss": 2.2597,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6156583629893244e-05,
+      "loss": 2.6044,
+      "step": 966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6152629497825226e-05,
+      "loss": 2.2036,
+      "step": 967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6148675365757215e-05,
+      "loss": 2.4152,
+      "step": 968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.614472123368921e-05,
+      "loss": 2.0628,
+      "step": 969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.61407671016212e-05,
+      "loss": 2.0857,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.613681296955318e-05,
+      "loss": 2.1386,
+      "step": 971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.613285883748518e-05,
+      "loss": 2.1654,
+      "step": 972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6128904705417166e-05,
+      "loss": 2.3326,
+      "step": 973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6124950573349154e-05,
+      "loss": 2.151,
+      "step": 974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6120996441281137e-05,
+      "loss": 2.0502,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.611704230921313e-05,
+      "loss": 2.459,
+      "step": 976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.611308817714512e-05,
+      "loss": 2.0725,
+      "step": 977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.610913404507711e-05,
+      "loss": 2.4039,
+      "step": 978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.610517991300909e-05,
+      "loss": 2.4225,
+      "step": 979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.610122578094109e-05,
+      "loss": 2.0799,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6097271648873076e-05,
+      "loss": 2.2708,
+      "step": 981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6093317516805065e-05,
+      "loss": 2.2068,
+      "step": 982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.608936338473705e-05,
+      "loss": 2.2917,
+      "step": 983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.608540925266904e-05,
+      "loss": 2.1273,
+      "step": 984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.608145512060103e-05,
+      "loss": 2.3619,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.607750098853302e-05,
+      "loss": 2.144,
+      "step": 986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.607354685646501e-05,
+      "loss": 2.2994,
+      "step": 987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6069592724397e-05,
+      "loss": 2.0675,
+      "step": 988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.606563859232899e-05,
+      "loss": 2.2647,
+      "step": 989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6061684460260976e-05,
+      "loss": 2.3505,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6057730328192965e-05,
+      "loss": 2.2674,
+      "step": 991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.605377619612495e-05,
+      "loss": 2.2451,
+      "step": 992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.604982206405694e-05,
+      "loss": 2.4339,
+      "step": 993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.604586793198893e-05,
+      "loss": 2.0169,
+      "step": 994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.604191379992092e-05,
+      "loss": 2.1316,
+      "step": 995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.603795966785291e-05,
+      "loss": 2.1342,
+      "step": 996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.60340055357849e-05,
+      "loss": 2.5163,
+      "step": 997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6030051403716886e-05,
+      "loss": 2.3677,
+      "step": 998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6026097271648875e-05,
+      "loss": 2.3414,
+      "step": 999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.6022143139580864e-05,
+      "loss": 2.1899,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.601818900751285e-05,
+      "loss": 2.5701,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.601423487544484e-05,
+      "loss": 2.2307,
+      "step": 1002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.601028074337683e-05,
+      "loss": 2.1473,
+      "step": 1003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.600632661130882e-05,
+      "loss": 2.3353,
+      "step": 1004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.600237247924081e-05,
+      "loss": 2.1581,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.59984183471728e-05,
+      "loss": 2.1135,
+      "step": 1006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5994464215104786e-05,
+      "loss": 2.1226,
+      "step": 1007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5990510083036775e-05,
+      "loss": 2.0165,
+      "step": 1008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5986555950968763e-05,
+      "loss": 2.1784,
+      "step": 1009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.598260181890076e-05,
+      "loss": 2.4959,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.597864768683274e-05,
+      "loss": 2.1368,
+      "step": 1011
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.597469355476473e-05,
+      "loss": 2.2598,
+      "step": 1012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.597073942269672e-05,
+      "loss": 2.0696,
+      "step": 1013
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5966785290628714e-05,
+      "loss": 2.2962,
+      "step": 1014
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5962831158560696e-05,
+      "loss": 2.1687,
+      "step": 1015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5958877026492685e-05,
+      "loss": 2.2026,
+      "step": 1016
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5954922894424674e-05,
+      "loss": 2.1367,
+      "step": 1017
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.595096876235667e-05,
+      "loss": 2.2,
+      "step": 1018
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.594701463028865e-05,
+      "loss": 2.0495,
+      "step": 1019
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.594306049822064e-05,
+      "loss": 2.1681,
+      "step": 1020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5939106366152636e-05,
+      "loss": 2.3618,
+      "step": 1021
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5935152234084625e-05,
+      "loss": 2.2785,
+      "step": 1022
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.593119810201661e-05,
+      "loss": 2.0355,
+      "step": 1023
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5927243969948596e-05,
+      "loss": 2.5434,
+      "step": 1024
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.592328983788059e-05,
+      "loss": 2.26,
+      "step": 1025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.591933570581258e-05,
+      "loss": 2.2096,
+      "step": 1026
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.591538157374456e-05,
+      "loss": 2.1815,
+      "step": 1027
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.591142744167655e-05,
+      "loss": 2.026,
+      "step": 1028
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.590747330960855e-05,
+      "loss": 2.3013,
+      "step": 1029
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5903519177540536e-05,
+      "loss": 2.3092,
+      "step": 1030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.589956504547252e-05,
+      "loss": 2.3664,
+      "step": 1031
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5895610913404506e-05,
+      "loss": 2.1294,
+      "step": 1032
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.58916567813365e-05,
+      "loss": 2.3074,
+      "step": 1033
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.588770264926849e-05,
+      "loss": 2.1126,
+      "step": 1034
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.588374851720047e-05,
+      "loss": 1.996,
+      "step": 1035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.587979438513246e-05,
+      "loss": 2.0866,
+      "step": 1036
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.587584025306446e-05,
+      "loss": 2.2885,
+      "step": 1037
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5871886120996446e-05,
+      "loss": 2.156,
+      "step": 1038
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.586793198892843e-05,
+      "loss": 2.2602,
+      "step": 1039
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5863977856860424e-05,
+      "loss": 2.2828,
+      "step": 1040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.586002372479241e-05,
+      "loss": 1.9647,
+      "step": 1041
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.58560695927244e-05,
+      "loss": 2.3748,
+      "step": 1042
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5852115460656384e-05,
+      "loss": 2.2986,
+      "step": 1043
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.584816132858838e-05,
+      "loss": 2.0221,
+      "step": 1044
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.584420719652037e-05,
+      "loss": 2.1357,
+      "step": 1045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.584025306445236e-05,
+      "loss": 2.4605,
+      "step": 1046
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.583629893238434e-05,
+      "loss": 2.4305,
+      "step": 1047
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5832344800316334e-05,
+      "loss": 2.1869,
+      "step": 1048
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.582839066824832e-05,
+      "loss": 2.3915,
+      "step": 1049
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.582443653618031e-05,
+      "loss": 2.0015,
+      "step": 1050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5820482404112294e-05,
+      "loss": 2.1571,
+      "step": 1051
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.581652827204429e-05,
+      "loss": 2.1117,
+      "step": 1052
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.581257413997628e-05,
+      "loss": 2.3075,
+      "step": 1053
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.580862000790827e-05,
+      "loss": 2.174,
+      "step": 1054
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5804665875840256e-05,
+      "loss": 2.125,
+      "step": 1055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5800711743772245e-05,
+      "loss": 2.0908,
+      "step": 1056
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5796757611704234e-05,
+      "loss": 2.3987,
+      "step": 1057
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.579280347963622e-05,
+      "loss": 1.9256,
+      "step": 1058
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.578884934756821e-05,
+      "loss": 2.2389,
+      "step": 1059
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.57848952155002e-05,
+      "loss": 1.8435,
+      "step": 1060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.578094108343219e-05,
+      "loss": 1.9078,
+      "step": 1061
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.577698695136418e-05,
+      "loss": 2.1213,
+      "step": 1062
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.577303281929617e-05,
+      "loss": 2.0243,
+      "step": 1063
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5769078687228156e-05,
+      "loss": 2.1971,
+      "step": 1064
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5765124555160144e-05,
+      "loss": 1.9158,
+      "step": 1065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.576117042309213e-05,
+      "loss": 2.1813,
+      "step": 1066
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.575721629102412e-05,
+      "loss": 2.1625,
+      "step": 1067
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.575326215895611e-05,
+      "loss": 2.161,
+      "step": 1068
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.57493080268881e-05,
+      "loss": 2.5311,
+      "step": 1069
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.574535389482009e-05,
+      "loss": 2.3882,
+      "step": 1070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.574139976275208e-05,
+      "loss": 2.4426,
+      "step": 1071
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5737445630684066e-05,
+      "loss": 2.0849,
+      "step": 1072
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.5733491498616055e-05,
+      "loss": 2.3297,
+      "step": 1073
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.572953736654805e-05,
+      "loss": 2.4922,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.572558323448003e-05,
+      "loss": 2.3662,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.572162910241202e-05,
+      "loss": 2.1291,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.571767497034401e-05,
+      "loss": 2.2278,
+      "step": 1077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5713720838276006e-05,
+      "loss": 2.3648,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.570976670620799e-05,
+      "loss": 2.2257,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.570581257413998e-05,
+      "loss": 2.319,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5701858442071966e-05,
+      "loss": 2.1695,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.569790431000396e-05,
+      "loss": 1.9927,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.569395017793594e-05,
+      "loss": 2.301,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.568999604586793e-05,
+      "loss": 2.4287,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.568604191379992e-05,
+      "loss": 2.126,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.568208778173192e-05,
+      "loss": 2.2052,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.56781336496639e-05,
+      "loss": 2.378,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.567417951759589e-05,
+      "loss": 2.2391,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5670225385527876e-05,
+      "loss": 2.382,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.566627125345987e-05,
+      "loss": 2.5005,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5662317121391854e-05,
+      "loss": 2.3145,
+      "step": 1091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.565836298932384e-05,
+      "loss": 2.1258,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.565440885725584e-05,
+      "loss": 2.0855,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.565045472518783e-05,
+      "loss": 2.0648,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.564650059311981e-05,
+      "loss": 2.2541,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.56425464610518e-05,
+      "loss": 2.2448,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5638592328983794e-05,
+      "loss": 1.9828,
+      "step": 1097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.563463819691578e-05,
+      "loss": 2.5077,
+      "step": 1098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5630684064847765e-05,
+      "loss": 2.0935,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5626729932779753e-05,
+      "loss": 2.3648,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.562277580071175e-05,
+      "loss": 2.4379,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.561882166864374e-05,
+      "loss": 1.9723,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.561486753657572e-05,
+      "loss": 2.4027,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.561091340450771e-05,
+      "loss": 2.103,
+      "step": 1104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5606959272439704e-05,
+      "loss": 2.37,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.560300514037169e-05,
+      "loss": 2.226,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5599051008303675e-05,
+      "loss": 2.0964,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.559509687623567e-05,
+      "loss": 2.3959,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.559114274416766e-05,
+      "loss": 2.0071,
+      "step": 1109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.558718861209965e-05,
+      "loss": 2.206,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.558323448003163e-05,
+      "loss": 2.3365,
+      "step": 1111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5579280347963626e-05,
+      "loss": 2.3236,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5575326215895615e-05,
+      "loss": 2.4677,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5571372083827604e-05,
+      "loss": 2.2042,
+      "step": 1114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5567417951759586e-05,
+      "loss": 2.3046,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.556346381969158e-05,
+      "loss": 2.2731,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.555950968762357e-05,
+      "loss": 1.9994,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.555555555555556e-05,
+      "loss": 2.2436,
+      "step": 1118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.555160142348754e-05,
+      "loss": 2.1884,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.554764729141954e-05,
+      "loss": 2.2754,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5543693159351526e-05,
+      "loss": 2.121,
+      "step": 1121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5539739027283514e-05,
+      "loss": 2.4957,
+      "step": 1122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5535784895215496e-05,
+      "loss": 2.3882,
+      "step": 1123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.553183076314749e-05,
+      "loss": 2.0761,
+      "step": 1124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.552787663107948e-05,
+      "loss": 1.9471,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.552392249901147e-05,
+      "loss": 2.254,
+      "step": 1126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.551996836694346e-05,
+      "loss": 2.2319,
+      "step": 1127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.551601423487545e-05,
+      "loss": 2.0617,
+      "step": 1128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5512060102807436e-05,
+      "loss": 2.3044,
+      "step": 1129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5508105970739425e-05,
+      "loss": 2.1825,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5504151838671414e-05,
+      "loss": 2.123,
+      "step": 1131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.55001977066034e-05,
+      "loss": 2.352,
+      "step": 1132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.549624357453539e-05,
+      "loss": 2.2347,
+      "step": 1133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.549228944246738e-05,
+      "loss": 2.4585,
+      "step": 1134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.548833531039937e-05,
+      "loss": 2.3281,
+      "step": 1135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.548438117833136e-05,
+      "loss": 2.2236,
+      "step": 1136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.548042704626335e-05,
+      "loss": 2.1792,
+      "step": 1137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5476472914195336e-05,
+      "loss": 2.0176,
+      "step": 1138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5472518782127324e-05,
+      "loss": 2.1284,
+      "step": 1139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.546856465005931e-05,
+      "loss": 2.2197,
+      "step": 1140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.54646105179913e-05,
+      "loss": 2.2208,
+      "step": 1141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.54606563859233e-05,
+      "loss": 2.4132,
+      "step": 1142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.545670225385528e-05,
+      "loss": 2.3592,
+      "step": 1143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.545274812178727e-05,
+      "loss": 2.3209,
+      "step": 1144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.544879398971926e-05,
+      "loss": 2.3641,
+      "step": 1145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.544483985765125e-05,
+      "loss": 2.2516,
+      "step": 1146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5440885725583235e-05,
+      "loss": 2.1105,
+      "step": 1147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5436931593515224e-05,
+      "loss": 2.4273,
+      "step": 1148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.543297746144721e-05,
+      "loss": 1.8066,
+      "step": 1149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.542902332937921e-05,
+      "loss": 2.2594,
+      "step": 1150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.542506919731119e-05,
+      "loss": 2.1246,
+      "step": 1151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.542111506524318e-05,
+      "loss": 2.1886,
+      "step": 1152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.541716093317517e-05,
+      "loss": 2.2391,
+      "step": 1153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5413206801107164e-05,
+      "loss": 2.3001,
+      "step": 1154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5409252669039146e-05,
+      "loss": 2.3262,
+      "step": 1155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5405298536971135e-05,
+      "loss": 2.1765,
+      "step": 1156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.540134440490312e-05,
+      "loss": 1.9879,
+      "step": 1157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.539739027283512e-05,
+      "loss": 2.5049,
+      "step": 1158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.53934361407671e-05,
+      "loss": 2.0744,
+      "step": 1159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.538948200869909e-05,
+      "loss": 2.2892,
+      "step": 1160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5385527876631085e-05,
+      "loss": 2.162,
+      "step": 1161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5381573744563074e-05,
+      "loss": 2.1814,
+      "step": 1162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5377619612495056e-05,
+      "loss": 2.3168,
+      "step": 1163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5373665480427045e-05,
+      "loss": 2.0158,
+      "step": 1164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.536971134835904e-05,
+      "loss": 2.0303,
+      "step": 1165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.536575721629103e-05,
+      "loss": 2.2147,
+      "step": 1166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.536180308422301e-05,
+      "loss": 2.3824,
+      "step": 1167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5357848952155e-05,
+      "loss": 2.1865,
+      "step": 1168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5353894820086996e-05,
+      "loss": 2.132,
+      "step": 1169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5349940688018985e-05,
+      "loss": 2.1499,
+      "step": 1170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.534598655595097e-05,
+      "loss": 2.3689,
+      "step": 1171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5342032423882956e-05,
+      "loss": 2.1931,
+      "step": 1172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.533807829181495e-05,
+      "loss": 2.4615,
+      "step": 1173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.533412415974694e-05,
+      "loss": 2.2037,
+      "step": 1174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.533017002767892e-05,
+      "loss": 2.0771,
+      "step": 1175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.532621589561091e-05,
+      "loss": 1.9844,
+      "step": 1176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.532226176354291e-05,
+      "loss": 2.327,
+      "step": 1177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5318307631474895e-05,
+      "loss": 1.9725,
+      "step": 1178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.531435349940688e-05,
+      "loss": 2.3536,
+      "step": 1179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.531039936733887e-05,
+      "loss": 2.155,
+      "step": 1180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.530644523527086e-05,
+      "loss": 2.0619,
+      "step": 1181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.530249110320285e-05,
+      "loss": 2.2591,
+      "step": 1182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.529853697113483e-05,
+      "loss": 2.4034,
+      "step": 1183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.529458283906683e-05,
+      "loss": 2.203,
+      "step": 1184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.529062870699882e-05,
+      "loss": 2.2351,
+      "step": 1185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5286674574930806e-05,
+      "loss": 2.0094,
+      "step": 1186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5282720442862795e-05,
+      "loss": 1.9726,
+      "step": 1187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5278766310794784e-05,
+      "loss": 2.4104,
+      "step": 1188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.527481217872677e-05,
+      "loss": 2.2534,
+      "step": 1189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.527085804665876e-05,
+      "loss": 1.9786,
+      "step": 1190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.526690391459075e-05,
+      "loss": 2.2218,
+      "step": 1191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.526294978252274e-05,
+      "loss": 2.2257,
+      "step": 1192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.525899565045473e-05,
+      "loss": 2.2366,
+      "step": 1193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.525504151838672e-05,
+      "loss": 2.2049,
+      "step": 1194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5251087386318706e-05,
+      "loss": 2.2071,
+      "step": 1195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.5247133254250694e-05,
+      "loss": 2.4032,
+      "step": 1196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.524317912218268e-05,
+      "loss": 2.3429,
+      "step": 1197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.523922499011467e-05,
+      "loss": 2.2886,
+      "step": 1198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.523527085804666e-05,
+      "loss": 2.4794,
+      "step": 1199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.523131672597865e-05,
+      "loss": 2.3191,
+      "step": 1200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.522736259391064e-05,
+      "loss": 2.2367,
+      "step": 1201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.522340846184263e-05,
+      "loss": 1.9883,
+      "step": 1202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5219454329774616e-05,
+      "loss": 2.4617,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5215500197706605e-05,
+      "loss": 2.1199,
+      "step": 1204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5211546065638594e-05,
+      "loss": 2.2445,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.520759193357058e-05,
+      "loss": 2.3128,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.520363780150257e-05,
+      "loss": 2.1108,
+      "step": 1207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.519968366943456e-05,
+      "loss": 2.0835,
+      "step": 1208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.519572953736655e-05,
+      "loss": 2.3132,
+      "step": 1209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.519177540529854e-05,
+      "loss": 2.3716,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.518782127323053e-05,
+      "loss": 2.1205,
+      "step": 1211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5183867141162516e-05,
+      "loss": 2.1125,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5179913009094504e-05,
+      "loss": 2.2397,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.51759588770265e-05,
+      "loss": 2.1603,
+      "step": 1214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.517200474495848e-05,
+      "loss": 2.0003,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.516805061289047e-05,
+      "loss": 2.198,
+      "step": 1216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.516409648082246e-05,
+      "loss": 2.196,
+      "step": 1217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5160142348754455e-05,
+      "loss": 2.365,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.515618821668644e-05,
+      "loss": 2.171,
+      "step": 1219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5152234084618426e-05,
+      "loss": 2.0493,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5148279952550415e-05,
+      "loss": 2.1779,
+      "step": 1221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.514432582048241e-05,
+      "loss": 2.4187,
+      "step": 1222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.514037168841439e-05,
+      "loss": 2.1074,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.513641755634638e-05,
+      "loss": 2.187,
+      "step": 1224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.513246342427837e-05,
+      "loss": 2.2908,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5128509292210366e-05,
+      "loss": 2.4962,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.512455516014235e-05,
+      "loss": 2.25,
+      "step": 1227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.512060102807434e-05,
+      "loss": 2.1648,
+      "step": 1228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.511664689600633e-05,
+      "loss": 2.3187,
+      "step": 1229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.511269276393832e-05,
+      "loss": 2.1187,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.51087386318703e-05,
+      "loss": 2.2281,
+      "step": 1231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.510478449980229e-05,
+      "loss": 2.0311,
+      "step": 1232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.510083036773429e-05,
+      "loss": 2.0738,
+      "step": 1233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5096876235666277e-05,
+      "loss": 2.1976,
+      "step": 1234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.509292210359826e-05,
+      "loss": 2.446,
+      "step": 1235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.508896797153025e-05,
+      "loss": 2.3525,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.508501383946224e-05,
+      "loss": 2.0492,
+      "step": 1237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.508105970739423e-05,
+      "loss": 2.2403,
+      "step": 1238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5077105575326214e-05,
+      "loss": 2.2154,
+      "step": 1239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.50731514432582e-05,
+      "loss": 2.1795,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.50691973111902e-05,
+      "loss": 2.3464,
+      "step": 1241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.506524317912219e-05,
+      "loss": 2.1684,
+      "step": 1242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5061289047054176e-05,
+      "loss": 2.2342,
+      "step": 1243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.505733491498616e-05,
+      "loss": 2.2634,
+      "step": 1244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5053380782918154e-05,
+      "loss": 2.3619,
+      "step": 1245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.504942665085014e-05,
+      "loss": 2.2816,
+      "step": 1246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.504547251878213e-05,
+      "loss": 2.0771,
+      "step": 1247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.504151838671412e-05,
+      "loss": 2.1559,
+      "step": 1248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.503756425464611e-05,
+      "loss": 2.2622,
+      "step": 1249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.50336101225781e-05,
+      "loss": 2.0443,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.502965599051009e-05,
+      "loss": 2.1727,
+      "step": 1251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5025701858442075e-05,
+      "loss": 2.2303,
+      "step": 1252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.5021747726374064e-05,
+      "loss": 2.3188,
+      "step": 1253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.501779359430605e-05,
+      "loss": 2.5526,
+      "step": 1254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.501383946223804e-05,
+      "loss": 2.1377,
+      "step": 1255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.500988533017003e-05,
+      "loss": 2.1755,
+      "step": 1256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.500593119810202e-05,
+      "loss": 2.1064,
+      "step": 1257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.500197706603401e-05,
+      "loss": 2.2141,
+      "step": 1258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4998022933966e-05,
+      "loss": 2.5257,
+      "step": 1259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4994068801897986e-05,
+      "loss": 2.3599,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4990114669829975e-05,
+      "loss": 2.2321,
+      "step": 1261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4986160537761964e-05,
+      "loss": 2.0408,
+      "step": 1262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.498220640569395e-05,
+      "loss": 2.036,
+      "step": 1263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.497825227362594e-05,
+      "loss": 2.4586,
+      "step": 1264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.497429814155793e-05,
+      "loss": 2.1358,
+      "step": 1265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.497034400948992e-05,
+      "loss": 2.3886,
+      "step": 1266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.496638987742191e-05,
+      "loss": 1.9919,
+      "step": 1267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.49624357453539e-05,
+      "loss": 2.1335,
+      "step": 1268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4958481613285886e-05,
+      "loss": 2.3784,
+      "step": 1269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4954527481217874e-05,
+      "loss": 2.0485,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.495057334914986e-05,
+      "loss": 2.0046,
+      "step": 1271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.494661921708185e-05,
+      "loss": 2.4236,
+      "step": 1272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.494266508501384e-05,
+      "loss": 2.2195,
+      "step": 1273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.493871095294583e-05,
+      "loss": 2.27,
+      "step": 1274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.493475682087782e-05,
+      "loss": 2.214,
+      "step": 1275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.493080268880981e-05,
+      "loss": 2.0861,
+      "step": 1276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4926848556741796e-05,
+      "loss": 2.337,
+      "step": 1277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4922894424673785e-05,
+      "loss": 2.2553,
+      "step": 1278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4918940292605774e-05,
+      "loss": 2.3796,
+      "step": 1279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.491498616053776e-05,
+      "loss": 2.3367,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.491103202846975e-05,
+      "loss": 2.0875,
+      "step": 1281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.490707789640175e-05,
+      "loss": 2.36,
+      "step": 1282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.490312376433373e-05,
+      "loss": 1.9607,
+      "step": 1283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.489916963226572e-05,
+      "loss": 2.1755,
+      "step": 1284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.489521550019771e-05,
+      "loss": 2.236,
+      "step": 1285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.48912613681297e-05,
+      "loss": 2.203,
+      "step": 1286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4887307236061684e-05,
+      "loss": 2.1029,
+      "step": 1287
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.488335310399367e-05,
+      "loss": 2.2851,
+      "step": 1288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.487939897192566e-05,
+      "loss": 2.1375,
+      "step": 1289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.487544483985766e-05,
+      "loss": 2.088,
+      "step": 1290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.487149070778964e-05,
+      "loss": 2.3434,
+      "step": 1291
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.486753657572163e-05,
+      "loss": 1.9806,
+      "step": 1292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.486358244365362e-05,
+      "loss": 2.1145,
+      "step": 1293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.485962831158561e-05,
+      "loss": 2.2877,
+      "step": 1294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4855674179517595e-05,
+      "loss": 2.2022,
+      "step": 1295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4851720047449584e-05,
+      "loss": 2.1753,
+      "step": 1296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.484776591538157e-05,
+      "loss": 2.3019,
+      "step": 1297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.484381178331357e-05,
+      "loss": 2.0785,
+      "step": 1298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.483985765124555e-05,
+      "loss": 2.2302,
+      "step": 1299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.483590351917754e-05,
+      "loss": 2.068,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4831949387109535e-05,
+      "loss": 2.1084,
+      "step": 1301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4827995255041524e-05,
+      "loss": 2.018,
+      "step": 1302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.482404112297351e-05,
+      "loss": 2.1119,
+      "step": 1303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4820086990905494e-05,
+      "loss": 2.06,
+      "step": 1304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.481613285883749e-05,
+      "loss": 2.2501,
+      "step": 1305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.481217872676948e-05,
+      "loss": 2.0779,
+      "step": 1306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.480822459470147e-05,
+      "loss": 2.18,
+      "step": 1307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.480427046263345e-05,
+      "loss": 2.351,
+      "step": 1308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4800316330565445e-05,
+      "loss": 2.35,
+      "step": 1309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4796362198497434e-05,
+      "loss": 1.9892,
+      "step": 1310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.479240806642942e-05,
+      "loss": 2.193,
+      "step": 1311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4788453934361405e-05,
+      "loss": 1.8503,
+      "step": 1312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.47844998022934e-05,
+      "loss": 2.1834,
+      "step": 1313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.478054567022539e-05,
+      "loss": 2.2147,
+      "step": 1314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.477659153815738e-05,
+      "loss": 2.2978,
+      "step": 1315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.477263740608937e-05,
+      "loss": 2.2963,
+      "step": 1316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4768683274021356e-05,
+      "loss": 2.1207,
+      "step": 1317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4764729141953345e-05,
+      "loss": 2.212,
+      "step": 1318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4760775009885334e-05,
+      "loss": 2.3257,
+      "step": 1319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.475682087781732e-05,
+      "loss": 2.2415,
+      "step": 1320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.475286674574931e-05,
+      "loss": 2.3268,
+      "step": 1321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.47489126136813e-05,
+      "loss": 1.994,
+      "step": 1322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.474495848161329e-05,
+      "loss": 2.0436,
+      "step": 1323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.474100434954528e-05,
+      "loss": 2.4008,
+      "step": 1324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4737050217477267e-05,
+      "loss": 2.0421,
+      "step": 1325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4733096085409255e-05,
+      "loss": 2.062,
+      "step": 1326
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.4729141953341244e-05,
+      "loss": 2.1815,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.472518782127323e-05,
+      "loss": 2.3056,
+      "step": 1328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.472123368920522e-05,
+      "loss": 2.2097,
+      "step": 1329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.471727955713721e-05,
+      "loss": 2.3,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.47133254250692e-05,
+      "loss": 1.922,
+      "step": 1331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.470937129300119e-05,
+      "loss": 2.3428,
+      "step": 1332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.470541716093318e-05,
+      "loss": 2.3569,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4701463028865166e-05,
+      "loss": 2.1095,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4697508896797155e-05,
+      "loss": 2.1855,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4693554764729144e-05,
+      "loss": 2.2776,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.468960063266113e-05,
+      "loss": 2.2373,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.468564650059312e-05,
+      "loss": 2.3693,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.468169236852511e-05,
+      "loss": 2.4697,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.46777382364571e-05,
+      "loss": 2.2203,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.467378410438909e-05,
+      "loss": 2.1917,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.466982997232108e-05,
+      "loss": 2.199,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4665875840253065e-05,
+      "loss": 2.1173,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4661921708185054e-05,
+      "loss": 2.0962,
+      "step": 1344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.465796757611704e-05,
+      "loss": 2.3233,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.465401344404903e-05,
+      "loss": 2.0192,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.465005931198102e-05,
+      "loss": 2.077,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.464610517991301e-05,
+      "loss": 2.2191,
+      "step": 1348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4642151047845e-05,
+      "loss": 2.1926,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.463819691577699e-05,
+      "loss": 2.1483,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4634242783708976e-05,
+      "loss": 2.2192,
+      "step": 1351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4630288651640965e-05,
+      "loss": 2.0813,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4626334519572954e-05,
+      "loss": 1.9013,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.462238038750495e-05,
+      "loss": 2.4048,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.461842625543693e-05,
+      "loss": 2.0394,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.461447212336892e-05,
+      "loss": 2.2436,
+      "step": 1356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.461051799130091e-05,
+      "loss": 1.9464,
+      "step": 1357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4606563859232905e-05,
+      "loss": 2.3875,
+      "step": 1358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4602609727164893e-05,
+      "loss": 2.2412,
+      "step": 1359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4598655595096876e-05,
+      "loss": 2.2194,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4594701463028864e-05,
+      "loss": 2.0825,
+      "step": 1361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.459074733096086e-05,
+      "loss": 2.2472,
+      "step": 1362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.458679319889285e-05,
+      "loss": 1.858,
+      "step": 1363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.458283906682483e-05,
+      "loss": 1.9738,
+      "step": 1364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.457888493475682e-05,
+      "loss": 2.0585,
+      "step": 1365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4574930802688815e-05,
+      "loss": 2.3671,
+      "step": 1366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4570976670620804e-05,
+      "loss": 2.1533,
+      "step": 1367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4567022538552786e-05,
+      "loss": 2.2508,
+      "step": 1368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.456306840648478e-05,
+      "loss": 2.085,
+      "step": 1369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.455911427441677e-05,
+      "loss": 2.4323,
+      "step": 1370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.455516014234876e-05,
+      "loss": 2.0644,
+      "step": 1371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.455120601028074e-05,
+      "loss": 2.2161,
+      "step": 1372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.454725187821274e-05,
+      "loss": 2.1114,
+      "step": 1373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4543297746144726e-05,
+      "loss": 2.1114,
+      "step": 1374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4539343614076715e-05,
+      "loss": 2.3101,
+      "step": 1375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.45353894820087e-05,
+      "loss": 2.0358,
+      "step": 1376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.453143534994069e-05,
+      "loss": 2.3582,
+      "step": 1377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.452748121787268e-05,
+      "loss": 1.9125,
+      "step": 1378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.452352708580467e-05,
+      "loss": 1.9791,
+      "step": 1379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.451957295373665e-05,
+      "loss": 2.1208,
+      "step": 1380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.451561882166865e-05,
+      "loss": 1.9733,
+      "step": 1381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4511664689600637e-05,
+      "loss": 2.1798,
+      "step": 1382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4507710557532625e-05,
+      "loss": 2.2475,
+      "step": 1383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.450375642546461e-05,
+      "loss": 2.5376,
+      "step": 1384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.44998022933966e-05,
+      "loss": 2.1835,
+      "step": 1385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.449584816132859e-05,
+      "loss": 2.1413,
+      "step": 1386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.449189402926058e-05,
+      "loss": 2.0173,
+      "step": 1387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.448793989719257e-05,
+      "loss": 2.0264,
+      "step": 1388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.448398576512456e-05,
+      "loss": 2.3711,
+      "step": 1389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.448003163305655e-05,
+      "loss": 2.0065,
+      "step": 1390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4476077500988536e-05,
+      "loss": 2.1365,
+      "step": 1391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4472123368920525e-05,
+      "loss": 1.9476,
+      "step": 1392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4468169236852514e-05,
+      "loss": 2.1073,
+      "step": 1393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.44642151047845e-05,
+      "loss": 2.0135,
+      "step": 1394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.446026097271649e-05,
+      "loss": 2.4029,
+      "step": 1395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.445630684064848e-05,
+      "loss": 2.1664,
+      "step": 1396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.445235270858047e-05,
+      "loss": 2.2722,
+      "step": 1397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.444839857651246e-05,
+      "loss": 2.394,
+      "step": 1398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 2.0491,
+      "step": 1399
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4440490312376435e-05,
+      "loss": 2.0879,
+      "step": 1400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4436536180308424e-05,
+      "loss": 2.2667,
+      "step": 1401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.443258204824041e-05,
+      "loss": 2.1445,
+      "step": 1402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.44286279161724e-05,
+      "loss": 2.2594,
+      "step": 1403
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.442467378410439e-05,
+      "loss": 2.177,
+      "step": 1404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.442071965203638e-05,
+      "loss": 2.2965,
+      "step": 1405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.441676551996837e-05,
+      "loss": 2.1615,
+      "step": 1406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.441281138790036e-05,
+      "loss": 2.2058,
+      "step": 1407
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4408857255832346e-05,
+      "loss": 2.0742,
+      "step": 1408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4404903123764335e-05,
+      "loss": 2.4269,
+      "step": 1409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4400948991696324e-05,
+      "loss": 2.428,
+      "step": 1410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.439699485962831e-05,
+      "loss": 2.0625,
+      "step": 1411
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.43930407275603e-05,
+      "loss": 2.1132,
+      "step": 1412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.438908659549229e-05,
+      "loss": 2.3008,
+      "step": 1413
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.438513246342428e-05,
+      "loss": 2.1842,
+      "step": 1414
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4381178331356275e-05,
+      "loss": 2.2771,
+      "step": 1415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.437722419928826e-05,
+      "loss": 2.4283,
+      "step": 1416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4373270067220245e-05,
+      "loss": 2.3819,
+      "step": 1417
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4369315935152234e-05,
+      "loss": 2.1795,
+      "step": 1418
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.436536180308423e-05,
+      "loss": 2.2099,
+      "step": 1419
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.436140767101621e-05,
+      "loss": 2.2741,
+      "step": 1420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.43574535389482e-05,
+      "loss": 2.1715,
+      "step": 1421
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4353499406880196e-05,
+      "loss": 1.9302,
+      "step": 1422
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4349545274812185e-05,
+      "loss": 2.2187,
+      "step": 1423
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.434559114274417e-05,
+      "loss": 1.9751,
+      "step": 1424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4341637010676156e-05,
+      "loss": 2.2672,
+      "step": 1425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.433768287860815e-05,
+      "loss": 2.3643,
+      "step": 1426
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.433372874654014e-05,
+      "loss": 2.1488,
+      "step": 1427
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.432977461447212e-05,
+      "loss": 1.9532,
+      "step": 1428
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.432582048240411e-05,
+      "loss": 2.2731,
+      "step": 1429
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.432186635033611e-05,
+      "loss": 2.2961,
+      "step": 1430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4317912218268096e-05,
+      "loss": 2.1685,
+      "step": 1431
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.431395808620008e-05,
+      "loss": 2.0936,
+      "step": 1432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.431000395413207e-05,
+      "loss": 2.0113,
+      "step": 1433
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.430604982206406e-05,
+      "loss": 1.9374,
+      "step": 1434
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.430209568999605e-05,
+      "loss": 2.1987,
+      "step": 1435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.429814155792803e-05,
+      "loss": 2.4427,
+      "step": 1436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.429418742586002e-05,
+      "loss": 2.0103,
+      "step": 1437
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.429023329379202e-05,
+      "loss": 2.3633,
+      "step": 1438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4286279161724006e-05,
+      "loss": 1.8683,
+      "step": 1439
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.428232502965599e-05,
+      "loss": 2.2151,
+      "step": 1440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4278370897587984e-05,
+      "loss": 2.1223,
+      "step": 1441
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.427441676551997e-05,
+      "loss": 2.0308,
+      "step": 1442
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.427046263345196e-05,
+      "loss": 2.1789,
+      "step": 1443
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4266508501383944e-05,
+      "loss": 1.9131,
+      "step": 1444
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.426255436931594e-05,
+      "loss": 2.2168,
+      "step": 1445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.425860023724793e-05,
+      "loss": 2.227,
+      "step": 1446
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.425464610517992e-05,
+      "loss": 2.3303,
+      "step": 1447
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.42506919731119e-05,
+      "loss": 2.0949,
+      "step": 1448
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4246737841043895e-05,
+      "loss": 2.2948,
+      "step": 1449
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4242783708975884e-05,
+      "loss": 1.9729,
+      "step": 1450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.423882957690787e-05,
+      "loss": 2.3956,
+      "step": 1451
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.4234875444839854e-05,
+      "loss": 2.1883,
+      "step": 1452
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.423092131277185e-05,
+      "loss": 1.9656,
+      "step": 1453
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.422696718070384e-05,
+      "loss": 2.3976,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.422301304863583e-05,
+      "loss": 1.8958,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4219058916567816e-05,
+      "loss": 2.2127,
+      "step": 1456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4215104784499805e-05,
+      "loss": 1.8343,
+      "step": 1457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4211150652431794e-05,
+      "loss": 2.2078,
+      "step": 1458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.420719652036378e-05,
+      "loss": 2.2208,
+      "step": 1459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.420324238829577e-05,
+      "loss": 2.1417,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.419928825622776e-05,
+      "loss": 2.1621,
+      "step": 1461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.419533412415975e-05,
+      "loss": 2.1199,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.419137999209174e-05,
+      "loss": 2.1356,
+      "step": 1463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.418742586002373e-05,
+      "loss": 2.0241,
+      "step": 1464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4183471727955716e-05,
+      "loss": 2.0592,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4179517595887705e-05,
+      "loss": 2.2197,
+      "step": 1466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4175563463819694e-05,
+      "loss": 2.1358,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.417160933175168e-05,
+      "loss": 2.0332,
+      "step": 1468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.416765519968367e-05,
+      "loss": 2.1192,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.416370106761566e-05,
+      "loss": 1.9823,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.415974693554765e-05,
+      "loss": 2.1601,
+      "step": 1471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.415579280347964e-05,
+      "loss": 2.3774,
+      "step": 1472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4151838671411627e-05,
+      "loss": 2.1106,
+      "step": 1473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4147884539343615e-05,
+      "loss": 2.1882,
+      "step": 1474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.414393040727561e-05,
+      "loss": 2.187,
+      "step": 1475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.413997627520759e-05,
+      "loss": 2.2756,
+      "step": 1476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.413602214313958e-05,
+      "loss": 1.8966,
+      "step": 1477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.413206801107157e-05,
+      "loss": 2.0899,
+      "step": 1478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4128113879003566e-05,
+      "loss": 1.8784,
+      "step": 1479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.412415974693555e-05,
+      "loss": 1.9422,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.412020561486754e-05,
+      "loss": 1.9136,
+      "step": 1481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4116251482799526e-05,
+      "loss": 2.0978,
+      "step": 1482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.411229735073152e-05,
+      "loss": 1.8018,
+      "step": 1483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4108343218663504e-05,
+      "loss": 2.3174,
+      "step": 1484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.410438908659549e-05,
+      "loss": 2.2407,
+      "step": 1485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.410043495452748e-05,
+      "loss": 1.857,
+      "step": 1486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.409648082245948e-05,
+      "loss": 1.9135,
+      "step": 1487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.409252669039146e-05,
+      "loss": 2.1642,
+      "step": 1488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.408857255832345e-05,
+      "loss": 2.3458,
+      "step": 1489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.408461842625544e-05,
+      "loss": 2.1078,
+      "step": 1490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.408066429418743e-05,
+      "loss": 2.0191,
+      "step": 1491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4076710162119414e-05,
+      "loss": 2.3301,
+      "step": 1492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.40727560300514e-05,
+      "loss": 2.3294,
+      "step": 1493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.40688018979834e-05,
+      "loss": 2.0827,
+      "step": 1494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.406484776591539e-05,
+      "loss": 2.1211,
+      "step": 1495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.406089363384737e-05,
+      "loss": 2.2364,
+      "step": 1496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.405693950177936e-05,
+      "loss": 2.0262,
+      "step": 1497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4052985369711354e-05,
+      "loss": 2.1885,
+      "step": 1498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.404903123764334e-05,
+      "loss": 2.2605,
+      "step": 1499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4045077105575325e-05,
+      "loss": 2.2175,
+      "step": 1500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4041122973507314e-05,
+      "loss": 2.1377,
+      "step": 1501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.403716884143931e-05,
+      "loss": 2.39,
+      "step": 1502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.40332147093713e-05,
+      "loss": 2.0585,
+      "step": 1503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.402926057730328e-05,
+      "loss": 2.2628,
+      "step": 1504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.402530644523527e-05,
+      "loss": 2.1881,
+      "step": 1505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4021352313167265e-05,
+      "loss": 2.285,
+      "step": 1506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4017398181099253e-05,
+      "loss": 1.8459,
+      "step": 1507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.4013444049031235e-05,
+      "loss": 2.11,
+      "step": 1508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.400948991696323e-05,
+      "loss": 1.9804,
+      "step": 1509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.400553578489522e-05,
+      "loss": 2.0626,
+      "step": 1510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.400158165282721e-05,
+      "loss": 2.0286,
+      "step": 1511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.399762752075919e-05,
+      "loss": 2.0115,
+      "step": 1512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3993673388691186e-05,
+      "loss": 2.1767,
+      "step": 1513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3989719256623175e-05,
+      "loss": 2.2913,
+      "step": 1514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3985765124555164e-05,
+      "loss": 2.2743,
+      "step": 1515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3981810992487146e-05,
+      "loss": 2.1195,
+      "step": 1516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.397785686041914e-05,
+      "loss": 2.4481,
+      "step": 1517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.397390272835113e-05,
+      "loss": 2.6538,
+      "step": 1518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.396994859628312e-05,
+      "loss": 2.2032,
+      "step": 1519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.39659944642151e-05,
+      "loss": 2.1558,
+      "step": 1520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.39620403321471e-05,
+      "loss": 2.0427,
+      "step": 1521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3958086200079086e-05,
+      "loss": 2.2106,
+      "step": 1522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3954132068011075e-05,
+      "loss": 2.1449,
+      "step": 1523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.395017793594306e-05,
+      "loss": 2.1856,
+      "step": 1524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.394622380387505e-05,
+      "loss": 2.3066,
+      "step": 1525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.394226967180704e-05,
+      "loss": 2.1884,
+      "step": 1526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.393831553973903e-05,
+      "loss": 1.8621,
+      "step": 1527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.393436140767102e-05,
+      "loss": 2.2622,
+      "step": 1528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.393040727560301e-05,
+      "loss": 2.2888,
+      "step": 1529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3926453143534996e-05,
+      "loss": 2.0579,
+      "step": 1530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3922499011466985e-05,
+      "loss": 2.2683,
+      "step": 1531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3918544879398974e-05,
+      "loss": 1.9688,
+      "step": 1532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.391459074733096e-05,
+      "loss": 2.1428,
+      "step": 1533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.391063661526295e-05,
+      "loss": 2.2792,
+      "step": 1534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.390668248319494e-05,
+      "loss": 2.0753,
+      "step": 1535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.390272835112693e-05,
+      "loss": 2.1292,
+      "step": 1536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.389877421905892e-05,
+      "loss": 2.0603,
+      "step": 1537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.389482008699091e-05,
+      "loss": 2.0828,
+      "step": 1538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3890865954922896e-05,
+      "loss": 2.1434,
+      "step": 1539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3886911822854885e-05,
+      "loss": 2.1945,
+      "step": 1540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3882957690786874e-05,
+      "loss": 2.255,
+      "step": 1541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.387900355871886e-05,
+      "loss": 2.3203,
+      "step": 1542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.387504942665086e-05,
+      "loss": 2.1962,
+      "step": 1543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.387109529458284e-05,
+      "loss": 2.1734,
+      "step": 1544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.386714116251483e-05,
+      "loss": 2.2098,
+      "step": 1545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.386318703044682e-05,
+      "loss": 2.1812,
+      "step": 1546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.385923289837881e-05,
+      "loss": 1.9974,
+      "step": 1547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3855278766310795e-05,
+      "loss": 2.005,
+      "step": 1548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3851324634242784e-05,
+      "loss": 2.3141,
+      "step": 1549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.384737050217477e-05,
+      "loss": 1.9977,
+      "step": 1550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.384341637010677e-05,
+      "loss": 2.2682,
+      "step": 1551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.383946223803875e-05,
+      "loss": 2.3279,
+      "step": 1552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.383550810597074e-05,
+      "loss": 2.0293,
+      "step": 1553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.383155397390273e-05,
+      "loss": 2.0688,
+      "step": 1554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3827599841834724e-05,
+      "loss": 1.8651,
+      "step": 1555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3823645709766706e-05,
+      "loss": 2.2969,
+      "step": 1556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3819691577698695e-05,
+      "loss": 2.1789,
+      "step": 1557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3815737445630684e-05,
+      "loss": 2.2299,
+      "step": 1558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.381178331356268e-05,
+      "loss": 2.0905,
+      "step": 1559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.380782918149466e-05,
+      "loss": 2.5117,
+      "step": 1560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.380387504942665e-05,
+      "loss": 2.3192,
+      "step": 1561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3799920917358646e-05,
+      "loss": 2.0178,
+      "step": 1562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3795966785290635e-05,
+      "loss": 1.9842,
+      "step": 1563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3792012653222617e-05,
+      "loss": 2.1927,
+      "step": 1564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3788058521154605e-05,
+      "loss": 2.2257,
+      "step": 1565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.37841043890866e-05,
+      "loss": 2.187,
+      "step": 1566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.378015025701859e-05,
+      "loss": 1.908,
+      "step": 1567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.377619612495057e-05,
+      "loss": 2.1532,
+      "step": 1568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.377224199288256e-05,
+      "loss": 2.2342,
+      "step": 1569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3768287860814556e-05,
+      "loss": 2.2636,
+      "step": 1570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3764333728746545e-05,
+      "loss": 2.0436,
+      "step": 1571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.376037959667853e-05,
+      "loss": 2.15,
+      "step": 1572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3756425464610516e-05,
+      "loss": 2.11,
+      "step": 1573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.375247133254251e-05,
+      "loss": 1.8604,
+      "step": 1574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.37485172004745e-05,
+      "loss": 2.3172,
+      "step": 1575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.374456306840648e-05,
+      "loss": 2.167,
+      "step": 1576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.374060893633848e-05,
+      "loss": 1.9465,
+      "step": 1577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.373665480427047e-05,
+      "loss": 2.1229,
+      "step": 1578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.3732700672202456e-05,
+      "loss": 2.091,
+      "step": 1579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.372874654013444e-05,
+      "loss": 2.1301,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3724792408066433e-05,
+      "loss": 2.219,
+      "step": 1581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.372083827599842e-05,
+      "loss": 2.2511,
+      "step": 1582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.371688414393041e-05,
+      "loss": 2.5377,
+      "step": 1583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.371293001186239e-05,
+      "loss": 2.1677,
+      "step": 1584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.370897587979439e-05,
+      "loss": 2.0943,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.370502174772638e-05,
+      "loss": 2.0927,
+      "step": 1586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3701067615658366e-05,
+      "loss": 2.3086,
+      "step": 1587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.369711348359035e-05,
+      "loss": 2.4628,
+      "step": 1588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3693159351522344e-05,
+      "loss": 2.1136,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.368920521945433e-05,
+      "loss": 2.138,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.368525108738632e-05,
+      "loss": 2.2154,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.368129695531831e-05,
+      "loss": 2.0531,
+      "step": 1592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.36773428232503e-05,
+      "loss": 2.0958,
+      "step": 1593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.367338869118229e-05,
+      "loss": 2.0811,
+      "step": 1594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.366943455911428e-05,
+      "loss": 2.2051,
+      "step": 1595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3665480427046266e-05,
+      "loss": 2.3428,
+      "step": 1596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3661526294978255e-05,
+      "loss": 2.2545,
+      "step": 1597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3657572162910243e-05,
+      "loss": 2.2039,
+      "step": 1598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.365361803084223e-05,
+      "loss": 2.1881,
+      "step": 1599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.364966389877422e-05,
+      "loss": 2.2055,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.364570976670621e-05,
+      "loss": 2.318,
+      "step": 1601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.36417556346382e-05,
+      "loss": 2.3909,
+      "step": 1602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.363780150257019e-05,
+      "loss": 2.2674,
+      "step": 1603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3633847370502176e-05,
+      "loss": 2.25,
+      "step": 1604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3629893238434165e-05,
+      "loss": 2.1345,
+      "step": 1605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3625939106366154e-05,
+      "loss": 2.2288,
+      "step": 1606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.362198497429814e-05,
+      "loss": 2.1755,
+      "step": 1607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.361803084223013e-05,
+      "loss": 2.3877,
+      "step": 1608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.361407671016212e-05,
+      "loss": 2.1462,
+      "step": 1609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.361012257809411e-05,
+      "loss": 2.0735,
+      "step": 1610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.36061684460261e-05,
+      "loss": 2.0082,
+      "step": 1611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.360221431395809e-05,
+      "loss": 2.2851,
+      "step": 1612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3598260181890076e-05,
+      "loss": 2.3463,
+      "step": 1613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3594306049822065e-05,
+      "loss": 2.2775,
+      "step": 1614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.359035191775406e-05,
+      "loss": 2.2074,
+      "step": 1615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.358639778568604e-05,
+      "loss": 2.2373,
+      "step": 1616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.358244365361803e-05,
+      "loss": 2.2376,
+      "step": 1617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.357848952155002e-05,
+      "loss": 2.3649,
+      "step": 1618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3574535389482016e-05,
+      "loss": 2.0919,
+      "step": 1619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3570581257414e-05,
+      "loss": 2.1629,
+      "step": 1620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3566627125345986e-05,
+      "loss": 1.7348,
+      "step": 1621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3562672993277975e-05,
+      "loss": 2.0295,
+      "step": 1622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.355871886120997e-05,
+      "loss": 2.033,
+      "step": 1623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.355476472914195e-05,
+      "loss": 2.0845,
+      "step": 1624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.355081059707394e-05,
+      "loss": 2.125,
+      "step": 1625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.354685646500593e-05,
+      "loss": 2.166,
+      "step": 1626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3542902332937926e-05,
+      "loss": 2.1144,
+      "step": 1627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.353894820086991e-05,
+      "loss": 2.125,
+      "step": 1628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.35349940688019e-05,
+      "loss": 2.0419,
+      "step": 1629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.353103993673389e-05,
+      "loss": 2.2202,
+      "step": 1630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.352708580466588e-05,
+      "loss": 2.1195,
+      "step": 1631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3523131672597864e-05,
+      "loss": 2.2703,
+      "step": 1632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.351917754052985e-05,
+      "loss": 2.0937,
+      "step": 1633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.351522340846185e-05,
+      "loss": 2.0439,
+      "step": 1634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.351126927639384e-05,
+      "loss": 2.1674,
+      "step": 1635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.350731514432582e-05,
+      "loss": 2.1281,
+      "step": 1636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.350336101225781e-05,
+      "loss": 2.1389,
+      "step": 1637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.34994068801898e-05,
+      "loss": 2.2708,
+      "step": 1638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.349545274812179e-05,
+      "loss": 1.7271,
+      "step": 1639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3491498616053774e-05,
+      "loss": 2.4766,
+      "step": 1640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.348754448398576e-05,
+      "loss": 2.2326,
+      "step": 1641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.348359035191776e-05,
+      "loss": 2.3979,
+      "step": 1642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.347963621984975e-05,
+      "loss": 2.3143,
+      "step": 1643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.347568208778173e-05,
+      "loss": 2.0402,
+      "step": 1644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.347172795571372e-05,
+      "loss": 2.1407,
+      "step": 1645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3467773823645714e-05,
+      "loss": 2.1095,
+      "step": 1646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.34638196915777e-05,
+      "loss": 2.174,
+      "step": 1647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3459865559509685e-05,
+      "loss": 2.1258,
+      "step": 1648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.345591142744168e-05,
+      "loss": 2.0558,
+      "step": 1649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.345195729537367e-05,
+      "loss": 2.0404,
+      "step": 1650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.344800316330566e-05,
+      "loss": 2.2392,
+      "step": 1651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.344404903123765e-05,
+      "loss": 2.0492,
+      "step": 1652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3440094899169636e-05,
+      "loss": 2.1681,
+      "step": 1653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3436140767101625e-05,
+      "loss": 2.2431,
+      "step": 1654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.343218663503361e-05,
+      "loss": 2.143,
+      "step": 1655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.34282325029656e-05,
+      "loss": 2.0377,
+      "step": 1656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.342427837089759e-05,
+      "loss": 2.0701,
+      "step": 1657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.342032423882958e-05,
+      "loss": 2.1195,
+      "step": 1658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.341637010676157e-05,
+      "loss": 2.1215,
+      "step": 1659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.341241597469356e-05,
+      "loss": 2.3479,
+      "step": 1660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3408461842625546e-05,
+      "loss": 2.0574,
+      "step": 1661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3404507710557535e-05,
+      "loss": 2.1238,
+      "step": 1662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3400553578489524e-05,
+      "loss": 1.9924,
+      "step": 1663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.339659944642151e-05,
+      "loss": 2.3687,
+      "step": 1664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.33926453143535e-05,
+      "loss": 2.0518,
+      "step": 1665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.338869118228549e-05,
+      "loss": 2.3209,
+      "step": 1666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.338473705021748e-05,
+      "loss": 1.9383,
+      "step": 1667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.338078291814947e-05,
+      "loss": 1.934,
+      "step": 1668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.337682878608146e-05,
+      "loss": 2.3365,
+      "step": 1669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3372874654013446e-05,
+      "loss": 1.9499,
+      "step": 1670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3368920521945435e-05,
+      "loss": 2.1684,
+      "step": 1671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3364966389877423e-05,
+      "loss": 2.2368,
+      "step": 1672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.336101225780941e-05,
+      "loss": 1.9998,
+      "step": 1673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.33570581257414e-05,
+      "loss": 1.8317,
+      "step": 1674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.335310399367339e-05,
+      "loss": 1.9506,
+      "step": 1675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.334914986160538e-05,
+      "loss": 2.2872,
+      "step": 1676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.334519572953737e-05,
+      "loss": 2.0771,
+      "step": 1677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3341241597469356e-05,
+      "loss": 2.2371,
+      "step": 1678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3337287465401345e-05,
+      "loss": 2.0835,
+      "step": 1679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 2.1059,
+      "step": 1680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.332937920126532e-05,
+      "loss": 2.2258,
+      "step": 1681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.332542506919731e-05,
+      "loss": 2.0482,
+      "step": 1682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.332147093712931e-05,
+      "loss": 2.2625,
+      "step": 1683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.331751680506129e-05,
+      "loss": 2.0233,
+      "step": 1684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.331356267299328e-05,
+      "loss": 2.3531,
+      "step": 1685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.330960854092527e-05,
+      "loss": 2.0847,
+      "step": 1686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.330565440885726e-05,
+      "loss": 2.145,
+      "step": 1687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3301700276789245e-05,
+      "loss": 2.2162,
+      "step": 1688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3297746144721233e-05,
+      "loss": 2.2096,
+      "step": 1689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.329379201265322e-05,
+      "loss": 2.1124,
+      "step": 1690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.328983788058522e-05,
+      "loss": 2.4278,
+      "step": 1691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.32858837485172e-05,
+      "loss": 2.2657,
+      "step": 1692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.328192961644919e-05,
+      "loss": 2.2301,
+      "step": 1693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.327797548438118e-05,
+      "loss": 1.9666,
+      "step": 1694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.327402135231317e-05,
+      "loss": 2.0297,
+      "step": 1695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3270067220245155e-05,
+      "loss": 1.9722,
+      "step": 1696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3266113088177144e-05,
+      "loss": 2.2915,
+      "step": 1697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.326215895610913e-05,
+      "loss": 2.3981,
+      "step": 1698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.325820482404113e-05,
+      "loss": 1.9275,
+      "step": 1699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.325425069197311e-05,
+      "loss": 2.0431,
+      "step": 1700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.32502965599051e-05,
+      "loss": 2.1256,
+      "step": 1701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3246342427837095e-05,
+      "loss": 2.113,
+      "step": 1702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3242388295769084e-05,
+      "loss": 2.3578,
+      "step": 1703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3238434163701066e-05,
+      "loss": 2.2524,
+      "step": 1704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3234480031633055e-05,
+      "loss": 2.0629,
+      "step": 1705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.323052589956505e-05,
+      "loss": 1.9928,
+      "step": 1706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.322657176749704e-05,
+      "loss": 2.2554,
+      "step": 1707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.322261763542903e-05,
+      "loss": 2.3808,
+      "step": 1708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.321866350336101e-05,
+      "loss": 2.1349,
+      "step": 1709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3214709371293006e-05,
+      "loss": 2.2019,
+      "step": 1710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3210755239224994e-05,
+      "loss": 1.8744,
+      "step": 1711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.320680110715698e-05,
+      "loss": 2.107,
+      "step": 1712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3202846975088965e-05,
+      "loss": 2.2414,
+      "step": 1713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.319889284302096e-05,
+      "loss": 2.1061,
+      "step": 1714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.319493871095295e-05,
+      "loss": 2.1497,
+      "step": 1715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.319098457888494e-05,
+      "loss": 2.3003,
+      "step": 1716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.318703044681693e-05,
+      "loss": 2.1104,
+      "step": 1717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3183076314748916e-05,
+      "loss": 2.0604,
+      "step": 1718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3179122182680905e-05,
+      "loss": 2.3357,
+      "step": 1719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3175168050612894e-05,
+      "loss": 2.2946,
+      "step": 1720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.317121391854488e-05,
+      "loss": 2.2225,
+      "step": 1721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.316725978647687e-05,
+      "loss": 2.0082,
+      "step": 1722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.316330565440886e-05,
+      "loss": 1.9356,
+      "step": 1723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.315935152234085e-05,
+      "loss": 2.009,
+      "step": 1724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.315539739027284e-05,
+      "loss": 2.0947,
+      "step": 1725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.315144325820483e-05,
+      "loss": 2.0442,
+      "step": 1726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3147489126136816e-05,
+      "loss": 2.1692,
+      "step": 1727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3143534994068805e-05,
+      "loss": 1.964,
+      "step": 1728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.313958086200079e-05,
+      "loss": 2.0454,
+      "step": 1729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.313562672993278e-05,
+      "loss": 1.8776,
+      "step": 1730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.313167259786477e-05,
+      "loss": 2.0128,
+      "step": 1731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.312771846579676e-05,
+      "loss": 2.1905,
+      "step": 1732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.312376433372875e-05,
+      "loss": 2.4289,
+      "step": 1733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.311981020166074e-05,
+      "loss": 2.0839,
+      "step": 1734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3115856069592726e-05,
+      "loss": 1.9966,
+      "step": 1735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3111901937524715e-05,
+      "loss": 1.9929,
+      "step": 1736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3107947805456704e-05,
+      "loss": 2.2715,
+      "step": 1737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.310399367338869e-05,
+      "loss": 2.2792,
+      "step": 1738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.310003954132068e-05,
+      "loss": 2.3055,
+      "step": 1739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.309608540925267e-05,
+      "loss": 2.1473,
+      "step": 1740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.309213127718466e-05,
+      "loss": 2.0921,
+      "step": 1741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.308817714511665e-05,
+      "loss": 2.1256,
+      "step": 1742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.308422301304864e-05,
+      "loss": 2.3213,
+      "step": 1743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3080268880980626e-05,
+      "loss": 1.8392,
+      "step": 1744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3076314748912615e-05,
+      "loss": 2.0336,
+      "step": 1745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3072360616844603e-05,
+      "loss": 2.1083,
+      "step": 1746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.306840648477659e-05,
+      "loss": 2.1919,
+      "step": 1747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.306445235270858e-05,
+      "loss": 2.1928,
+      "step": 1748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.306049822064057e-05,
+      "loss": 2.2648,
+      "step": 1749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.305654408857256e-05,
+      "loss": 2.1941,
+      "step": 1750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3052589956504554e-05,
+      "loss": 2.2538,
+      "step": 1751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3048635824436536e-05,
+      "loss": 2.2059,
+      "step": 1752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3044681692368525e-05,
+      "loss": 2.0649,
+      "step": 1753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3040727560300514e-05,
+      "loss": 2.0816,
+      "step": 1754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.303677342823251e-05,
+      "loss": 2.2555,
+      "step": 1755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.303281929616449e-05,
+      "loss": 2.2658,
+      "step": 1756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.302886516409648e-05,
+      "loss": 2.1823,
+      "step": 1757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.302491103202847e-05,
+      "loss": 2.8293,
+      "step": 1758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3020956899960465e-05,
+      "loss": 2.1987,
+      "step": 1759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.301700276789245e-05,
+      "loss": 1.9684,
+      "step": 1760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3013048635824436e-05,
+      "loss": 2.1986,
+      "step": 1761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.3009094503756425e-05,
+      "loss": 2.0434,
+      "step": 1762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.300514037168842e-05,
+      "loss": 2.6419,
+      "step": 1763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.300118623962041e-05,
+      "loss": 2.3615,
+      "step": 1764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.299723210755239e-05,
+      "loss": 1.9672,
+      "step": 1765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.299327797548438e-05,
+      "loss": 2.317,
+      "step": 1766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2989323843416376e-05,
+      "loss": 2.2102,
+      "step": 1767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2985369711348364e-05,
+      "loss": 2.0698,
+      "step": 1768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2981415579280346e-05,
+      "loss": 2.2829,
+      "step": 1769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.297746144721234e-05,
+      "loss": 2.0363,
+      "step": 1770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.297350731514433e-05,
+      "loss": 2.0153,
+      "step": 1771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.296955318307632e-05,
+      "loss": 2.0366,
+      "step": 1772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.29655990510083e-05,
+      "loss": 2.1873,
+      "step": 1773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.29616449189403e-05,
+      "loss": 2.0183,
+      "step": 1774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2957690786872286e-05,
+      "loss": 2.0846,
+      "step": 1775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2953736654804275e-05,
+      "loss": 2.0159,
+      "step": 1776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.294978252273626e-05,
+      "loss": 1.9823,
+      "step": 1777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.294582839066825e-05,
+      "loss": 2.1655,
+      "step": 1778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.294187425860024e-05,
+      "loss": 2.2247,
+      "step": 1779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.293792012653223e-05,
+      "loss": 2.1384,
+      "step": 1780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.293396599446421e-05,
+      "loss": 2.1974,
+      "step": 1781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.293001186239621e-05,
+      "loss": 2.0486,
+      "step": 1782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.29260577303282e-05,
+      "loss": 2.206,
+      "step": 1783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2922103598260186e-05,
+      "loss": 2.2111,
+      "step": 1784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.291814946619217e-05,
+      "loss": 2.3166,
+      "step": 1785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.291419533412416e-05,
+      "loss": 2.0246,
+      "step": 1786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.291024120205615e-05,
+      "loss": 1.8454,
+      "step": 1787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.290628706998814e-05,
+      "loss": 2.0989,
+      "step": 1788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.290233293792013e-05,
+      "loss": 1.8266,
+      "step": 1789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.289837880585212e-05,
+      "loss": 2.1982,
+      "step": 1790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.289442467378411e-05,
+      "loss": 2.0001,
+      "step": 1791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2890470541716096e-05,
+      "loss": 2.1166,
+      "step": 1792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2886516409648085e-05,
+      "loss": 2.08,
+      "step": 1793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2882562277580074e-05,
+      "loss": 2.0147,
+      "step": 1794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.287860814551206e-05,
+      "loss": 2.1567,
+      "step": 1795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.287465401344405e-05,
+      "loss": 2.0851,
+      "step": 1796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.287069988137604e-05,
+      "loss": 2.1161,
+      "step": 1797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.286674574930803e-05,
+      "loss": 2.1182,
+      "step": 1798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.286279161724002e-05,
+      "loss": 2.09,
+      "step": 1799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.285883748517201e-05,
+      "loss": 1.9575,
+      "step": 1800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2854883353103996e-05,
+      "loss": 2.2757,
+      "step": 1801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2850929221035984e-05,
+      "loss": 2.0328,
+      "step": 1802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.284697508896797e-05,
+      "loss": 2.0582,
+      "step": 1803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.284302095689996e-05,
+      "loss": 2.0294,
+      "step": 1804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.283906682483195e-05,
+      "loss": 2.0037,
+      "step": 1805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.283511269276394e-05,
+      "loss": 2.1976,
+      "step": 1806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.283115856069593e-05,
+      "loss": 2.073,
+      "step": 1807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.282720442862792e-05,
+      "loss": 2.417,
+      "step": 1808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2823250296559906e-05,
+      "loss": 1.9054,
+      "step": 1809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2819296164491895e-05,
+      "loss": 2.2745,
+      "step": 1810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2815342032423884e-05,
+      "loss": 2.0952,
+      "step": 1811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.281138790035587e-05,
+      "loss": 2.2015,
+      "step": 1812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.280743376828786e-05,
+      "loss": 2.1251,
+      "step": 1813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.280347963621985e-05,
+      "loss": 2.2441,
+      "step": 1814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.279952550415184e-05,
+      "loss": 1.8957,
+      "step": 1815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.279557137208383e-05,
+      "loss": 2.499,
+      "step": 1816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.279161724001582e-05,
+      "loss": 2.4525,
+      "step": 1817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2787663107947806e-05,
+      "loss": 2.0305,
+      "step": 1818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2783708975879795e-05,
+      "loss": 2.3995,
+      "step": 1819
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.277975484381179e-05,
+      "loss": 1.8104,
+      "step": 1820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.277580071174377e-05,
+      "loss": 2.053,
+      "step": 1821
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.277184657967576e-05,
+      "loss": 2.267,
+      "step": 1822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.276789244760776e-05,
+      "loss": 2.1519,
+      "step": 1823
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2763938315539745e-05,
+      "loss": 2.3916,
+      "step": 1824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.275998418347173e-05,
+      "loss": 2.1309,
+      "step": 1825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2756030051403716e-05,
+      "loss": 2.3812,
+      "step": 1826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.275207591933571e-05,
+      "loss": 2.161,
+      "step": 1827
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.27481217872677e-05,
+      "loss": 2.153,
+      "step": 1828
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.274416765519968e-05,
+      "loss": 1.9564,
+      "step": 1829
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.274021352313167e-05,
+      "loss": 2.2886,
+      "step": 1830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.273625939106367e-05,
+      "loss": 2.315,
+      "step": 1831
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.2732305258995656e-05,
+      "loss": 2.0626,
+      "step": 1832
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.272835112692764e-05,
+      "loss": 2.4074,
+      "step": 1833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.272439699485963e-05,
+      "loss": 2.183,
+      "step": 1834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.272044286279162e-05,
+      "loss": 2.0696,
+      "step": 1835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.271648873072361e-05,
+      "loss": 1.9224,
+      "step": 1836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2712534598655593e-05,
+      "loss": 2.4179,
+      "step": 1837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.270858046658759e-05,
+      "loss": 1.88,
+      "step": 1838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.270462633451958e-05,
+      "loss": 2.2718,
+      "step": 1839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.270067220245157e-05,
+      "loss": 2.143,
+      "step": 1840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.269671807038355e-05,
+      "loss": 2.3855,
+      "step": 1841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2692763938315544e-05,
+      "loss": 2.242,
+      "step": 1842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.268880980624753e-05,
+      "loss": 2.2531,
+      "step": 1843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.268485567417952e-05,
+      "loss": 2.0172,
+      "step": 1844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2680901542111504e-05,
+      "loss": 2.1463,
+      "step": 1845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.26769474100435e-05,
+      "loss": 2.0646,
+      "step": 1846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.267299327797549e-05,
+      "loss": 2.3612,
+      "step": 1847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.266903914590748e-05,
+      "loss": 2.1146,
+      "step": 1848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.266508501383946e-05,
+      "loss": 2.0894,
+      "step": 1849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2661130881771455e-05,
+      "loss": 2.0503,
+      "step": 1850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2657176749703444e-05,
+      "loss": 2.338,
+      "step": 1851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.265322261763543e-05,
+      "loss": 2.1313,
+      "step": 1852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2649268485567415e-05,
+      "loss": 1.9769,
+      "step": 1853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.264531435349941e-05,
+      "loss": 2.1585,
+      "step": 1854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.26413602214314e-05,
+      "loss": 2.2034,
+      "step": 1855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.263740608936339e-05,
+      "loss": 2.0286,
+      "step": 1856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.263345195729538e-05,
+      "loss": 2.0795,
+      "step": 1857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2629497825227366e-05,
+      "loss": 2.3236,
+      "step": 1858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2625543693159354e-05,
+      "loss": 1.897,
+      "step": 1859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.262158956109134e-05,
+      "loss": 1.8858,
+      "step": 1860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.261763542902333e-05,
+      "loss": 2.4599,
+      "step": 1861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.261368129695532e-05,
+      "loss": 2.1038,
+      "step": 1862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.260972716488731e-05,
+      "loss": 2.06,
+      "step": 1863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.26057730328193e-05,
+      "loss": 2.0221,
+      "step": 1864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.260181890075129e-05,
+      "loss": 2.3268,
+      "step": 1865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2597864768683276e-05,
+      "loss": 1.8193,
+      "step": 1866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2593910636615265e-05,
+      "loss": 2.1635,
+      "step": 1867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2589956504547254e-05,
+      "loss": 1.9864,
+      "step": 1868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.258600237247924e-05,
+      "loss": 2.2399,
+      "step": 1869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.258204824041123e-05,
+      "loss": 2.287,
+      "step": 1870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.257809410834322e-05,
+      "loss": 1.8657,
+      "step": 1871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.257413997627521e-05,
+      "loss": 2.2445,
+      "step": 1872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.25701858442072e-05,
+      "loss": 2.0418,
+      "step": 1873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.256623171213919e-05,
+      "loss": 2.0711,
+      "step": 1874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2562277580071176e-05,
+      "loss": 2.3589,
+      "step": 1875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2558323448003164e-05,
+      "loss": 2.1833,
+      "step": 1876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.255436931593515e-05,
+      "loss": 2.2374,
+      "step": 1877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.255041518386714e-05,
+      "loss": 1.8367,
+      "step": 1878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.254646105179913e-05,
+      "loss": 2.0905,
+      "step": 1879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2542506919731127e-05,
+      "loss": 2.2095,
+      "step": 1880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.253855278766311e-05,
+      "loss": 1.9482,
+      "step": 1881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.25345986555951e-05,
+      "loss": 2.2436,
+      "step": 1882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2530644523527086e-05,
+      "loss": 2.3261,
+      "step": 1883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.252669039145908e-05,
+      "loss": 2.2497,
+      "step": 1884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2522736259391064e-05,
+      "loss": 2.3084,
+      "step": 1885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.251878212732305e-05,
+      "loss": 2.2561,
+      "step": 1886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.251482799525504e-05,
+      "loss": 2.019,
+      "step": 1887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.251087386318704e-05,
+      "loss": 2.4595,
+      "step": 1888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.250691973111902e-05,
+      "loss": 2.4242,
+      "step": 1889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.250296559905101e-05,
+      "loss": 2.2227,
+      "step": 1890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2499011466983004e-05,
+      "loss": 1.9639,
+      "step": 1891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.249505733491499e-05,
+      "loss": 2.1472,
+      "step": 1892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2491103202846975e-05,
+      "loss": 2.1813,
+      "step": 1893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.248714907077896e-05,
+      "loss": 2.0212,
+      "step": 1894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.248319493871096e-05,
+      "loss": 2.1489,
+      "step": 1895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.247924080664295e-05,
+      "loss": 2.2819,
+      "step": 1896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.247528667457493e-05,
+      "loss": 2.2,
+      "step": 1897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.247133254250692e-05,
+      "loss": 2.2124,
+      "step": 1898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2467378410438914e-05,
+      "loss": 2.104,
+      "step": 1899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.24634242783709e-05,
+      "loss": 2.0727,
+      "step": 1900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2459470146302885e-05,
+      "loss": 2.1374,
+      "step": 1901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2455516014234874e-05,
+      "loss": 2.144,
+      "step": 1902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.245156188216687e-05,
+      "loss": 2.3215,
+      "step": 1903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.244760775009886e-05,
+      "loss": 2.2443,
+      "step": 1904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.244365361803084e-05,
+      "loss": 2.1168,
+      "step": 1905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.243969948596283e-05,
+      "loss": 2.3657,
+      "step": 1906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2435745353894825e-05,
+      "loss": 1.9924,
+      "step": 1907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2431791221826814e-05,
+      "loss": 2.2515,
+      "step": 1908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2427837089758796e-05,
+      "loss": 2.0654,
+      "step": 1909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.242388295769079e-05,
+      "loss": 2.2788,
+      "step": 1910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.241992882562278e-05,
+      "loss": 2.1614,
+      "step": 1911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.241597469355477e-05,
+      "loss": 2.0865,
+      "step": 1912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.241202056148675e-05,
+      "loss": 2.1076,
+      "step": 1913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.240806642941875e-05,
+      "loss": 1.8694,
+      "step": 1914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2404112297350735e-05,
+      "loss": 2.0988,
+      "step": 1915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2400158165282724e-05,
+      "loss": 2.0025,
+      "step": 1916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2396204033214706e-05,
+      "loss": 2.2497,
+      "step": 1917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.23922499011467e-05,
+      "loss": 2.2924,
+      "step": 1918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.238829576907869e-05,
+      "loss": 2.1651,
+      "step": 1919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.238434163701068e-05,
+      "loss": 2.029,
+      "step": 1920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.238038750494266e-05,
+      "loss": 2.165,
+      "step": 1921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.237643337287466e-05,
+      "loss": 2.3529,
+      "step": 1922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2372479240806646e-05,
+      "loss": 2.264,
+      "step": 1923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2368525108738635e-05,
+      "loss": 1.9949,
+      "step": 1924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2364570976670624e-05,
+      "loss": 2.0874,
+      "step": 1925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.236061684460261e-05,
+      "loss": 1.9545,
+      "step": 1926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.23566627125346e-05,
+      "loss": 2.1472,
+      "step": 1927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.235270858046659e-05,
+      "loss": 2.1278,
+      "step": 1928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.234875444839858e-05,
+      "loss": 2.1735,
+      "step": 1929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.234480031633057e-05,
+      "loss": 2.2213,
+      "step": 1930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.234084618426256e-05,
+      "loss": 2.2944,
+      "step": 1931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2336892052194546e-05,
+      "loss": 1.9236,
+      "step": 1932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2332937920126534e-05,
+      "loss": 2.2469,
+      "step": 1933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.232898378805852e-05,
+      "loss": 2.0726,
+      "step": 1934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.232502965599051e-05,
+      "loss": 2.1872,
+      "step": 1935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.23210755239225e-05,
+      "loss": 2.2534,
+      "step": 1936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.231712139185449e-05,
+      "loss": 2.1243,
+      "step": 1937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.231316725978648e-05,
+      "loss": 2.0543,
+      "step": 1938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.230921312771847e-05,
+      "loss": 2.2386,
+      "step": 1939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2305258995650456e-05,
+      "loss": 2.0054,
+      "step": 1940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2301304863582445e-05,
+      "loss": 2.0058,
+      "step": 1941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2297350731514434e-05,
+      "loss": 2.3751,
+      "step": 1942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.229339659944642e-05,
+      "loss": 2.1947,
+      "step": 1943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.228944246737842e-05,
+      "loss": 2.244,
+      "step": 1944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.22854883353104e-05,
+      "loss": 2.1312,
+      "step": 1945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.228153420324239e-05,
+      "loss": 2.059,
+      "step": 1946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.227758007117438e-05,
+      "loss": 1.8968,
+      "step": 1947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2273625939106374e-05,
+      "loss": 2.121,
+      "step": 1948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2269671807038356e-05,
+      "loss": 2.1246,
+      "step": 1949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2265717674970344e-05,
+      "loss": 2.0181,
+      "step": 1950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.226176354290233e-05,
+      "loss": 2.0188,
+      "step": 1951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.225780941083433e-05,
+      "loss": 2.2791,
+      "step": 1952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.225385527876631e-05,
+      "loss": 2.0816,
+      "step": 1953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.22499011466983e-05,
+      "loss": 2.2286,
+      "step": 1954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.224594701463029e-05,
+      "loss": 2.1266,
+      "step": 1955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2241992882562284e-05,
+      "loss": 1.9943,
+      "step": 1956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2238038750494266e-05,
+      "loss": 2.1987,
+      "step": 1957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2234084618426255e-05,
+      "loss": 2.063,
+      "step": 1958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.2230130486358244e-05,
+      "loss": 2.0576,
+      "step": 1959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.222617635429024e-05,
+      "loss": 2.1114,
+      "step": 1960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 2.3331,
+      "step": 1961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.221826809015421e-05,
+      "loss": 2.4527,
+      "step": 1962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2214313958086206e-05,
+      "loss": 2.0106,
+      "step": 1963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2210359826018195e-05,
+      "loss": 1.9531,
+      "step": 1964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.220640569395018e-05,
+      "loss": 2.1696,
+      "step": 1965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2202451561882166e-05,
+      "loss": 2.002,
+      "step": 1966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.219849742981416e-05,
+      "loss": 2.0936,
+      "step": 1967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.219454329774615e-05,
+      "loss": 2.0151,
+      "step": 1968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.219058916567813e-05,
+      "loss": 2.0381,
+      "step": 1969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.218663503361012e-05,
+      "loss": 2.0371,
+      "step": 1970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2182680901542117e-05,
+      "loss": 1.9898,
+      "step": 1971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2178726769474105e-05,
+      "loss": 2.4553,
+      "step": 1972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.217477263740609e-05,
+      "loss": 2.3536,
+      "step": 1973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2170818505338076e-05,
+      "loss": 2.1078,
+      "step": 1974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.216686437327007e-05,
+      "loss": 1.9218,
+      "step": 1975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.216291024120206e-05,
+      "loss": 2.2704,
+      "step": 1976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.215895610913404e-05,
+      "loss": 1.8651,
+      "step": 1977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.215500197706604e-05,
+      "loss": 2.1023,
+      "step": 1978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.215104784499803e-05,
+      "loss": 2.4526,
+      "step": 1979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2147093712930016e-05,
+      "loss": 2.2334,
+      "step": 1980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2143139580862e-05,
+      "loss": 2.1456,
+      "step": 1981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2139185448793994e-05,
+      "loss": 2.1082,
+      "step": 1982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.213523131672598e-05,
+      "loss": 2.2585,
+      "step": 1983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.213127718465797e-05,
+      "loss": 2.2003,
+      "step": 1984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.212732305258995e-05,
+      "loss": 1.9925,
+      "step": 1985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.212336892052195e-05,
+      "loss": 2.236,
+      "step": 1986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.211941478845394e-05,
+      "loss": 1.9841,
+      "step": 1987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.211546065638593e-05,
+      "loss": 2.0295,
+      "step": 1988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.211150652431791e-05,
+      "loss": 2.0619,
+      "step": 1989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2107552392249904e-05,
+      "loss": 2.0383,
+      "step": 1990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.210359826018189e-05,
+      "loss": 2.1665,
+      "step": 1991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.209964412811388e-05,
+      "loss": 2.1039,
+      "step": 1992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2095689996045864e-05,
+      "loss": 2.1515,
+      "step": 1993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.209173586397786e-05,
+      "loss": 2.0852,
+      "step": 1994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.208778173190985e-05,
+      "loss": 2.2143,
+      "step": 1995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.208382759984184e-05,
+      "loss": 2.3761,
+      "step": 1996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2079873467773826e-05,
+      "loss": 2.1757,
+      "step": 1997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2075919335705815e-05,
+      "loss": 2.1693,
+      "step": 1998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2071965203637804e-05,
+      "loss": 2.3507,
+      "step": 1999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.206801107156979e-05,
+      "loss": 2.039,
+      "step": 2000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.206405693950178e-05,
+      "loss": 2.6678,
+      "step": 2001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.206010280743377e-05,
+      "loss": 2.1044,
+      "step": 2002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.205614867536576e-05,
+      "loss": 2.1182,
+      "step": 2003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.205219454329775e-05,
+      "loss": 1.9088,
+      "step": 2004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.204824041122974e-05,
+      "loss": 2.0986,
+      "step": 2005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2044286279161726e-05,
+      "loss": 1.8852,
+      "step": 2006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2040332147093714e-05,
+      "loss": 2.1745,
+      "step": 2007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.20363780150257e-05,
+      "loss": 2.09,
+      "step": 2008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.203242388295769e-05,
+      "loss": 2.0195,
+      "step": 2009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.202846975088968e-05,
+      "loss": 2.1226,
+      "step": 2010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.202451561882167e-05,
+      "loss": 2.084,
+      "step": 2011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2020561486753665e-05,
+      "loss": 2.1781,
+      "step": 2012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.201660735468565e-05,
+      "loss": 2.1288,
+      "step": 2013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2012653222617636e-05,
+      "loss": 2.0779,
+      "step": 2014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2008699090549625e-05,
+      "loss": 2.0738,
+      "step": 2015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.200474495848162e-05,
+      "loss": 2.072,
+      "step": 2016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.20007908264136e-05,
+      "loss": 1.8962,
+      "step": 2017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.199683669434559e-05,
+      "loss": 2.2413,
+      "step": 2018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.199288256227758e-05,
+      "loss": 2.2243,
+      "step": 2019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1988928430209576e-05,
+      "loss": 2.0944,
+      "step": 2020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.198497429814156e-05,
+      "loss": 1.9788,
+      "step": 2021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.198102016607355e-05,
+      "loss": 2.0162,
+      "step": 2022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1977066034005536e-05,
+      "loss": 2.0654,
+      "step": 2023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.197311190193753e-05,
+      "loss": 2.2339,
+      "step": 2024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.196915776986951e-05,
+      "loss": 2.1801,
+      "step": 2025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.19652036378015e-05,
+      "loss": 1.8418,
+      "step": 2026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.196124950573349e-05,
+      "loss": 2.2396,
+      "step": 2027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1957295373665486e-05,
+      "loss": 2.3149,
+      "step": 2028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.195334124159747e-05,
+      "loss": 2.1867,
+      "step": 2029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.194938710952946e-05,
+      "loss": 1.9393,
+      "step": 2030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.194543297746145e-05,
+      "loss": 2.0178,
+      "step": 2031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.194147884539344e-05,
+      "loss": 2.2063,
+      "step": 2032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1937524713325424e-05,
+      "loss": 2.0037,
+      "step": 2033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.193357058125741e-05,
+      "loss": 2.3198,
+      "step": 2034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.192961644918941e-05,
+      "loss": 1.9853,
+      "step": 2035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.19256623171214e-05,
+      "loss": 2.1737,
+      "step": 2036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.192170818505338e-05,
+      "loss": 2.2683,
+      "step": 2037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.191775405298537e-05,
+      "loss": 2.0816,
+      "step": 2038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1913799920917364e-05,
+      "loss": 2.2254,
+      "step": 2039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.190984578884935e-05,
+      "loss": 2.4886,
+      "step": 2040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1905891656781334e-05,
+      "loss": 2.0776,
+      "step": 2041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.190193752471332e-05,
+      "loss": 2.2934,
+      "step": 2042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.189798339264532e-05,
+      "loss": 2.4351,
+      "step": 2043
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.189402926057731e-05,
+      "loss": 2.0382,
+      "step": 2044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.189007512850929e-05,
+      "loss": 2.4467,
+      "step": 2045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.188612099644128e-05,
+      "loss": 2.3947,
+      "step": 2046
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1882166864373274e-05,
+      "loss": 1.9468,
+      "step": 2047
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.187821273230526e-05,
+      "loss": 1.9657,
+      "step": 2048
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1874258600237245e-05,
+      "loss": 2.0434,
+      "step": 2049
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.187030446816924e-05,
+      "loss": 2.1874,
+      "step": 2050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.186635033610123e-05,
+      "loss": 2.1352,
+      "step": 2051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.186239620403322e-05,
+      "loss": 2.3601,
+      "step": 2052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.18584420719652e-05,
+      "loss": 2.129,
+      "step": 2053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1854487939897196e-05,
+      "loss": 2.0334,
+      "step": 2054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1850533807829185e-05,
+      "loss": 2.2304,
+      "step": 2055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1846579675761174e-05,
+      "loss": 2.1976,
+      "step": 2056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.184262554369316e-05,
+      "loss": 1.8078,
+      "step": 2057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.183867141162515e-05,
+      "loss": 2.2448,
+      "step": 2058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.183471727955714e-05,
+      "loss": 2.0624,
+      "step": 2059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.183076314748913e-05,
+      "loss": 2.2307,
+      "step": 2060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.182680901542112e-05,
+      "loss": 2.0819,
+      "step": 2061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1822854883353107e-05,
+      "loss": 2.127,
+      "step": 2062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1818900751285095e-05,
+      "loss": 2.3654,
+      "step": 2063
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1814946619217084e-05,
+      "loss": 2.1874,
+      "step": 2064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.181099248714907e-05,
+      "loss": 2.4245,
+      "step": 2065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.180703835508106e-05,
+      "loss": 1.9695,
+      "step": 2066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.180308422301305e-05,
+      "loss": 2.1488,
+      "step": 2067
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.179913009094504e-05,
+      "loss": 1.9806,
+      "step": 2068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.179517595887703e-05,
+      "loss": 2.1618,
+      "step": 2069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.179122182680902e-05,
+      "loss": 2.2365,
+      "step": 2070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1787267694741006e-05,
+      "loss": 1.9745,
+      "step": 2071
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1783313562672995e-05,
+      "loss": 2.0764,
+      "step": 2072
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1779359430604984e-05,
+      "loss": 1.9649,
+      "step": 2073
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.177540529853697e-05,
+      "loss": 2.1795,
+      "step": 2074
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.177145116646896e-05,
+      "loss": 2.0693,
+      "step": 2075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.176749703440095e-05,
+      "loss": 2.069,
+      "step": 2076
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.176354290233294e-05,
+      "loss": 2.311,
+      "step": 2077
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.175958877026493e-05,
+      "loss": 2.2212,
+      "step": 2078
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.175563463819692e-05,
+      "loss": 2.5094,
+      "step": 2079
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1751680506128905e-05,
+      "loss": 2.2597,
+      "step": 2080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.1747726374060894e-05,
+      "loss": 1.8214,
+      "step": 2081
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.174377224199288e-05,
+      "loss": 2.2232,
+      "step": 2082
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.173981810992487e-05,
+      "loss": 2.2079,
+      "step": 2083
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.173586397785687e-05,
+      "loss": 2.1352,
+      "step": 2084
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.173190984578885e-05,
+      "loss": 2.1173,
+      "step": 2085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.172795571372084e-05,
+      "loss": 1.8851,
+      "step": 2086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.172400158165283e-05,
+      "loss": 2.1673,
+      "step": 2087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.172004744958482e-05,
+      "loss": 2.1767,
+      "step": 2088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1716093317516805e-05,
+      "loss": 2.0289,
+      "step": 2089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1712139185448794e-05,
+      "loss": 2.0692,
+      "step": 2090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.170818505338078e-05,
+      "loss": 2.0487,
+      "step": 2091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.170423092131278e-05,
+      "loss": 2.164,
+      "step": 2092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.170027678924476e-05,
+      "loss": 2.0992,
+      "step": 2093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.169632265717675e-05,
+      "loss": 2.1632,
+      "step": 2094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.169236852510874e-05,
+      "loss": 2.1498,
+      "step": 2095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1688414393040734e-05,
+      "loss": 2.2367,
+      "step": 2096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1684460260972716e-05,
+      "loss": 2.1368,
+      "step": 2097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1680506128904704e-05,
+      "loss": 2.0826,
+      "step": 2098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.16765519968367e-05,
+      "loss": 2.1981,
+      "step": 2099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.167259786476869e-05,
+      "loss": 1.7697,
+      "step": 2100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.166864373270067e-05,
+      "loss": 2.1708,
+      "step": 2101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.166468960063266e-05,
+      "loss": 2.1512,
+      "step": 2102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1660735468564655e-05,
+      "loss": 1.9641,
+      "step": 2103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1656781336496644e-05,
+      "loss": 2.1268,
+      "step": 2104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1652827204428626e-05,
+      "loss": 2.1222,
+      "step": 2105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1648873072360615e-05,
+      "loss": 2.0999,
+      "step": 2106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.164491894029261e-05,
+      "loss": 2.275,
+      "step": 2107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.16409648082246e-05,
+      "loss": 2.0334,
+      "step": 2108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.163701067615658e-05,
+      "loss": 2.2206,
+      "step": 2109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.163305654408857e-05,
+      "loss": 2.2008,
+      "step": 2110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1629102412020566e-05,
+      "loss": 2.0267,
+      "step": 2111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1625148279952555e-05,
+      "loss": 2.0102,
+      "step": 2112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1621194147884544e-05,
+      "loss": 2.2545,
+      "step": 2113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1617240015816526e-05,
+      "loss": 2.1484,
+      "step": 2114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.161328588374852e-05,
+      "loss": 2.2808,
+      "step": 2115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.160933175168051e-05,
+      "loss": 2.1461,
+      "step": 2116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.16053776196125e-05,
+      "loss": 2.0483,
+      "step": 2117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.160142348754449e-05,
+      "loss": 2.1804,
+      "step": 2118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1597469355476477e-05,
+      "loss": 2.2487,
+      "step": 2119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1593515223408465e-05,
+      "loss": 2.1737,
+      "step": 2120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1589561091340454e-05,
+      "loss": 2.0269,
+      "step": 2121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.158560695927244e-05,
+      "loss": 2.1302,
+      "step": 2122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.158165282720443e-05,
+      "loss": 2.0664,
+      "step": 2123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.157769869513642e-05,
+      "loss": 2.2381,
+      "step": 2124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.157374456306841e-05,
+      "loss": 2.2951,
+      "step": 2125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.15697904310004e-05,
+      "loss": 2.1754,
+      "step": 2126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.156583629893239e-05,
+      "loss": 2.0999,
+      "step": 2127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1561882166864376e-05,
+      "loss": 1.8801,
+      "step": 2128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1557928034796365e-05,
+      "loss": 1.9847,
+      "step": 2129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1553973902728354e-05,
+      "loss": 2.0041,
+      "step": 2130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.155001977066034e-05,
+      "loss": 2.3881,
+      "step": 2131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.154606563859233e-05,
+      "loss": 2.0386,
+      "step": 2132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.154211150652432e-05,
+      "loss": 1.9995,
+      "step": 2133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.153815737445631e-05,
+      "loss": 1.9782,
+      "step": 2134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.15342032423883e-05,
+      "loss": 1.9822,
+      "step": 2135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1530249110320287e-05,
+      "loss": 2.0976,
+      "step": 2136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1526294978252275e-05,
+      "loss": 2.2635,
+      "step": 2137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1522340846184264e-05,
+      "loss": 1.9867,
+      "step": 2138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.151838671411625e-05,
+      "loss": 2.0707,
+      "step": 2139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.151443258204824e-05,
+      "loss": 1.9845,
+      "step": 2140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.151047844998023e-05,
+      "loss": 2.0354,
+      "step": 2141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.150652431791222e-05,
+      "loss": 2.4929,
+      "step": 2142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.150257018584421e-05,
+      "loss": 2.0947,
+      "step": 2143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.14986160537762e-05,
+      "loss": 1.8276,
+      "step": 2144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1494661921708186e-05,
+      "loss": 1.9528,
+      "step": 2145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1490707789640175e-05,
+      "loss": 2.382,
+      "step": 2146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1486753657572164e-05,
+      "loss": 1.9429,
+      "step": 2147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.148279952550415e-05,
+      "loss": 2.29,
+      "step": 2148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.147884539343614e-05,
+      "loss": 2.0835,
+      "step": 2149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.147489126136813e-05,
+      "loss": 2.027,
+      "step": 2150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.147093712930012e-05,
+      "loss": 2.3377,
+      "step": 2151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1466982997232115e-05,
+      "loss": 2.0428,
+      "step": 2152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.14630288651641e-05,
+      "loss": 2.0033,
+      "step": 2153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1459074733096085e-05,
+      "loss": 2.0044,
+      "step": 2154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1455120601028074e-05,
+      "loss": 2.0807,
+      "step": 2155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.145116646896007e-05,
+      "loss": 2.22,
+      "step": 2156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.144721233689205e-05,
+      "loss": 2.1646,
+      "step": 2157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.144325820482404e-05,
+      "loss": 2.1633,
+      "step": 2158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.143930407275603e-05,
+      "loss": 1.8306,
+      "step": 2159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1435349940688025e-05,
+      "loss": 1.8543,
+      "step": 2160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.143139580862001e-05,
+      "loss": 2.4466,
+      "step": 2161
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1427441676551996e-05,
+      "loss": 1.9049,
+      "step": 2162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1423487544483985e-05,
+      "loss": 2.2609,
+      "step": 2163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.141953341241598e-05,
+      "loss": 2.103,
+      "step": 2164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.141557928034796e-05,
+      "loss": 1.7086,
+      "step": 2165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.141162514827995e-05,
+      "loss": 1.9428,
+      "step": 2166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.140767101621194e-05,
+      "loss": 2.07,
+      "step": 2167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1403716884143936e-05,
+      "loss": 2.2376,
+      "step": 2168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.139976275207592e-05,
+      "loss": 2.0472,
+      "step": 2169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.139580862000791e-05,
+      "loss": 2.0315,
+      "step": 2170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.13918544879399e-05,
+      "loss": 2.324,
+      "step": 2171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.138790035587189e-05,
+      "loss": 2.2847,
+      "step": 2172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.138394622380388e-05,
+      "loss": 1.8457,
+      "step": 2173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.137999209173586e-05,
+      "loss": 2.0071,
+      "step": 2174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.137603795966786e-05,
+      "loss": 2.1951,
+      "step": 2175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1372083827599846e-05,
+      "loss": 2.0922,
+      "step": 2176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1368129695531835e-05,
+      "loss": 1.8952,
+      "step": 2177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.136417556346382e-05,
+      "loss": 2.1854,
+      "step": 2178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.136022143139581e-05,
+      "loss": 2.0875,
+      "step": 2179
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.13562672993278e-05,
+      "loss": 1.9303,
+      "step": 2180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.135231316725979e-05,
+      "loss": 2.114,
+      "step": 2181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.134835903519177e-05,
+      "loss": 2.196,
+      "step": 2182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.134440490312377e-05,
+      "loss": 2.371,
+      "step": 2183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.134045077105576e-05,
+      "loss": 2.0466,
+      "step": 2184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1336496638987746e-05,
+      "loss": 2.3747,
+      "step": 2185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1332542506919735e-05,
+      "loss": 2.1579,
+      "step": 2186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1328588374851724e-05,
+      "loss": 2.0322,
+      "step": 2187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.132463424278371e-05,
+      "loss": 1.9971,
+      "step": 2188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.13206801107157e-05,
+      "loss": 2.3535,
+      "step": 2189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.131672597864769e-05,
+      "loss": 2.1345,
+      "step": 2190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.131277184657968e-05,
+      "loss": 2.1642,
+      "step": 2191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.130881771451167e-05,
+      "loss": 2.2003,
+      "step": 2192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1304863582443656e-05,
+      "loss": 2.1983,
+      "step": 2193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1300909450375645e-05,
+      "loss": 2.1094,
+      "step": 2194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1296955318307634e-05,
+      "loss": 2.1517,
+      "step": 2195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.129300118623962e-05,
+      "loss": 2.1242,
+      "step": 2196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.128904705417161e-05,
+      "loss": 1.9328,
+      "step": 2197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.12850929221036e-05,
+      "loss": 1.9982,
+      "step": 2198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.128113879003559e-05,
+      "loss": 2.084,
+      "step": 2199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.127718465796758e-05,
+      "loss": 2.0546,
+      "step": 2200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.127323052589957e-05,
+      "loss": 2.0096,
+      "step": 2201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1269276393831556e-05,
+      "loss": 2.2487,
+      "step": 2202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1265322261763545e-05,
+      "loss": 2.0428,
+      "step": 2203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1261368129695534e-05,
+      "loss": 2.074,
+      "step": 2204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.125741399762752e-05,
+      "loss": 2.0662,
+      "step": 2205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.125345986555951e-05,
+      "loss": 2.3129,
+      "step": 2206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.12495057334915e-05,
+      "loss": 2.1331,
+      "step": 2207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.124555160142349e-05,
+      "loss": 2.1415,
+      "step": 2208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.124159746935548e-05,
+      "loss": 1.9089,
+      "step": 2209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1237643337287467e-05,
+      "loss": 2.1674,
+      "step": 2210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1233689205219455e-05,
+      "loss": 2.0391,
+      "step": 2211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.1229735073151444e-05,
+      "loss": 1.9646,
+      "step": 2212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.122578094108343e-05,
+      "loss": 2.2103,
+      "step": 2213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.122182680901542e-05,
+      "loss": 2.2719,
+      "step": 2214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.121787267694741e-05,
+      "loss": 2.2091,
+      "step": 2215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.12139185448794e-05,
+      "loss": 2.0269,
+      "step": 2216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.120996441281139e-05,
+      "loss": 1.9663,
+      "step": 2217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.120601028074338e-05,
+      "loss": 2.2671,
+      "step": 2218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1202056148675366e-05,
+      "loss": 2.2743,
+      "step": 2219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1198102016607355e-05,
+      "loss": 2.104,
+      "step": 2220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1194147884539344e-05,
+      "loss": 2.1928,
+      "step": 2221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.119019375247133e-05,
+      "loss": 2.0757,
+      "step": 2222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.118623962040332e-05,
+      "loss": 2.1477,
+      "step": 2223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.118228548833532e-05,
+      "loss": 2.0573,
+      "step": 2224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.11783313562673e-05,
+      "loss": 2.1308,
+      "step": 2225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.117437722419929e-05,
+      "loss": 2.2202,
+      "step": 2226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1170423092131277e-05,
+      "loss": 1.985,
+      "step": 2227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.116646896006327e-05,
+      "loss": 2.2249,
+      "step": 2228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.116251482799526e-05,
+      "loss": 2.0594,
+      "step": 2229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.115856069592724e-05,
+      "loss": 2.0372,
+      "step": 2230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.115460656385923e-05,
+      "loss": 2.171,
+      "step": 2231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.115065243179123e-05,
+      "loss": 2.168,
+      "step": 2232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1146698299723216e-05,
+      "loss": 2.047,
+      "step": 2233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.11427441676552e-05,
+      "loss": 2.0923,
+      "step": 2234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.113879003558719e-05,
+      "loss": 2.2373,
+      "step": 2235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.113483590351918e-05,
+      "loss": 1.9798,
+      "step": 2236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.113088177145117e-05,
+      "loss": 2.2085,
+      "step": 2237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1126927639383154e-05,
+      "loss": 2.18,
+      "step": 2238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.112297350731515e-05,
+      "loss": 2.2379,
+      "step": 2239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.111901937524714e-05,
+      "loss": 1.9138,
+      "step": 2240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.111506524317913e-05,
+      "loss": 1.9825,
+      "step": 2241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.111111111111111e-05,
+      "loss": 2.2606,
+      "step": 2242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1107156979043105e-05,
+      "loss": 2.1325,
+      "step": 2243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1103202846975093e-05,
+      "loss": 2.1827,
+      "step": 2244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.109924871490708e-05,
+      "loss": 2.1362,
+      "step": 2245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1095294582839064e-05,
+      "loss": 2.0645,
+      "step": 2246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.109134045077106e-05,
+      "loss": 1.9128,
+      "step": 2247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.108738631870305e-05,
+      "loss": 2.2752,
+      "step": 2248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.108343218663504e-05,
+      "loss": 1.976,
+      "step": 2249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.107947805456702e-05,
+      "loss": 2.288,
+      "step": 2250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1075523922499015e-05,
+      "loss": 1.9371,
+      "step": 2251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1071569790431004e-05,
+      "loss": 1.8285,
+      "step": 2252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.106761565836299e-05,
+      "loss": 2.1701,
+      "step": 2253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1063661526294975e-05,
+      "loss": 2.1295,
+      "step": 2254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.105970739422697e-05,
+      "loss": 2.3121,
+      "step": 2255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.105575326215896e-05,
+      "loss": 2.015,
+      "step": 2256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.105179913009095e-05,
+      "loss": 2.0144,
+      "step": 2257
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.104784499802294e-05,
+      "loss": 2.1472,
+      "step": 2258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1043890865954926e-05,
+      "loss": 2.0173,
+      "step": 2259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1039936733886915e-05,
+      "loss": 2.0456,
+      "step": 2260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1035982601818904e-05,
+      "loss": 2.1169,
+      "step": 2261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.103202846975089e-05,
+      "loss": 2.1025,
+      "step": 2262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.102807433768288e-05,
+      "loss": 2.1808,
+      "step": 2263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.102412020561487e-05,
+      "loss": 1.9555,
+      "step": 2264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.102016607354686e-05,
+      "loss": 2.1573,
+      "step": 2265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.101621194147885e-05,
+      "loss": 2.0366,
+      "step": 2266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1012257809410836e-05,
+      "loss": 2.2883,
+      "step": 2267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1008303677342825e-05,
+      "loss": 2.0128,
+      "step": 2268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.1004349545274814e-05,
+      "loss": 2.1149,
+      "step": 2269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.10003954132068e-05,
+      "loss": 1.938,
+      "step": 2270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.099644128113879e-05,
+      "loss": 2.1481,
+      "step": 2271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.099248714907078e-05,
+      "loss": 2.2266,
+      "step": 2272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.098853301700277e-05,
+      "loss": 2.2919,
+      "step": 2273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.098457888493476e-05,
+      "loss": 2.2512,
+      "step": 2274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.098062475286675e-05,
+      "loss": 1.9945,
+      "step": 2275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0976670620798736e-05,
+      "loss": 2.2289,
+      "step": 2276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0972716488730725e-05,
+      "loss": 2.1356,
+      "step": 2277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0968762356662714e-05,
+      "loss": 1.9576,
+      "step": 2278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.09648082245947e-05,
+      "loss": 1.9942,
+      "step": 2279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.096085409252669e-05,
+      "loss": 2.0527,
+      "step": 2280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.095689996045868e-05,
+      "loss": 2.2358,
+      "step": 2281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.095294582839067e-05,
+      "loss": 2.1163,
+      "step": 2282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.094899169632266e-05,
+      "loss": 2.1632,
+      "step": 2283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0945037564254647e-05,
+      "loss": 2.3483,
+      "step": 2284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.094108343218664e-05,
+      "loss": 2.0431,
+      "step": 2285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0937129300118624e-05,
+      "loss": 1.9302,
+      "step": 2286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.093317516805061e-05,
+      "loss": 2.0536,
+      "step": 2287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.09292210359826e-05,
+      "loss": 2.2509,
+      "step": 2288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.09252669039146e-05,
+      "loss": 2.0504,
+      "step": 2289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.092131277184658e-05,
+      "loss": 2.2373,
+      "step": 2290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.091735863977857e-05,
+      "loss": 2.0717,
+      "step": 2291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0913404507710564e-05,
+      "loss": 2.1734,
+      "step": 2292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.090945037564255e-05,
+      "loss": 2.111,
+      "step": 2293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0905496243574535e-05,
+      "loss": 2.1839,
+      "step": 2294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0901542111506524e-05,
+      "loss": 2.1006,
+      "step": 2295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.089758797943852e-05,
+      "loss": 2.1335,
+      "step": 2296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.089363384737051e-05,
+      "loss": 2.027,
+      "step": 2297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.088967971530249e-05,
+      "loss": 2.1053,
+      "step": 2298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.088572558323448e-05,
+      "loss": 2.1264,
+      "step": 2299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0881771451166475e-05,
+      "loss": 2.2242,
+      "step": 2300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.087781731909846e-05,
+      "loss": 2.0221,
+      "step": 2301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0873863187030445e-05,
+      "loss": 1.9683,
+      "step": 2302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0869909054962434e-05,
+      "loss": 2.0131,
+      "step": 2303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.086595492289443e-05,
+      "loss": 2.1102,
+      "step": 2304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.086200079082642e-05,
+      "loss": 2.1242,
+      "step": 2305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.08580466587584e-05,
+      "loss": 2.2331,
+      "step": 2306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.085409252669039e-05,
+      "loss": 2.0922,
+      "step": 2307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0850138394622385e-05,
+      "loss": 2.306,
+      "step": 2308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0846184262554374e-05,
+      "loss": 2.1043,
+      "step": 2309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0842230130486356e-05,
+      "loss": 1.9831,
+      "step": 2310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.083827599841835e-05,
+      "loss": 1.9511,
+      "step": 2311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.083432186635034e-05,
+      "loss": 1.925,
+      "step": 2312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.083036773428233e-05,
+      "loss": 2.163,
+      "step": 2313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.082641360221431e-05,
+      "loss": 2.156,
+      "step": 2314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.082245947014631e-05,
+      "loss": 2.0909,
+      "step": 2315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0818505338078296e-05,
+      "loss": 2.0486,
+      "step": 2316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0814551206010285e-05,
+      "loss": 1.9348,
+      "step": 2317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.081059707394227e-05,
+      "loss": 2.2632,
+      "step": 2318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.080664294187426e-05,
+      "loss": 2.2231,
+      "step": 2319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.080268880980625e-05,
+      "loss": 2.3993,
+      "step": 2320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.079873467773824e-05,
+      "loss": 2.0481,
+      "step": 2321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.079478054567022e-05,
+      "loss": 2.0003,
+      "step": 2322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.079082641360222e-05,
+      "loss": 2.0817,
+      "step": 2323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0786872281534206e-05,
+      "loss": 2.1998,
+      "step": 2324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0782918149466195e-05,
+      "loss": 1.9192,
+      "step": 2325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0778964017398184e-05,
+      "loss": 2.0312,
+      "step": 2326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.077500988533017e-05,
+      "loss": 1.9737,
+      "step": 2327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.077105575326216e-05,
+      "loss": 1.8559,
+      "step": 2328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.076710162119415e-05,
+      "loss": 2.3074,
+      "step": 2329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.076314748912614e-05,
+      "loss": 2.2294,
+      "step": 2330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.075919335705813e-05,
+      "loss": 2.36,
+      "step": 2331
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.075523922499012e-05,
+      "loss": 2.2038,
+      "step": 2332
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0751285092922106e-05,
+      "loss": 2.146,
+      "step": 2333
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0747330960854095e-05,
+      "loss": 2.0949,
+      "step": 2334
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.0743376828786083e-05,
+      "loss": 2.0125,
+      "step": 2335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.073942269671807e-05,
+      "loss": 2.2712,
+      "step": 2336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.073546856465006e-05,
+      "loss": 2.3317,
+      "step": 2337
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.073151443258205e-05,
+      "loss": 1.989,
+      "step": 2338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.072756030051404e-05,
+      "loss": 2.149,
+      "step": 2339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.072360616844603e-05,
+      "loss": 2.0939,
+      "step": 2340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0719652036378016e-05,
+      "loss": 2.0367,
+      "step": 2341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0715697904310005e-05,
+      "loss": 2.3441,
+      "step": 2342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0711743772241994e-05,
+      "loss": 2.1681,
+      "step": 2343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.070778964017398e-05,
+      "loss": 2.2164,
+      "step": 2344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.070383550810598e-05,
+      "loss": 2.1681,
+      "step": 2345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.069988137603796e-05,
+      "loss": 2.3255,
+      "step": 2346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.069592724396995e-05,
+      "loss": 1.889,
+      "step": 2347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.069197311190194e-05,
+      "loss": 2.1639,
+      "step": 2348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0688018979833934e-05,
+      "loss": 2.2791,
+      "step": 2349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0684064847765916e-05,
+      "loss": 2.3205,
+      "step": 2350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0680110715697905e-05,
+      "loss": 2.3133,
+      "step": 2351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0676156583629894e-05,
+      "loss": 2.2522,
+      "step": 2352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.067220245156189e-05,
+      "loss": 1.9706,
+      "step": 2353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.066824831949387e-05,
+      "loss": 1.9356,
+      "step": 2354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.066429418742586e-05,
+      "loss": 2.02,
+      "step": 2355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.066034005535785e-05,
+      "loss": 2.0136,
+      "step": 2356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0656385923289844e-05,
+      "loss": 2.0145,
+      "step": 2357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0652431791221826e-05,
+      "loss": 2.0599,
+      "step": 2358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0648477659153815e-05,
+      "loss": 1.9781,
+      "step": 2359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.064452352708581e-05,
+      "loss": 2.0653,
+      "step": 2360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.06405693950178e-05,
+      "loss": 2.324,
+      "step": 2361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.063661526294978e-05,
+      "loss": 2.2422,
+      "step": 2362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.063266113088177e-05,
+      "loss": 2.088,
+      "step": 2363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0628706998813766e-05,
+      "loss": 2.0916,
+      "step": 2364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0624752866745755e-05,
+      "loss": 1.9927,
+      "step": 2365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.062079873467774e-05,
+      "loss": 1.8271,
+      "step": 2366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0616844602609726e-05,
+      "loss": 2.2179,
+      "step": 2367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.061289047054172e-05,
+      "loss": 2.0386,
+      "step": 2368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.060893633847371e-05,
+      "loss": 2.1393,
+      "step": 2369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.060498220640569e-05,
+      "loss": 2.081,
+      "step": 2370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.060102807433768e-05,
+      "loss": 2.0844,
+      "step": 2371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.059707394226968e-05,
+      "loss": 2.0484,
+      "step": 2372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0593119810201666e-05,
+      "loss": 2.1156,
+      "step": 2373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.058916567813365e-05,
+      "loss": 2.1159,
+      "step": 2374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0585211546065637e-05,
+      "loss": 2.0223,
+      "step": 2375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.058125741399763e-05,
+      "loss": 2.1128,
+      "step": 2376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.057730328192962e-05,
+      "loss": 2.4541,
+      "step": 2377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.05733491498616e-05,
+      "loss": 2.0874,
+      "step": 2378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.05693950177936e-05,
+      "loss": 2.2028,
+      "step": 2379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.056544088572559e-05,
+      "loss": 1.9174,
+      "step": 2380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0561486753657576e-05,
+      "loss": 2.3014,
+      "step": 2381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.055753262158956e-05,
+      "loss": 2.119,
+      "step": 2382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0553578489521554e-05,
+      "loss": 2.1435,
+      "step": 2383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.054962435745354e-05,
+      "loss": 2.0135,
+      "step": 2384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.054567022538553e-05,
+      "loss": 2.293,
+      "step": 2385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0541716093317514e-05,
+      "loss": 2.0262,
+      "step": 2386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.053776196124951e-05,
+      "loss": 2.3592,
+      "step": 2387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.05338078291815e-05,
+      "loss": 2.0344,
+      "step": 2388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.052985369711349e-05,
+      "loss": 2.2419,
+      "step": 2389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.052589956504547e-05,
+      "loss": 1.9673,
+      "step": 2390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0521945432977465e-05,
+      "loss": 2.385,
+      "step": 2391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.051799130090945e-05,
+      "loss": 1.9428,
+      "step": 2392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.051403716884144e-05,
+      "loss": 2.1386,
+      "step": 2393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0510083036773424e-05,
+      "loss": 1.7928,
+      "step": 2394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.050612890470542e-05,
+      "loss": 2.0787,
+      "step": 2395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.050217477263741e-05,
+      "loss": 2.2224,
+      "step": 2396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.04982206405694e-05,
+      "loss": 2.096,
+      "step": 2397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0494266508501386e-05,
+      "loss": 2.4017,
+      "step": 2398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0490312376433375e-05,
+      "loss": 2.1421,
+      "step": 2399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0486358244365364e-05,
+      "loss": 2.2765,
+      "step": 2400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.048240411229735e-05,
+      "loss": 2.1788,
+      "step": 2401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.047844998022934e-05,
+      "loss": 2.2348,
+      "step": 2402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.047449584816133e-05,
+      "loss": 2.128,
+      "step": 2403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.047054171609332e-05,
+      "loss": 2.137,
+      "step": 2404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.046658758402531e-05,
+      "loss": 2.1826,
+      "step": 2405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.04626334519573e-05,
+      "loss": 2.2602,
+      "step": 2406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0458679319889286e-05,
+      "loss": 2.3057,
+      "step": 2407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0454725187821275e-05,
+      "loss": 2.093,
+      "step": 2408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0450771055753263e-05,
+      "loss": 2.0008,
+      "step": 2409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.044681692368525e-05,
+      "loss": 2.2867,
+      "step": 2410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.044286279161724e-05,
+      "loss": 2.1127,
+      "step": 2411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.043890865954923e-05,
+      "loss": 2.2327,
+      "step": 2412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0434954527481226e-05,
+      "loss": 2.293,
+      "step": 2413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.043100039541321e-05,
+      "loss": 2.2088,
+      "step": 2414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0427046263345196e-05,
+      "loss": 2.0825,
+      "step": 2415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0423092131277185e-05,
+      "loss": 1.9606,
+      "step": 2416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.041913799920918e-05,
+      "loss": 2.1409,
+      "step": 2417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.041518386714116e-05,
+      "loss": 2.2039,
+      "step": 2418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.041122973507315e-05,
+      "loss": 2.1696,
+      "step": 2419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.040727560300514e-05,
+      "loss": 2.0355,
+      "step": 2420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0403321470937136e-05,
+      "loss": 2.0388,
+      "step": 2421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.039936733886912e-05,
+      "loss": 2.144,
+      "step": 2422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.039541320680111e-05,
+      "loss": 2.2626,
+      "step": 2423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0391459074733096e-05,
+      "loss": 2.1566,
+      "step": 2424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.038750494266509e-05,
+      "loss": 2.1221,
+      "step": 2425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0383550810597074e-05,
+      "loss": 1.9737,
+      "step": 2426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.037959667852906e-05,
+      "loss": 2.3191,
+      "step": 2427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.037564254646105e-05,
+      "loss": 2.0846,
+      "step": 2428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.037168841439305e-05,
+      "loss": 2.0614,
+      "step": 2429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.036773428232503e-05,
+      "loss": 2.2362,
+      "step": 2430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.036378015025702e-05,
+      "loss": 2.2718,
+      "step": 2431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.035982601818901e-05,
+      "loss": 2.0567,
+      "step": 2432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0355871886121e-05,
+      "loss": 2.1727,
+      "step": 2433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0351917754052984e-05,
+      "loss": 1.9176,
+      "step": 2434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.034796362198497e-05,
+      "loss": 2.1118,
+      "step": 2435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.034400948991697e-05,
+      "loss": 2.0002,
+      "step": 2436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.034005535784896e-05,
+      "loss": 2.2326,
+      "step": 2437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.033610122578094e-05,
+      "loss": 2.0344,
+      "step": 2438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.033214709371293e-05,
+      "loss": 1.9504,
+      "step": 2439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0328192961644924e-05,
+      "loss": 2.2243,
+      "step": 2440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.032423882957691e-05,
+      "loss": 2.0886,
+      "step": 2441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0320284697508895e-05,
+      "loss": 2.1055,
+      "step": 2442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0316330565440884e-05,
+      "loss": 2.2565,
+      "step": 2443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.031237643337288e-05,
+      "loss": 1.9724,
+      "step": 2444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.030842230130487e-05,
+      "loss": 2.0202,
+      "step": 2445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.030446816923685e-05,
+      "loss": 1.9767,
+      "step": 2446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0300514037168846e-05,
+      "loss": 1.8245,
+      "step": 2447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0296559905100834e-05,
+      "loss": 2.3606,
+      "step": 2448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.029260577303282e-05,
+      "loss": 2.16,
+      "step": 2449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0288651640964805e-05,
+      "loss": 2.0833,
+      "step": 2450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.02846975088968e-05,
+      "loss": 1.9739,
+      "step": 2451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.028074337682879e-05,
+      "loss": 2.2684,
+      "step": 2452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.027678924476078e-05,
+      "loss": 2.1786,
+      "step": 2453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.027283511269276e-05,
+      "loss": 2.2576,
+      "step": 2454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0268880980624756e-05,
+      "loss": 1.8746,
+      "step": 2455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0264926848556745e-05,
+      "loss": 2.1307,
+      "step": 2456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0260972716488734e-05,
+      "loss": 2.1212,
+      "step": 2457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0257018584420716e-05,
+      "loss": 1.9765,
+      "step": 2458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.025306445235271e-05,
+      "loss": 2.1731,
+      "step": 2459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.02491103202847e-05,
+      "loss": 2.254,
+      "step": 2460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.024515618821669e-05,
+      "loss": 2.1942,
+      "step": 2461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.024120205614868e-05,
+      "loss": 2.0497,
+      "step": 2462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.023724792408067e-05,
+      "loss": 2.2624,
+      "step": 2463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0233293792012656e-05,
+      "loss": 2.18,
+      "step": 2464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.0229339659944645e-05,
+      "loss": 2.2617,
+      "step": 2465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.022538552787663e-05,
+      "loss": 2.3443,
+      "step": 2466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.022143139580862e-05,
+      "loss": 2.1191,
+      "step": 2467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.021747726374061e-05,
+      "loss": 2.1339,
+      "step": 2468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.02135231316726e-05,
+      "loss": 1.8179,
+      "step": 2469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.020956899960459e-05,
+      "loss": 2.0771,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.020561486753658e-05,
+      "loss": 2.0561,
+      "step": 2471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0201660735468566e-05,
+      "loss": 2.1199,
+      "step": 2472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0197706603400555e-05,
+      "loss": 2.2404,
+      "step": 2473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0193752471332544e-05,
+      "loss": 1.9883,
+      "step": 2474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.018979833926453e-05,
+      "loss": 2.3769,
+      "step": 2475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.018584420719652e-05,
+      "loss": 1.7709,
+      "step": 2476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.018189007512851e-05,
+      "loss": 2.4139,
+      "step": 2477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.01779359430605e-05,
+      "loss": 2.2643,
+      "step": 2478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.017398181099249e-05,
+      "loss": 2.0292,
+      "step": 2479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.017002767892448e-05,
+      "loss": 1.908,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0166073546856466e-05,
+      "loss": 2.0038,
+      "step": 2481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0162119414788455e-05,
+      "loss": 2.2701,
+      "step": 2482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0158165282720443e-05,
+      "loss": 1.935,
+      "step": 2483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.015421115065243e-05,
+      "loss": 2.0964,
+      "step": 2484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.015025701858443e-05,
+      "loss": 2.1154,
+      "step": 2485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.014630288651641e-05,
+      "loss": 2.1303,
+      "step": 2486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.01423487544484e-05,
+      "loss": 2.3068,
+      "step": 2487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.013839462238039e-05,
+      "loss": 2.0639,
+      "step": 2488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.013444049031238e-05,
+      "loss": 2.3051,
+      "step": 2489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0130486358244365e-05,
+      "loss": 2.1278,
+      "step": 2490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0126532226176354e-05,
+      "loss": 2.0193,
+      "step": 2491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.012257809410834e-05,
+      "loss": 2.0098,
+      "step": 2492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.011862396204034e-05,
+      "loss": 1.9218,
+      "step": 2493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.011466982997232e-05,
+      "loss": 2.3169,
+      "step": 2494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.011071569790431e-05,
+      "loss": 2.2511,
+      "step": 2495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.01067615658363e-05,
+      "loss": 1.9117,
+      "step": 2496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0102807433768294e-05,
+      "loss": 2.0334,
+      "step": 2497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0098853301700276e-05,
+      "loss": 1.97,
+      "step": 2498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0094899169632265e-05,
+      "loss": 2.1845,
+      "step": 2499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.009094503756426e-05,
+      "loss": 2.2361,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.008699090549625e-05,
+      "loss": 2.0282,
+      "step": 2501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.008303677342823e-05,
+      "loss": 2.1556,
+      "step": 2502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.007908264136022e-05,
+      "loss": 2.1334,
+      "step": 2503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0075128509292216e-05,
+      "loss": 2.1238,
+      "step": 2504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0071174377224204e-05,
+      "loss": 2.0882,
+      "step": 2505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0067220245156186e-05,
+      "loss": 2.0631,
+      "step": 2506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0063266113088175e-05,
+      "loss": 1.8445,
+      "step": 2507
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.005931198102017e-05,
+      "loss": 2.0481,
+      "step": 2508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.005535784895216e-05,
+      "loss": 2.4162,
+      "step": 2509
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.005140371688414e-05,
+      "loss": 1.8695,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.004744958481613e-05,
+      "loss": 2.099,
+      "step": 2511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0043495452748126e-05,
+      "loss": 2.0826,
+      "step": 2512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0039541320680115e-05,
+      "loss": 2.0114,
+      "step": 2513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.00355871886121e-05,
+      "loss": 2.2248,
+      "step": 2514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0031633056544086e-05,
+      "loss": 2.1903,
+      "step": 2515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.002767892447608e-05,
+      "loss": 2.0972,
+      "step": 2516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.002372479240807e-05,
+      "loss": 2.3765,
+      "step": 2517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.001977066034005e-05,
+      "loss": 1.9664,
+      "step": 2518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.001581652827205e-05,
+      "loss": 2.0242,
+      "step": 2519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.001186239620404e-05,
+      "loss": 2.1286,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0007908264136026e-05,
+      "loss": 1.986,
+      "step": 2521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0003954132068014e-05,
+      "loss": 2.217,
+      "step": 2522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4e-05,
+      "loss": 2.2881,
+      "step": 2523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.999604586793199e-05,
+      "loss": 2.1071,
+      "step": 2524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.999209173586398e-05,
+      "loss": 2.1272,
+      "step": 2525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.998813760379597e-05,
+      "loss": 2.0097,
+      "step": 2526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.998418347172796e-05,
+      "loss": 2.1826,
+      "step": 2527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.998022933965995e-05,
+      "loss": 2.3214,
+      "step": 2528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9976275207591936e-05,
+      "loss": 2.098,
+      "step": 2529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9972321075523925e-05,
+      "loss": 2.0727,
+      "step": 2530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9968366943455914e-05,
+      "loss": 2.0823,
+      "step": 2531
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.99644128113879e-05,
+      "loss": 2.1075,
+      "step": 2532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.996045867931989e-05,
+      "loss": 2.0491,
+      "step": 2533
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.995650454725188e-05,
+      "loss": 2.3448,
+      "step": 2534
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.995255041518387e-05,
+      "loss": 2.0116,
+      "step": 2535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.994859628311586e-05,
+      "loss": 2.1555,
+      "step": 2536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.994464215104785e-05,
+      "loss": 2.0473,
+      "step": 2537
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9940688018979836e-05,
+      "loss": 2.1284,
+      "step": 2538
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9936733886911825e-05,
+      "loss": 1.829,
+      "step": 2539
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.993277975484381e-05,
+      "loss": 2.1548,
+      "step": 2540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.99288256227758e-05,
+      "loss": 1.8301,
+      "step": 2541
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.992487149070779e-05,
+      "loss": 1.8924,
+      "step": 2542
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.992091735863978e-05,
+      "loss": 2.4121,
+      "step": 2543
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.991696322657177e-05,
+      "loss": 1.9158,
+      "step": 2544
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.991300909450376e-05,
+      "loss": 2.0371,
+      "step": 2545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9909054962435746e-05,
+      "loss": 1.9998,
+      "step": 2546
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9905100830367735e-05,
+      "loss": 2.0549,
+      "step": 2547
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9901146698299724e-05,
+      "loss": 1.9549,
+      "step": 2548
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.989719256623171e-05,
+      "loss": 2.2416,
+      "step": 2549
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.98932384341637e-05,
+      "loss": 2.1399,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.988928430209569e-05,
+      "loss": 2.1405,
+      "step": 2551
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.988533017002768e-05,
+      "loss": 2.0026,
+      "step": 2552
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9881376037959675e-05,
+      "loss": 2.1928,
+      "step": 2553
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.987742190589166e-05,
+      "loss": 2.0753,
+      "step": 2554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9873467773823646e-05,
+      "loss": 2.0846,
+      "step": 2555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9869513641755635e-05,
+      "loss": 1.8865,
+      "step": 2556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.986555950968763e-05,
+      "loss": 2.3177,
+      "step": 2557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.986160537761961e-05,
+      "loss": 1.9402,
+      "step": 2558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.98576512455516e-05,
+      "loss": 2.0304,
+      "step": 2559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.985369711348359e-05,
+      "loss": 2.1257,
+      "step": 2560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9849742981415585e-05,
+      "loss": 2.2477,
+      "step": 2561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.984578884934757e-05,
+      "loss": 2.0278,
+      "step": 2562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9841834717279556e-05,
+      "loss": 2.0226,
+      "step": 2563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9837880585211545e-05,
+      "loss": 2.3059,
+      "step": 2564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.983392645314354e-05,
+      "loss": 2.3605,
+      "step": 2565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.982997232107552e-05,
+      "loss": 2.1511,
+      "step": 2566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.982601818900751e-05,
+      "loss": 1.914,
+      "step": 2567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.98220640569395e-05,
+      "loss": 2.2901,
+      "step": 2568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9818109924871496e-05,
+      "loss": 1.9848,
+      "step": 2569
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.981415579280348e-05,
+      "loss": 2.0238,
+      "step": 2570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.981020166073547e-05,
+      "loss": 2.0334,
+      "step": 2571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.980624752866746e-05,
+      "loss": 2.2009,
+      "step": 2572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.980229339659945e-05,
+      "loss": 1.9906,
+      "step": 2573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9798339264531433e-05,
+      "loss": 2.1106,
+      "step": 2574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.979438513246342e-05,
+      "loss": 2.0845,
+      "step": 2575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.979043100039542e-05,
+      "loss": 2.1727,
+      "step": 2576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.978647686832741e-05,
+      "loss": 2.1842,
+      "step": 2577
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9782522736259396e-05,
+      "loss": 2.1291,
+      "step": 2578
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.977856860419138e-05,
+      "loss": 2.1878,
+      "step": 2579
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.977461447212337e-05,
+      "loss": 1.8254,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.977066034005536e-05,
+      "loss": 2.1466,
+      "step": 2581
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.976670620798735e-05,
+      "loss": 1.9555,
+      "step": 2582
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.976275207591933e-05,
+      "loss": 2.0855,
+      "step": 2583
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.975879794385133e-05,
+      "loss": 2.0749,
+      "step": 2584
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.975484381178332e-05,
+      "loss": 2.0,
+      "step": 2585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9750889679715306e-05,
+      "loss": 2.0664,
+      "step": 2586
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9746935547647295e-05,
+      "loss": 2.1502,
+      "step": 2587
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9742981415579284e-05,
+      "loss": 1.998,
+      "step": 2588
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.973902728351127e-05,
+      "loss": 2.0711,
+      "step": 2589
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.973507315144326e-05,
+      "loss": 2.2107,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.973111901937525e-05,
+      "loss": 2.0902,
+      "step": 2591
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.972716488730724e-05,
+      "loss": 2.0667,
+      "step": 2592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.972321075523923e-05,
+      "loss": 1.8852,
+      "step": 2593
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.971925662317122e-05,
+      "loss": 2.1886,
+      "step": 2594
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9715302491103206e-05,
+      "loss": 2.0411,
+      "step": 2595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9711348359035194e-05,
+      "loss": 1.9047,
+      "step": 2596
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.970739422696718e-05,
+      "loss": 1.9954,
+      "step": 2597
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.970344009489917e-05,
+      "loss": 1.7701,
+      "step": 2598
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.969948596283116e-05,
+      "loss": 2.2032,
+      "step": 2599
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.969553183076315e-05,
+      "loss": 1.8505,
+      "step": 2600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.969157769869514e-05,
+      "loss": 1.9605,
+      "step": 2601
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.968762356662713e-05,
+      "loss": 1.8522,
+      "step": 2602
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9683669434559116e-05,
+      "loss": 2.2913,
+      "step": 2603
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9679715302491105e-05,
+      "loss": 1.7797,
+      "step": 2604
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9675761170423094e-05,
+      "loss": 2.1038,
+      "step": 2605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.967180703835508e-05,
+      "loss": 1.796,
+      "step": 2606
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.966785290628707e-05,
+      "loss": 1.8675,
+      "step": 2607
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.966389877421906e-05,
+      "loss": 2.0232,
+      "step": 2608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.965994464215105e-05,
+      "loss": 1.9109,
+      "step": 2609
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.965599051008304e-05,
+      "loss": 2.3296,
+      "step": 2610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.965203637801503e-05,
+      "loss": 2.2233,
+      "step": 2611
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9648082245947016e-05,
+      "loss": 2.2683,
+      "step": 2612
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9644128113879004e-05,
+      "loss": 2.2693,
+      "step": 2613
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.964017398181099e-05,
+      "loss": 2.1271,
+      "step": 2614
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.963621984974298e-05,
+      "loss": 2.0174,
+      "step": 2615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.963226571767497e-05,
+      "loss": 2.1729,
+      "step": 2616
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.962831158560696e-05,
+      "loss": 2.2601,
+      "step": 2617
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.962435745353895e-05,
+      "loss": 2.1046,
+      "step": 2618
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.962040332147094e-05,
+      "loss": 2.1667,
+      "step": 2619
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9616449189402926e-05,
+      "loss": 2.1562,
+      "step": 2620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.961249505733492e-05,
+      "loss": 2.1012,
+      "step": 2621
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9608540925266904e-05,
+      "loss": 2.0336,
+      "step": 2622
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.960458679319889e-05,
+      "loss": 2.3497,
+      "step": 2623
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.960063266113088e-05,
+      "loss": 1.9063,
+      "step": 2624
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.959667852906288e-05,
+      "loss": 2.1465,
+      "step": 2625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.959272439699486e-05,
+      "loss": 2.2024,
+      "step": 2626
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.958877026492685e-05,
+      "loss": 1.9143,
+      "step": 2627
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.958481613285884e-05,
+      "loss": 2.1227,
+      "step": 2628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.958086200079083e-05,
+      "loss": 2.2538,
+      "step": 2629
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9576907868722815e-05,
+      "loss": 2.0529,
+      "step": 2630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.95729537366548e-05,
+      "loss": 2.0069,
+      "step": 2631
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.956899960458679e-05,
+      "loss": 2.0699,
+      "step": 2632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.956504547251879e-05,
+      "loss": 1.9483,
+      "step": 2633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.956109134045078e-05,
+      "loss": 1.9623,
+      "step": 2634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.955713720838276e-05,
+      "loss": 1.9405,
+      "step": 2635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.955318307631475e-05,
+      "loss": 2.1853,
+      "step": 2636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.954922894424674e-05,
+      "loss": 2.0256,
+      "step": 2637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.954527481217873e-05,
+      "loss": 2.1213,
+      "step": 2638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9541320680110714e-05,
+      "loss": 2.1218,
+      "step": 2639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.953736654804271e-05,
+      "loss": 2.0703,
+      "step": 2640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.95334124159747e-05,
+      "loss": 2.1027,
+      "step": 2641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.952945828390669e-05,
+      "loss": 2.0943,
+      "step": 2642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.952550415183867e-05,
+      "loss": 2.0788,
+      "step": 2643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9521550019770665e-05,
+      "loss": 2.0765,
+      "step": 2644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9517595887702654e-05,
+      "loss": 2.1028,
+      "step": 2645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.951364175563464e-05,
+      "loss": 1.9074,
+      "step": 2646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9509687623566625e-05,
+      "loss": 2.0088,
+      "step": 2647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.950573349149862e-05,
+      "loss": 1.9679,
+      "step": 2648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.950177935943061e-05,
+      "loss": 1.9051,
+      "step": 2649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.94978252273626e-05,
+      "loss": 2.0755,
+      "step": 2650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.949387109529458e-05,
+      "loss": 1.9262,
+      "step": 2651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9489916963226576e-05,
+      "loss": 2.0221,
+      "step": 2652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9485962831158564e-05,
+      "loss": 1.8358,
+      "step": 2653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.948200869909055e-05,
+      "loss": 1.9282,
+      "step": 2654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9478054567022535e-05,
+      "loss": 2.1006,
+      "step": 2655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.947410043495453e-05,
+      "loss": 1.8396,
+      "step": 2656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.947014630288652e-05,
+      "loss": 2.0293,
+      "step": 2657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.946619217081851e-05,
+      "loss": 2.2562,
+      "step": 2658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.94622380387505e-05,
+      "loss": 2.0971,
+      "step": 2659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9458283906682486e-05,
+      "loss": 2.0681,
+      "step": 2660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9454329774614475e-05,
+      "loss": 2.1802,
+      "step": 2661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9450375642546464e-05,
+      "loss": 2.019,
+      "step": 2662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.944642151047845e-05,
+      "loss": 2.0747,
+      "step": 2663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.944246737841044e-05,
+      "loss": 2.3106,
+      "step": 2664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.943851324634243e-05,
+      "loss": 2.1651,
+      "step": 2665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.943455911427442e-05,
+      "loss": 1.9221,
+      "step": 2666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.943060498220641e-05,
+      "loss": 1.9933,
+      "step": 2667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.94266508501384e-05,
+      "loss": 2.038,
+      "step": 2668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9422696718070386e-05,
+      "loss": 2.2511,
+      "step": 2669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9418742586002374e-05,
+      "loss": 1.8992,
+      "step": 2670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.941478845393436e-05,
+      "loss": 2.0066,
+      "step": 2671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.941083432186635e-05,
+      "loss": 1.8132,
+      "step": 2672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.940688018979834e-05,
+      "loss": 2.1428,
+      "step": 2673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.940292605773033e-05,
+      "loss": 2.3585,
+      "step": 2674
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.939897192566232e-05,
+      "loss": 2.1333,
+      "step": 2675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.939501779359431e-05,
+      "loss": 2.1998,
+      "step": 2676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9391063661526296e-05,
+      "loss": 2.1144,
+      "step": 2677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9387109529458285e-05,
+      "loss": 1.8721,
+      "step": 2678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9383155397390274e-05,
+      "loss": 2.2213,
+      "step": 2679
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.937920126532226e-05,
+      "loss": 2.3358,
+      "step": 2680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.937524713325425e-05,
+      "loss": 2.4769,
+      "step": 2681
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.937129300118624e-05,
+      "loss": 1.883,
+      "step": 2682
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.936733886911823e-05,
+      "loss": 2.1714,
+      "step": 2683
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.936338473705022e-05,
+      "loss": 2.3441,
+      "step": 2684
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.935943060498221e-05,
+      "loss": 2.278,
+      "step": 2685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9355476472914196e-05,
+      "loss": 1.9325,
+      "step": 2686
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9351522340846184e-05,
+      "loss": 1.8427,
+      "step": 2687
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.934756820877817e-05,
+      "loss": 1.9606,
+      "step": 2688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.934361407671016e-05,
+      "loss": 2.1916,
+      "step": 2689
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.933965994464216e-05,
+      "loss": 2.0188,
+      "step": 2690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.933570581257414e-05,
+      "loss": 2.1885,
+      "step": 2691
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.933175168050613e-05,
+      "loss": 1.9896,
+      "step": 2692
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9327797548438124e-05,
+      "loss": 1.9605,
+      "step": 2693
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.932384341637011e-05,
+      "loss": 2.0283,
+      "step": 2694
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9319889284302095e-05,
+      "loss": 2.3267,
+      "step": 2695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9315935152234084e-05,
+      "loss": 1.8315,
+      "step": 2696
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.931198102016608e-05,
+      "loss": 2.1925,
+      "step": 2697
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.930802688809807e-05,
+      "loss": 2.1628,
+      "step": 2698
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.930407275603005e-05,
+      "loss": 1.9907,
+      "step": 2699
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.930011862396204e-05,
+      "loss": 2.1185,
+      "step": 2700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9296164491894035e-05,
+      "loss": 2.1724,
+      "step": 2701
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9292210359826024e-05,
+      "loss": 2.3897,
+      "step": 2702
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9288256227758006e-05,
+      "loss": 2.0796,
+      "step": 2703
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9284302095689995e-05,
+      "loss": 1.8133,
+      "step": 2704
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.928034796362199e-05,
+      "loss": 1.8482,
+      "step": 2705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.927639383155398e-05,
+      "loss": 2.0485,
+      "step": 2706
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.927243969948596e-05,
+      "loss": 2.0096,
+      "step": 2707
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9268485567417957e-05,
+      "loss": 2.1781,
+      "step": 2708
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9264531435349945e-05,
+      "loss": 2.1602,
+      "step": 2709
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9260577303281934e-05,
+      "loss": 2.3257,
+      "step": 2710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9256623171213916e-05,
+      "loss": 2.1784,
+      "step": 2711
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.925266903914591e-05,
+      "loss": 2.0675,
+      "step": 2712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.92487149070779e-05,
+      "loss": 1.8566,
+      "step": 2713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.924476077500989e-05,
+      "loss": 1.9566,
+      "step": 2714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.924080664294187e-05,
+      "loss": 2.3083,
+      "step": 2715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.923685251087387e-05,
+      "loss": 2.214,
+      "step": 2716
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9232898378805856e-05,
+      "loss": 1.9888,
+      "step": 2717
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.9228944246737845e-05,
+      "loss": 1.8544,
+      "step": 2718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.922499011466983e-05,
+      "loss": 2.218,
+      "step": 2719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.922103598260182e-05,
+      "loss": 2.076,
+      "step": 2720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.921708185053381e-05,
+      "loss": 2.0196,
+      "step": 2721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.92131277184658e-05,
+      "loss": 1.9723,
+      "step": 2722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.920917358639778e-05,
+      "loss": 2.0133,
+      "step": 2723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.920521945432978e-05,
+      "loss": 2.1353,
+      "step": 2724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.920126532226177e-05,
+      "loss": 2.2201,
+      "step": 2725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9197311190193755e-05,
+      "loss": 2.0833,
+      "step": 2726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9193357058125744e-05,
+      "loss": 2.2326,
+      "step": 2727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.918940292605773e-05,
+      "loss": 2.0104,
+      "step": 2728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.918544879398972e-05,
+      "loss": 2.1658,
+      "step": 2729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.918149466192171e-05,
+      "loss": 2.2237,
+      "step": 2730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.91775405298537e-05,
+      "loss": 1.9677,
+      "step": 2731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.917358639778569e-05,
+      "loss": 2.1838,
+      "step": 2732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.916963226571768e-05,
+      "loss": 2.1395,
+      "step": 2733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9165678133649666e-05,
+      "loss": 1.9206,
+      "step": 2734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9161724001581655e-05,
+      "loss": 1.9466,
+      "step": 2735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9157769869513644e-05,
+      "loss": 1.9837,
+      "step": 2736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.915381573744563e-05,
+      "loss": 2.0053,
+      "step": 2737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.914986160537762e-05,
+      "loss": 1.9454,
+      "step": 2738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.914590747330961e-05,
+      "loss": 2.0738,
+      "step": 2739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.91419533412416e-05,
+      "loss": 2.0415,
+      "step": 2740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.913799920917359e-05,
+      "loss": 2.2763,
+      "step": 2741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.913404507710558e-05,
+      "loss": 2.0473,
+      "step": 2742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9130090945037566e-05,
+      "loss": 2.1056,
+      "step": 2743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9126136812969554e-05,
+      "loss": 2.2193,
+      "step": 2744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.912218268090154e-05,
+      "loss": 1.9366,
+      "step": 2745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.911822854883353e-05,
+      "loss": 1.9941,
+      "step": 2746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.911427441676552e-05,
+      "loss": 2.1396,
+      "step": 2747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.911032028469751e-05,
+      "loss": 2.255,
+      "step": 2748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.91063661526295e-05,
+      "loss": 1.8502,
+      "step": 2749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9102412020561494e-05,
+      "loss": 2.177,
+      "step": 2750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9098457888493476e-05,
+      "loss": 2.025,
+      "step": 2751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9094503756425465e-05,
+      "loss": 1.9863,
+      "step": 2752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9090549624357454e-05,
+      "loss": 2.3773,
+      "step": 2753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.908659549228945e-05,
+      "loss": 2.0877,
+      "step": 2754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.908264136022143e-05,
+      "loss": 1.8884,
+      "step": 2755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.907868722815342e-05,
+      "loss": 2.0464,
+      "step": 2756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.907473309608541e-05,
+      "loss": 2.1351,
+      "step": 2757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9070778964017405e-05,
+      "loss": 2.2045,
+      "step": 2758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.906682483194939e-05,
+      "loss": 1.7499,
+      "step": 2759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9062870699881376e-05,
+      "loss": 2.0933,
+      "step": 2760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.905891656781337e-05,
+      "loss": 1.9474,
+      "step": 2761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.905496243574536e-05,
+      "loss": 2.3341,
+      "step": 2762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.905100830367734e-05,
+      "loss": 1.994,
+      "step": 2763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.904705417160933e-05,
+      "loss": 2.2174,
+      "step": 2764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9043100039541327e-05,
+      "loss": 2.227,
+      "step": 2765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9039145907473315e-05,
+      "loss": 2.2129,
+      "step": 2766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.90351917754053e-05,
+      "loss": 2.1186,
+      "step": 2767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9031237643337286e-05,
+      "loss": 2.1259,
+      "step": 2768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.902728351126928e-05,
+      "loss": 2.1862,
+      "step": 2769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.902332937920127e-05,
+      "loss": 1.763,
+      "step": 2770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.901937524713325e-05,
+      "loss": 2.05,
+      "step": 2771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.901542111506524e-05,
+      "loss": 1.8652,
+      "step": 2772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.901146698299724e-05,
+      "loss": 2.2568,
+      "step": 2773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.9007512850929226e-05,
+      "loss": 1.8198,
+      "step": 2774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.900355871886121e-05,
+      "loss": 2.0993,
+      "step": 2775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.89996045867932e-05,
+      "loss": 2.3792,
+      "step": 2776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.899565045472519e-05,
+      "loss": 2.1922,
+      "step": 2777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.899169632265718e-05,
+      "loss": 2.0057,
+      "step": 2778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.898774219058916e-05,
+      "loss": 1.8338,
+      "step": 2779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.898378805852116e-05,
+      "loss": 1.9407,
+      "step": 2780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.897983392645315e-05,
+      "loss": 1.9615,
+      "step": 2781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8975879794385137e-05,
+      "loss": 1.8847,
+      "step": 2782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.897192566231712e-05,
+      "loss": 2.0598,
+      "step": 2783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8967971530249114e-05,
+      "loss": 2.3181,
+      "step": 2784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.89640173981811e-05,
+      "loss": 1.9732,
+      "step": 2785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.896006326611309e-05,
+      "loss": 2.0982,
+      "step": 2786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8956109134045074e-05,
+      "loss": 2.0244,
+      "step": 2787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.895215500197707e-05,
+      "loss": 2.3774,
+      "step": 2788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.894820086990906e-05,
+      "loss": 1.9128,
+      "step": 2789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.894424673784105e-05,
+      "loss": 2.0402,
+      "step": 2790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.894029260577303e-05,
+      "loss": 1.8066,
+      "step": 2791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8936338473705025e-05,
+      "loss": 2.3647,
+      "step": 2792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8932384341637014e-05,
+      "loss": 2.2011,
+      "step": 2793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8928430209569e-05,
+      "loss": 2.1849,
+      "step": 2794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.892447607750099e-05,
+      "loss": 2.1382,
+      "step": 2795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.892052194543298e-05,
+      "loss": 2.3107,
+      "step": 2796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.891656781336497e-05,
+      "loss": 1.9788,
+      "step": 2797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.891261368129696e-05,
+      "loss": 2.1202,
+      "step": 2798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.890865954922895e-05,
+      "loss": 1.8792,
+      "step": 2799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8904705417160935e-05,
+      "loss": 2.0174,
+      "step": 2800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8900751285092924e-05,
+      "loss": 2.0586,
+      "step": 2801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.889679715302491e-05,
+      "loss": 2.1848,
+      "step": 2802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.88928430209569e-05,
+      "loss": 1.9751,
+      "step": 2803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 1.8404,
+      "step": 2804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.888493475682088e-05,
+      "loss": 2.0362,
+      "step": 2805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.888098062475287e-05,
+      "loss": 1.9829,
+      "step": 2806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.887702649268486e-05,
+      "loss": 2.096,
+      "step": 2807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8873072360616846e-05,
+      "loss": 2.187,
+      "step": 2808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8869118228548835e-05,
+      "loss": 1.9736,
+      "step": 2809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8865164096480824e-05,
+      "loss": 2.0704,
+      "step": 2810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.886120996441281e-05,
+      "loss": 2.1234,
+      "step": 2811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.88572558323448e-05,
+      "loss": 2.0416,
+      "step": 2812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.885330170027679e-05,
+      "loss": 2.1677,
+      "step": 2813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8849347568208786e-05,
+      "loss": 2.2166,
+      "step": 2814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.884539343614077e-05,
+      "loss": 2.0144,
+      "step": 2815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.884143930407276e-05,
+      "loss": 2.1815,
+      "step": 2816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8837485172004746e-05,
+      "loss": 2.057,
+      "step": 2817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.883353103993674e-05,
+      "loss": 2.1941,
+      "step": 2818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.882957690786872e-05,
+      "loss": 2.1345,
+      "step": 2819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.882562277580071e-05,
+      "loss": 2.115,
+      "step": 2820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.88216686437327e-05,
+      "loss": 2.1999,
+      "step": 2821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8817714511664696e-05,
+      "loss": 2.0327,
+      "step": 2822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.881376037959668e-05,
+      "loss": 2.2046,
+      "step": 2823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.880980624752867e-05,
+      "loss": 1.9374,
+      "step": 2824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8805852115460656e-05,
+      "loss": 2.2106,
+      "step": 2825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.880189798339265e-05,
+      "loss": 2.0665,
+      "step": 2826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8797943851324634e-05,
+      "loss": 2.014,
+      "step": 2827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.879398971925662e-05,
+      "loss": 2.1912,
+      "step": 2828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.879003558718861e-05,
+      "loss": 1.7987,
+      "step": 2829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.878608145512061e-05,
+      "loss": 1.9013,
+      "step": 2830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.878212732305259e-05,
+      "loss": 2.043,
+      "step": 2831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.877817319098458e-05,
+      "loss": 1.8779,
+      "step": 2832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8774219058916574e-05,
+      "loss": 2.0968,
+      "step": 2833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.877026492684856e-05,
+      "loss": 2.117,
+      "step": 2834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8766310794780544e-05,
+      "loss": 1.8647,
+      "step": 2835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.876235666271253e-05,
+      "loss": 2.0549,
+      "step": 2836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.875840253064453e-05,
+      "loss": 2.0664,
+      "step": 2837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.875444839857652e-05,
+      "loss": 2.276,
+      "step": 2838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.87504942665085e-05,
+      "loss": 2.2546,
+      "step": 2839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.874654013444049e-05,
+      "loss": 2.177,
+      "step": 2840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8742586002372484e-05,
+      "loss": 2.0655,
+      "step": 2841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.873863187030447e-05,
+      "loss": 1.991,
+      "step": 2842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8734677738236455e-05,
+      "loss": 2.2531,
+      "step": 2843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.8730723606168444e-05,
+      "loss": 1.9147,
+      "step": 2844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.872676947410044e-05,
+      "loss": 2.0921,
+      "step": 2845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.872281534203243e-05,
+      "loss": 2.273,
+      "step": 2846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.871886120996441e-05,
+      "loss": 1.845,
+      "step": 2847
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8714907077896406e-05,
+      "loss": 1.9428,
+      "step": 2848
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8710952945828395e-05,
+      "loss": 2.2635,
+      "step": 2849
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8706998813760384e-05,
+      "loss": 1.9708,
+      "step": 2850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8703044681692366e-05,
+      "loss": 2.0094,
+      "step": 2851
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.869909054962436e-05,
+      "loss": 2.1593,
+      "step": 2852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.869513641755635e-05,
+      "loss": 1.9921,
+      "step": 2853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.869118228548834e-05,
+      "loss": 2.109,
+      "step": 2854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.868722815342032e-05,
+      "loss": 2.1718,
+      "step": 2855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8683274021352317e-05,
+      "loss": 2.3272,
+      "step": 2856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8679319889284305e-05,
+      "loss": 1.932,
+      "step": 2857
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8675365757216294e-05,
+      "loss": 2.1561,
+      "step": 2858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8671411625148276e-05,
+      "loss": 2.0867,
+      "step": 2859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.866745749308027e-05,
+      "loss": 2.062,
+      "step": 2860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.866350336101226e-05,
+      "loss": 2.3113,
+      "step": 2861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.865954922894425e-05,
+      "loss": 2.0134,
+      "step": 2862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.865559509687623e-05,
+      "loss": 2.1398,
+      "step": 2863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.865164096480823e-05,
+      "loss": 2.0791,
+      "step": 2864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8647686832740216e-05,
+      "loss": 1.9502,
+      "step": 2865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8643732700672205e-05,
+      "loss": 2.1152,
+      "step": 2866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8639778568604194e-05,
+      "loss": 2.0417,
+      "step": 2867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.863582443653618e-05,
+      "loss": 1.9899,
+      "step": 2868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.863187030446817e-05,
+      "loss": 2.0992,
+      "step": 2869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.862791617240016e-05,
+      "loss": 2.0481,
+      "step": 2870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.862396204033215e-05,
+      "loss": 2.0811,
+      "step": 2871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.862000790826414e-05,
+      "loss": 2.2032,
+      "step": 2872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8616053776196127e-05,
+      "loss": 1.9679,
+      "step": 2873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8612099644128115e-05,
+      "loss": 1.8967,
+      "step": 2874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8608145512060104e-05,
+      "loss": 2.0348,
+      "step": 2875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.860419137999209e-05,
+      "loss": 2.0735,
+      "step": 2876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.860023724792408e-05,
+      "loss": 2.0992,
+      "step": 2877
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.859628311585607e-05,
+      "loss": 2.2837,
+      "step": 2878
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.859232898378806e-05,
+      "loss": 2.0914,
+      "step": 2879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.858837485172005e-05,
+      "loss": 2.2546,
+      "step": 2880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.858442071965204e-05,
+      "loss": 2.2913,
+      "step": 2881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.858046658758403e-05,
+      "loss": 2.4242,
+      "step": 2882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8576512455516015e-05,
+      "loss": 2.0064,
+      "step": 2883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8572558323448004e-05,
+      "loss": 2.2361,
+      "step": 2884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.856860419137999e-05,
+      "loss": 2.0869,
+      "step": 2885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.856465005931199e-05,
+      "loss": 2.0433,
+      "step": 2886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.856069592724397e-05,
+      "loss": 2.2049,
+      "step": 2887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.855674179517596e-05,
+      "loss": 1.8688,
+      "step": 2888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.855278766310795e-05,
+      "loss": 2.0932,
+      "step": 2889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8548833531039943e-05,
+      "loss": 1.7947,
+      "step": 2890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8544879398971925e-05,
+      "loss": 1.9116,
+      "step": 2891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8540925266903914e-05,
+      "loss": 2.0222,
+      "step": 2892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.85369711348359e-05,
+      "loss": 2.2045,
+      "step": 2893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.85330170027679e-05,
+      "loss": 2.0945,
+      "step": 2894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.852906287069988e-05,
+      "loss": 2.0142,
+      "step": 2895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.852510873863187e-05,
+      "loss": 2.0762,
+      "step": 2896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.852115460656386e-05,
+      "loss": 2.1258,
+      "step": 2897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8517200474495854e-05,
+      "loss": 1.9779,
+      "step": 2898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8513246342427836e-05,
+      "loss": 2.2535,
+      "step": 2899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8509292210359825e-05,
+      "loss": 2.2444,
+      "step": 2900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.850533807829182e-05,
+      "loss": 2.3633,
+      "step": 2901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.850138394622381e-05,
+      "loss": 2.0297,
+      "step": 2902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.849742981415579e-05,
+      "loss": 2.1344,
+      "step": 2903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.849347568208778e-05,
+      "loss": 2.1785,
+      "step": 2904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8489521550019776e-05,
+      "loss": 2.0092,
+      "step": 2905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8485567417951765e-05,
+      "loss": 1.8819,
+      "step": 2906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.848161328588375e-05,
+      "loss": 2.3313,
+      "step": 2907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8477659153815736e-05,
+      "loss": 1.8997,
+      "step": 2908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.847370502174773e-05,
+      "loss": 2.3442,
+      "step": 2909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.846975088967972e-05,
+      "loss": 2.0151,
+      "step": 2910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.84657967576117e-05,
+      "loss": 2.1871,
+      "step": 2911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.846184262554369e-05,
+      "loss": 2.1615,
+      "step": 2912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8457888493475686e-05,
+      "loss": 2.0028,
+      "step": 2913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8453934361407675e-05,
+      "loss": 1.8481,
+      "step": 2914
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.844998022933966e-05,
+      "loss": 2.1006,
+      "step": 2915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8446026097271646e-05,
+      "loss": 2.2385,
+      "step": 2916
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.844207196520364e-05,
+      "loss": 2.1758,
+      "step": 2917
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.843811783313563e-05,
+      "loss": 2.1717,
+      "step": 2918
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.843416370106761e-05,
+      "loss": 2.3134,
+      "step": 2919
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.843020956899961e-05,
+      "loss": 2.1201,
+      "step": 2920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.84262554369316e-05,
+      "loss": 2.0612,
+      "step": 2921
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8422301304863586e-05,
+      "loss": 1.8117,
+      "step": 2922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.841834717279557e-05,
+      "loss": 1.972,
+      "step": 2923
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8414393040727564e-05,
+      "loss": 1.8907,
+      "step": 2924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.841043890865955e-05,
+      "loss": 2.0163,
+      "step": 2925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.840648477659154e-05,
+      "loss": 2.075,
+      "step": 2926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.840253064452353e-05,
+      "loss": 2.2695,
+      "step": 2927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.839857651245552e-05,
+      "loss": 2.0122,
+      "step": 2928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.839462238038751e-05,
+      "loss": 1.9387,
+      "step": 2929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8390668248319497e-05,
+      "loss": 2.2532,
+      "step": 2930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8386714116251485e-05,
+      "loss": 2.0071,
+      "step": 2931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8382759984183474e-05,
+      "loss": 1.9274,
+      "step": 2932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.837880585211546e-05,
+      "loss": 2.0238,
+      "step": 2933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.837485172004745e-05,
+      "loss": 2.1386,
+      "step": 2934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.837089758797944e-05,
+      "loss": 2.0958,
+      "step": 2935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.836694345591143e-05,
+      "loss": 2.1078,
+      "step": 2936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.836298932384342e-05,
+      "loss": 2.2493,
+      "step": 2937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.835903519177541e-05,
+      "loss": 2.0861,
+      "step": 2938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8355081059707396e-05,
+      "loss": 2.0928,
+      "step": 2939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8351126927639385e-05,
+      "loss": 2.1422,
+      "step": 2940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8347172795571374e-05,
+      "loss": 1.9769,
+      "step": 2941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.834321866350336e-05,
+      "loss": 2.1578,
+      "step": 2942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.833926453143535e-05,
+      "loss": 1.9122,
+      "step": 2943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.833531039936734e-05,
+      "loss": 1.8855,
+      "step": 2944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.833135626729933e-05,
+      "loss": 2.0496,
+      "step": 2945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.832740213523132e-05,
+      "loss": 1.9701,
+      "step": 2946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8323448003163307e-05,
+      "loss": 2.2271,
+      "step": 2947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8319493871095295e-05,
+      "loss": 1.9942,
+      "step": 2948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8315539739027284e-05,
+      "loss": 2.1148,
+      "step": 2949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.831158560695927e-05,
+      "loss": 1.9426,
+      "step": 2950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.830763147489126e-05,
+      "loss": 1.9978,
+      "step": 2951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.830367734282325e-05,
+      "loss": 2.4771,
+      "step": 2952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.829972321075524e-05,
+      "loss": 1.8648,
+      "step": 2953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8295769078687235e-05,
+      "loss": 2.0698,
+      "step": 2954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.829181494661922e-05,
+      "loss": 1.9564,
+      "step": 2955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8287860814551206e-05,
+      "loss": 1.8485,
+      "step": 2956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8283906682483195e-05,
+      "loss": 2.339,
+      "step": 2957
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.827995255041519e-05,
+      "loss": 1.9371,
+      "step": 2958
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.827599841834717e-05,
+      "loss": 2.0506,
+      "step": 2959
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.827204428627916e-05,
+      "loss": 2.3263,
+      "step": 2960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.826809015421115e-05,
+      "loss": 2.297,
+      "step": 2961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8264136022143146e-05,
+      "loss": 2.2403,
+      "step": 2962
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.826018189007513e-05,
+      "loss": 2.0297,
+      "step": 2963
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.825622775800712e-05,
+      "loss": 2.0204,
+      "step": 2964
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8252273625939105e-05,
+      "loss": 2.2868,
+      "step": 2965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.82483194938711e-05,
+      "loss": 2.0636,
+      "step": 2966
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.824436536180308e-05,
+      "loss": 2.4771,
+      "step": 2967
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.824041122973507e-05,
+      "loss": 2.2456,
+      "step": 2968
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.823645709766707e-05,
+      "loss": 2.3298,
+      "step": 2969
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.8232502965599056e-05,
+      "loss": 1.8628,
+      "step": 2970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.822854883353104e-05,
+      "loss": 2.1003,
+      "step": 2971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.822459470146303e-05,
+      "loss": 1.8802,
+      "step": 2972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.822064056939502e-05,
+      "loss": 1.7676,
+      "step": 2973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.821668643732701e-05,
+      "loss": 1.9405,
+      "step": 2974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8212732305258994e-05,
+      "loss": 2.1526,
+      "step": 2975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.820877817319098e-05,
+      "loss": 1.9769,
+      "step": 2976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.820482404112298e-05,
+      "loss": 2.1309,
+      "step": 2977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.820086990905497e-05,
+      "loss": 2.3781,
+      "step": 2978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.819691577698695e-05,
+      "loss": 2.1444,
+      "step": 2979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.819296164491894e-05,
+      "loss": 2.0403,
+      "step": 2980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8189007512850933e-05,
+      "loss": 1.847,
+      "step": 2981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.818505338078292e-05,
+      "loss": 2.2469,
+      "step": 2982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.818109924871491e-05,
+      "loss": 2.0317,
+      "step": 2983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.817714511664689e-05,
+      "loss": 2.0234,
+      "step": 2984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.817319098457889e-05,
+      "loss": 1.8418,
+      "step": 2985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.816923685251088e-05,
+      "loss": 2.146,
+      "step": 2986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8165282720442866e-05,
+      "loss": 2.0309,
+      "step": 2987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8161328588374855e-05,
+      "loss": 2.001,
+      "step": 2988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8157374456306844e-05,
+      "loss": 2.2603,
+      "step": 2989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.815342032423883e-05,
+      "loss": 2.337,
+      "step": 2990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.814946619217082e-05,
+      "loss": 1.9395,
+      "step": 2991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.814551206010281e-05,
+      "loss": 2.3305,
+      "step": 2992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.81415579280348e-05,
+      "loss": 2.199,
+      "step": 2993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.813760379596679e-05,
+      "loss": 2.1309,
+      "step": 2994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.813364966389878e-05,
+      "loss": 2.1714,
+      "step": 2995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8129695531830766e-05,
+      "loss": 2.379,
+      "step": 2996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8125741399762755e-05,
+      "loss": 1.8593,
+      "step": 2997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8121787267694744e-05,
+      "loss": 2.1133,
+      "step": 2998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.811783313562673e-05,
+      "loss": 2.0323,
+      "step": 2999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.811387900355872e-05,
+      "loss": 2.0009,
+      "step": 3000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.810992487149071e-05,
+      "loss": 2.176,
+      "step": 3001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.81059707394227e-05,
+      "loss": 1.9871,
+      "step": 3002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.810201660735469e-05,
+      "loss": 1.7269,
+      "step": 3003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8098062475286676e-05,
+      "loss": 2.045,
+      "step": 3004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8094108343218665e-05,
+      "loss": 1.8104,
+      "step": 3005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8090154211150654e-05,
+      "loss": 1.8466,
+      "step": 3006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.808620007908264e-05,
+      "loss": 2.2195,
+      "step": 3007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.808224594701463e-05,
+      "loss": 2.0167,
+      "step": 3008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.807829181494662e-05,
+      "loss": 2.0539,
+      "step": 3009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.807433768287861e-05,
+      "loss": 2.0517,
+      "step": 3010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.80703835508106e-05,
+      "loss": 2.2441,
+      "step": 3011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.806642941874259e-05,
+      "loss": 2.1763,
+      "step": 3012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8062475286674576e-05,
+      "loss": 2.1773,
+      "step": 3013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8058521154606565e-05,
+      "loss": 2.028,
+      "step": 3014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8054567022538554e-05,
+      "loss": 2.0671,
+      "step": 3015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.805061289047054e-05,
+      "loss": 2.0324,
+      "step": 3016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.804665875840253e-05,
+      "loss": 2.2751,
+      "step": 3017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.804270462633452e-05,
+      "loss": 2.4887,
+      "step": 3018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.803875049426651e-05,
+      "loss": 2.0214,
+      "step": 3019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.80347963621985e-05,
+      "loss": 1.9691,
+      "step": 3020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8030842230130487e-05,
+      "loss": 1.999,
+      "step": 3021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.802688809806248e-05,
+      "loss": 1.93,
+      "step": 3022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.8022933965994464e-05,
+      "loss": 2.2508,
+      "step": 3023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.801897983392645e-05,
+      "loss": 2.2654,
+      "step": 3024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.801502570185844e-05,
+      "loss": 1.8715,
+      "step": 3025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.801107156979044e-05,
+      "loss": 1.8549,
+      "step": 3026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.800711743772242e-05,
+      "loss": 2.324,
+      "step": 3027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.800316330565441e-05,
+      "loss": 1.903,
+      "step": 3028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.79992091735864e-05,
+      "loss": 1.9585,
+      "step": 3029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.799525504151839e-05,
+      "loss": 2.0406,
+      "step": 3030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7991300909450375e-05,
+      "loss": 1.8821,
+      "step": 3031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7987346777382364e-05,
+      "loss": 2.0182,
+      "step": 3032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.798339264531435e-05,
+      "loss": 1.9627,
+      "step": 3033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.797943851324635e-05,
+      "loss": 2.0688,
+      "step": 3034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.797548438117833e-05,
+      "loss": 1.9878,
+      "step": 3035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.797153024911032e-05,
+      "loss": 2.4249,
+      "step": 3036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.796757611704231e-05,
+      "loss": 1.9271,
+      "step": 3037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.79636219849743e-05,
+      "loss": 2.1965,
+      "step": 3038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7959667852906285e-05,
+      "loss": 2.006,
+      "step": 3039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7955713720838274e-05,
+      "loss": 2.4484,
+      "step": 3040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.795175958877027e-05,
+      "loss": 2.2246,
+      "step": 3041
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.794780545670226e-05,
+      "loss": 1.988,
+      "step": 3042
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.794385132463425e-05,
+      "loss": 2.0773,
+      "step": 3043
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.793989719256623e-05,
+      "loss": 1.9985,
+      "step": 3044
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7935943060498225e-05,
+      "loss": 2.2173,
+      "step": 3045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7931988928430214e-05,
+      "loss": 1.9039,
+      "step": 3046
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.79280347963622e-05,
+      "loss": 2.2535,
+      "step": 3047
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7924080664294185e-05,
+      "loss": 2.2704,
+      "step": 3048
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.792012653222618e-05,
+      "loss": 2.1005,
+      "step": 3049
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.791617240015817e-05,
+      "loss": 1.9712,
+      "step": 3050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.791221826809016e-05,
+      "loss": 2.2914,
+      "step": 3051
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.790826413602214e-05,
+      "loss": 2.1124,
+      "step": 3052
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7904310003954136e-05,
+      "loss": 2.1686,
+      "step": 3053
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7900355871886125e-05,
+      "loss": 2.1357,
+      "step": 3054
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7896401739818113e-05,
+      "loss": 2.3517,
+      "step": 3055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.78924476077501e-05,
+      "loss": 2.0255,
+      "step": 3056
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.788849347568209e-05,
+      "loss": 2.0883,
+      "step": 3057
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.788453934361408e-05,
+      "loss": 2.4023,
+      "step": 3058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.788058521154607e-05,
+      "loss": 2.25,
+      "step": 3059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.787663107947806e-05,
+      "loss": 2.154,
+      "step": 3060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7872676947410046e-05,
+      "loss": 2.2959,
+      "step": 3061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7868722815342035e-05,
+      "loss": 1.8417,
+      "step": 3062
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7864768683274024e-05,
+      "loss": 1.9167,
+      "step": 3063
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.786081455120601e-05,
+      "loss": 2.34,
+      "step": 3064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7856860419138e-05,
+      "loss": 1.9737,
+      "step": 3065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.785290628706999e-05,
+      "loss": 2.0037,
+      "step": 3066
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.784895215500198e-05,
+      "loss": 2.219,
+      "step": 3067
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.784499802293397e-05,
+      "loss": 1.9916,
+      "step": 3068
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.784104389086596e-05,
+      "loss": 2.0479,
+      "step": 3069
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7837089758797946e-05,
+      "loss": 2.0023,
+      "step": 3070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7833135626729935e-05,
+      "loss": 1.9758,
+      "step": 3071
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7829181494661923e-05,
+      "loss": 1.911,
+      "step": 3072
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.782522736259391e-05,
+      "loss": 1.9744,
+      "step": 3073
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.78212732305259e-05,
+      "loss": 1.9663,
+      "step": 3074
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.781731909845789e-05,
+      "loss": 2.0647,
+      "step": 3075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.781336496638988e-05,
+      "loss": 2.0947,
+      "step": 3076
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.780941083432187e-05,
+      "loss": 1.8406,
+      "step": 3077
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7805456702253856e-05,
+      "loss": 1.9992,
+      "step": 3078
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7801502570185845e-05,
+      "loss": 1.9608,
+      "step": 3079
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7797548438117834e-05,
+      "loss": 1.9382,
+      "step": 3080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.779359430604982e-05,
+      "loss": 2.0367,
+      "step": 3081
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.778964017398181e-05,
+      "loss": 2.1309,
+      "step": 3082
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.77856860419138e-05,
+      "loss": 2.2042,
+      "step": 3083
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.778173190984579e-05,
+      "loss": 2.0622,
+      "step": 3084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.777777777777778e-05,
+      "loss": 2.1075,
+      "step": 3085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.777382364570977e-05,
+      "loss": 2.3593,
+      "step": 3086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7769869513641756e-05,
+      "loss": 2.1651,
+      "step": 3087
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7765915381573745e-05,
+      "loss": 1.99,
+      "step": 3088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7761961249505734e-05,
+      "loss": 1.9763,
+      "step": 3089
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.775800711743772e-05,
+      "loss": 2.1038,
+      "step": 3090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.775405298536971e-05,
+      "loss": 1.9469,
+      "step": 3091
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.77500988533017e-05,
+      "loss": 2.1631,
+      "step": 3092
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.774614472123369e-05,
+      "loss": 1.9978,
+      "step": 3093
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7742190589165684e-05,
+      "loss": 1.9595,
+      "step": 3094
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7738236457097667e-05,
+      "loss": 2.0739,
+      "step": 3095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7734282325029655e-05,
+      "loss": 1.9281,
+      "step": 3096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.7730328192961644e-05,
+      "loss": 2.2191,
+      "step": 3097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.772637406089364e-05,
+      "loss": 2.2014,
+      "step": 3098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.772241992882563e-05,
+      "loss": 2.1319,
+      "step": 3099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.771846579675761e-05,
+      "loss": 2.3297,
+      "step": 3100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.77145116646896e-05,
+      "loss": 2.0991,
+      "step": 3101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7710557532621595e-05,
+      "loss": 2.1573,
+      "step": 3102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7706603400553584e-05,
+      "loss": 2.0663,
+      "step": 3103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7702649268485566e-05,
+      "loss": 2.1252,
+      "step": 3104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7698695136417555e-05,
+      "loss": 2.1901,
+      "step": 3105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.769474100434955e-05,
+      "loss": 1.8661,
+      "step": 3106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.769078687228154e-05,
+      "loss": 2.2275,
+      "step": 3107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.768683274021352e-05,
+      "loss": 2.0647,
+      "step": 3108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.768287860814552e-05,
+      "loss": 1.879,
+      "step": 3109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7678924476077506e-05,
+      "loss": 2.1004,
+      "step": 3110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7674970344009495e-05,
+      "loss": 1.8893,
+      "step": 3111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7671016211941477e-05,
+      "loss": 2.078,
+      "step": 3112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.766706207987347e-05,
+      "loss": 2.1116,
+      "step": 3113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.766310794780546e-05,
+      "loss": 2.1474,
+      "step": 3114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.765915381573745e-05,
+      "loss": 2.0069,
+      "step": 3115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.765519968366943e-05,
+      "loss": 1.9496,
+      "step": 3116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.765124555160143e-05,
+      "loss": 2.0843,
+      "step": 3117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7647291419533416e-05,
+      "loss": 1.8827,
+      "step": 3118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7643337287465405e-05,
+      "loss": 1.986,
+      "step": 3119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.763938315539739e-05,
+      "loss": 2.1746,
+      "step": 3120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.763542902332938e-05,
+      "loss": 2.1572,
+      "step": 3121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.763147489126137e-05,
+      "loss": 2.0641,
+      "step": 3122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.762752075919336e-05,
+      "loss": 2.148,
+      "step": 3123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.762356662712534e-05,
+      "loss": 2.1574,
+      "step": 3124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.761961249505734e-05,
+      "loss": 2.0283,
+      "step": 3125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.761565836298933e-05,
+      "loss": 2.0927,
+      "step": 3126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7611704230921316e-05,
+      "loss": 2.0399,
+      "step": 3127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7607750098853305e-05,
+      "loss": 2.0046,
+      "step": 3128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7603795966785293e-05,
+      "loss": 2.0419,
+      "step": 3129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.759984183471728e-05,
+      "loss": 1.9481,
+      "step": 3130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.759588770264927e-05,
+      "loss": 2.2056,
+      "step": 3131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.759193357058126e-05,
+      "loss": 1.9515,
+      "step": 3132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.758797943851325e-05,
+      "loss": 2.1519,
+      "step": 3133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.758402530644524e-05,
+      "loss": 2.1466,
+      "step": 3134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7580071174377226e-05,
+      "loss": 1.8388,
+      "step": 3135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7576117042309215e-05,
+      "loss": 2.1446,
+      "step": 3136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7572162910241204e-05,
+      "loss": 2.1618,
+      "step": 3137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.756820877817319e-05,
+      "loss": 1.9849,
+      "step": 3138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.756425464610518e-05,
+      "loss": 2.2559,
+      "step": 3139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.756030051403717e-05,
+      "loss": 2.1869,
+      "step": 3140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.755634638196916e-05,
+      "loss": 2.0881,
+      "step": 3141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.755239224990115e-05,
+      "loss": 1.9263,
+      "step": 3142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.754843811783314e-05,
+      "loss": 2.243,
+      "step": 3143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7544483985765126e-05,
+      "loss": 1.9869,
+      "step": 3144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7540529853697115e-05,
+      "loss": 2.1901,
+      "step": 3145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7536575721629103e-05,
+      "loss": 1.9409,
+      "step": 3146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.753262158956109e-05,
+      "loss": 1.9964,
+      "step": 3147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.752866745749308e-05,
+      "loss": 2.0324,
+      "step": 3148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.752471332542507e-05,
+      "loss": 2.1078,
+      "step": 3149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.752075919335706e-05,
+      "loss": 2.0772,
+      "step": 3150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.751680506128905e-05,
+      "loss": 2.1745,
+      "step": 3151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7512850929221036e-05,
+      "loss": 1.9246,
+      "step": 3152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7508896797153025e-05,
+      "loss": 1.9243,
+      "step": 3153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7504942665085014e-05,
+      "loss": 2.0551,
+      "step": 3154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.750098853301701e-05,
+      "loss": 1.9521,
+      "step": 3155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.749703440094899e-05,
+      "loss": 1.999,
+      "step": 3156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.749308026888098e-05,
+      "loss": 1.9219,
+      "step": 3157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.748912613681297e-05,
+      "loss": 2.2365,
+      "step": 3158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7485172004744965e-05,
+      "loss": 2.44,
+      "step": 3159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.748121787267695e-05,
+      "loss": 2.0461,
+      "step": 3160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7477263740608936e-05,
+      "loss": 2.2152,
+      "step": 3161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.747330960854093e-05,
+      "loss": 2.1219,
+      "step": 3162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.746935547647292e-05,
+      "loss": 2.0587,
+      "step": 3163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.74654013444049e-05,
+      "loss": 2.2096,
+      "step": 3164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.746144721233689e-05,
+      "loss": 2.0162,
+      "step": 3165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.745749308026889e-05,
+      "loss": 1.9892,
+      "step": 3166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7453538948200876e-05,
+      "loss": 1.9845,
+      "step": 3167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.744958481613286e-05,
+      "loss": 2.0294,
+      "step": 3168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7445630684064846e-05,
+      "loss": 2.2454,
+      "step": 3169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.744167655199684e-05,
+      "loss": 2.2355,
+      "step": 3170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.743772241992883e-05,
+      "loss": 2.0482,
+      "step": 3171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.743376828786081e-05,
+      "loss": 1.8415,
+      "step": 3172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.74298141557928e-05,
+      "loss": 2.1775,
+      "step": 3173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.74258600237248e-05,
+      "loss": 2.5316,
+      "step": 3174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7421905891656786e-05,
+      "loss": 2.1402,
+      "step": 3175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.741795175958877e-05,
+      "loss": 2.3858,
+      "step": 3176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7413997627520764e-05,
+      "loss": 2.1547,
+      "step": 3177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.741004349545275e-05,
+      "loss": 1.9543,
+      "step": 3178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.740608936338474e-05,
+      "loss": 1.7588,
+      "step": 3179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7402135231316724e-05,
+      "loss": 2.0138,
+      "step": 3180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.739818109924872e-05,
+      "loss": 2.3848,
+      "step": 3181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.739422696718071e-05,
+      "loss": 1.9896,
+      "step": 3182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.73902728351127e-05,
+      "loss": 2.0706,
+      "step": 3183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.738631870304468e-05,
+      "loss": 2.0421,
+      "step": 3184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7382364570976674e-05,
+      "loss": 1.9149,
+      "step": 3185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.737841043890866e-05,
+      "loss": 2.2152,
+      "step": 3186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.737445630684065e-05,
+      "loss": 2.0001,
+      "step": 3187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7370502174772634e-05,
+      "loss": 2.1787,
+      "step": 3188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.736654804270463e-05,
+      "loss": 2.005,
+      "step": 3189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.736259391063662e-05,
+      "loss": 1.9781,
+      "step": 3190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.735863977856861e-05,
+      "loss": 2.0554,
+      "step": 3191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.735468564650059e-05,
+      "loss": 1.9465,
+      "step": 3192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7350731514432585e-05,
+      "loss": 1.9776,
+      "step": 3193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7346777382364574e-05,
+      "loss": 1.8292,
+      "step": 3194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.734282325029656e-05,
+      "loss": 1.9692,
+      "step": 3195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.733886911822855e-05,
+      "loss": 2.2352,
+      "step": 3196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.733491498616054e-05,
+      "loss": 2.5312,
+      "step": 3197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.733096085409253e-05,
+      "loss": 1.9335,
+      "step": 3198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.732700672202452e-05,
+      "loss": 2.1067,
+      "step": 3199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.732305258995651e-05,
+      "loss": 2.1023,
+      "step": 3200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7319098457888496e-05,
+      "loss": 2.0295,
+      "step": 3201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7315144325820485e-05,
+      "loss": 2.127,
+      "step": 3202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.731119019375247e-05,
+      "loss": 2.0111,
+      "step": 3203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.730723606168446e-05,
+      "loss": 1.992,
+      "step": 3204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.730328192961645e-05,
+      "loss": 2.1056,
+      "step": 3205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.729932779754844e-05,
+      "loss": 2.0461,
+      "step": 3206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.729537366548043e-05,
+      "loss": 2.0316,
+      "step": 3207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.729141953341242e-05,
+      "loss": 2.3381,
+      "step": 3208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7287465401344406e-05,
+      "loss": 1.9812,
+      "step": 3209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7283511269276395e-05,
+      "loss": 1.9795,
+      "step": 3210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7279557137208384e-05,
+      "loss": 1.8575,
+      "step": 3211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.727560300514037e-05,
+      "loss": 2.1287,
+      "step": 3212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.727164887307236e-05,
+      "loss": 2.1388,
+      "step": 3213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.726769474100435e-05,
+      "loss": 2.1093,
+      "step": 3214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7263740608936346e-05,
+      "loss": 1.9168,
+      "step": 3215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.725978647686833e-05,
+      "loss": 1.8889,
+      "step": 3216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.725583234480032e-05,
+      "loss": 2.0553,
+      "step": 3217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7251878212732306e-05,
+      "loss": 1.8764,
+      "step": 3218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.72479240806643e-05,
+      "loss": 2.188,
+      "step": 3219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.7243969948596283e-05,
+      "loss": 1.7997,
+      "step": 3220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.724001581652827e-05,
+      "loss": 2.2556,
+      "step": 3221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.723606168446026e-05,
+      "loss": 2.2371,
+      "step": 3222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.723210755239226e-05,
+      "loss": 2.0945,
+      "step": 3223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.722815342032424e-05,
+      "loss": 2.0734,
+      "step": 3224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.722419928825623e-05,
+      "loss": 2.1459,
+      "step": 3225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7220245156188216e-05,
+      "loss": 2.182,
+      "step": 3226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.721629102412021e-05,
+      "loss": 2.0643,
+      "step": 3227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7212336892052194e-05,
+      "loss": 2.0204,
+      "step": 3228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.720838275998418e-05,
+      "loss": 2.0034,
+      "step": 3229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.720442862791618e-05,
+      "loss": 1.7331,
+      "step": 3230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.720047449584817e-05,
+      "loss": 2.2365,
+      "step": 3231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.719652036378015e-05,
+      "loss": 2.3697,
+      "step": 3232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.719256623171214e-05,
+      "loss": 1.7983,
+      "step": 3233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7188612099644134e-05,
+      "loss": 2.0292,
+      "step": 3234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.718465796757612e-05,
+      "loss": 1.9554,
+      "step": 3235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7180703835508105e-05,
+      "loss": 2.2339,
+      "step": 3236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7176749703440093e-05,
+      "loss": 2.056,
+      "step": 3237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.717279557137209e-05,
+      "loss": 1.955,
+      "step": 3238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.716884143930408e-05,
+      "loss": 2.0233,
+      "step": 3239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.716488730723606e-05,
+      "loss": 2.1399,
+      "step": 3240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.716093317516805e-05,
+      "loss": 2.0281,
+      "step": 3241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7156979043100044e-05,
+      "loss": 2.3115,
+      "step": 3242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.715302491103203e-05,
+      "loss": 2.1165,
+      "step": 3243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7149070778964015e-05,
+      "loss": 2.3737,
+      "step": 3244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7145116646896004e-05,
+      "loss": 2.1192,
+      "step": 3245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7141162514828e-05,
+      "loss": 1.8961,
+      "step": 3246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.713720838275999e-05,
+      "loss": 1.904,
+      "step": 3247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.713325425069197e-05,
+      "loss": 2.0754,
+      "step": 3248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7129300118623966e-05,
+      "loss": 2.2589,
+      "step": 3249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7125345986555955e-05,
+      "loss": 2.0484,
+      "step": 3250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7121391854487944e-05,
+      "loss": 2.0559,
+      "step": 3251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7117437722419926e-05,
+      "loss": 2.32,
+      "step": 3252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.711348359035192e-05,
+      "loss": 2.0,
+      "step": 3253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.710952945828391e-05,
+      "loss": 2.0534,
+      "step": 3254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.71055753262159e-05,
+      "loss": 2.1082,
+      "step": 3255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.710162119414788e-05,
+      "loss": 1.9924,
+      "step": 3256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.709766706207988e-05,
+      "loss": 1.9798,
+      "step": 3257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7093712930011866e-05,
+      "loss": 1.8352,
+      "step": 3258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7089758797943854e-05,
+      "loss": 2.0057,
+      "step": 3259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7085804665875837e-05,
+      "loss": 1.9071,
+      "step": 3260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.708185053380783e-05,
+      "loss": 2.0397,
+      "step": 3261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.707789640173982e-05,
+      "loss": 1.7977,
+      "step": 3262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.707394226967181e-05,
+      "loss": 2.2886,
+      "step": 3263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.70699881376038e-05,
+      "loss": 2.0903,
+      "step": 3264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.706603400553579e-05,
+      "loss": 2.0851,
+      "step": 3265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7062079873467776e-05,
+      "loss": 2.067,
+      "step": 3266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7058125741399765e-05,
+      "loss": 2.0873,
+      "step": 3267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7054171609331754e-05,
+      "loss": 2.0888,
+      "step": 3268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.705021747726374e-05,
+      "loss": 1.8873,
+      "step": 3269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.704626334519573e-05,
+      "loss": 1.8477,
+      "step": 3270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.704230921312772e-05,
+      "loss": 2.1256,
+      "step": 3271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.703835508105971e-05,
+      "loss": 2.0855,
+      "step": 3272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.70344009489917e-05,
+      "loss": 2.2231,
+      "step": 3273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.703044681692369e-05,
+      "loss": 2.1455,
+      "step": 3274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7026492684855676e-05,
+      "loss": 1.8377,
+      "step": 3275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.7022538552787665e-05,
+      "loss": 1.9834,
+      "step": 3276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.701858442071965e-05,
+      "loss": 2.0811,
+      "step": 3277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.701463028865164e-05,
+      "loss": 2.2462,
+      "step": 3278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.701067615658363e-05,
+      "loss": 1.981,
+      "step": 3279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.700672202451562e-05,
+      "loss": 1.9943,
+      "step": 3280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.700276789244761e-05,
+      "loss": 1.8786,
+      "step": 3281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.69988137603796e-05,
+      "loss": 2.1315,
+      "step": 3282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.699485962831159e-05,
+      "loss": 2.0118,
+      "step": 3283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6990905496243575e-05,
+      "loss": 2.0255,
+      "step": 3284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6986951364175564e-05,
+      "loss": 1.8736,
+      "step": 3285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.698299723210755e-05,
+      "loss": 2.3529,
+      "step": 3286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.697904310003955e-05,
+      "loss": 2.213,
+      "step": 3287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.697508896797153e-05,
+      "loss": 1.8926,
+      "step": 3288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.697113483590352e-05,
+      "loss": 1.9013,
+      "step": 3289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.696718070383551e-05,
+      "loss": 2.1907,
+      "step": 3290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6963226571767504e-05,
+      "loss": 2.1925,
+      "step": 3291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6959272439699486e-05,
+      "loss": 1.9778,
+      "step": 3292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6955318307631475e-05,
+      "loss": 2.0841,
+      "step": 3293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6951364175563463e-05,
+      "loss": 2.2214,
+      "step": 3294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.694741004349546e-05,
+      "loss": 2.1095,
+      "step": 3295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.694345591142744e-05,
+      "loss": 1.9831,
+      "step": 3296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.693950177935943e-05,
+      "loss": 2.1165,
+      "step": 3297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.693554764729142e-05,
+      "loss": 2.0373,
+      "step": 3298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6931593515223414e-05,
+      "loss": 1.9326,
+      "step": 3299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6927639383155396e-05,
+      "loss": 2.2294,
+      "step": 3300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6923685251087385e-05,
+      "loss": 2.014,
+      "step": 3301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.691973111901938e-05,
+      "loss": 2.1786,
+      "step": 3302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.691577698695137e-05,
+      "loss": 2.0589,
+      "step": 3303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.691182285488335e-05,
+      "loss": 2.1696,
+      "step": 3304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.690786872281534e-05,
+      "loss": 2.0353,
+      "step": 3305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6903914590747336e-05,
+      "loss": 2.2412,
+      "step": 3306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6899960458679325e-05,
+      "loss": 2.1877,
+      "step": 3307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.689600632661131e-05,
+      "loss": 2.0806,
+      "step": 3308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6892052194543296e-05,
+      "loss": 2.1383,
+      "step": 3309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.688809806247529e-05,
+      "loss": 2.1702,
+      "step": 3310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.688414393040728e-05,
+      "loss": 2.1617,
+      "step": 3311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.688018979833926e-05,
+      "loss": 1.9522,
+      "step": 3312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.687623566627125e-05,
+      "loss": 2.0845,
+      "step": 3313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.687228153420325e-05,
+      "loss": 2.0669,
+      "step": 3314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6868327402135236e-05,
+      "loss": 1.9635,
+      "step": 3315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.686437327006722e-05,
+      "loss": 1.9459,
+      "step": 3316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.686041913799921e-05,
+      "loss": 1.7161,
+      "step": 3317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.68564650059312e-05,
+      "loss": 1.9459,
+      "step": 3318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.685251087386319e-05,
+      "loss": 2.0021,
+      "step": 3319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.684855674179517e-05,
+      "loss": 2.0509,
+      "step": 3320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.684460260972717e-05,
+      "loss": 1.9187,
+      "step": 3321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.684064847765916e-05,
+      "loss": 1.9668,
+      "step": 3322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6836694345591146e-05,
+      "loss": 2.3108,
+      "step": 3323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.683274021352313e-05,
+      "loss": 1.9309,
+      "step": 3324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6828786081455124e-05,
+      "loss": 2.0705,
+      "step": 3325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.682483194938711e-05,
+      "loss": 2.1138,
+      "step": 3326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.68208778173191e-05,
+      "loss": 2.2183,
+      "step": 3327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6816923685251084e-05,
+      "loss": 1.9024,
+      "step": 3328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.681296955318308e-05,
+      "loss": 2.1235,
+      "step": 3329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.680901542111507e-05,
+      "loss": 1.9404,
+      "step": 3330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.680506128904706e-05,
+      "loss": 1.8942,
+      "step": 3331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6801107156979046e-05,
+      "loss": 2.1141,
+      "step": 3332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6797153024911034e-05,
+      "loss": 2.1418,
+      "step": 3333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.679319889284302e-05,
+      "loss": 2.1844,
+      "step": 3334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.678924476077501e-05,
+      "loss": 2.0404,
+      "step": 3335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6785290628707e-05,
+      "loss": 2.1008,
+      "step": 3336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.678133649663899e-05,
+      "loss": 1.6359,
+      "step": 3337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.677738236457098e-05,
+      "loss": 2.062,
+      "step": 3338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.677342823250297e-05,
+      "loss": 2.1554,
+      "step": 3339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6769474100434956e-05,
+      "loss": 2.0936,
+      "step": 3340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6765519968366945e-05,
+      "loss": 2.2673,
+      "step": 3341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6761565836298934e-05,
+      "loss": 2.1234,
+      "step": 3342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.675761170423092e-05,
+      "loss": 2.2088,
+      "step": 3343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.675365757216291e-05,
+      "loss": 1.9241,
+      "step": 3344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.67497034400949e-05,
+      "loss": 2.049,
+      "step": 3345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.674574930802689e-05,
+      "loss": 2.2414,
+      "step": 3346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.674179517595888e-05,
+      "loss": 2.0403,
+      "step": 3347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.673784104389087e-05,
+      "loss": 1.8833,
+      "step": 3348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6733886911822856e-05,
+      "loss": 2.2889,
+      "step": 3349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.6729932779754844e-05,
+      "loss": 2.2679,
+      "step": 3350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.672597864768684e-05,
+      "loss": 2.2861,
+      "step": 3351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.672202451561882e-05,
+      "loss": 2.2909,
+      "step": 3352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.671807038355081e-05,
+      "loss": 2.1545,
+      "step": 3353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.67141162514828e-05,
+      "loss": 1.8342,
+      "step": 3354
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6710162119414795e-05,
+      "loss": 2.136,
+      "step": 3355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.670620798734678e-05,
+      "loss": 2.0286,
+      "step": 3356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6702253855278766e-05,
+      "loss": 1.8894,
+      "step": 3357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6698299723210755e-05,
+      "loss": 1.9721,
+      "step": 3358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.669434559114275e-05,
+      "loss": 2.1592,
+      "step": 3359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.669039145907473e-05,
+      "loss": 1.9929,
+      "step": 3360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.668643732700672e-05,
+      "loss": 2.0794,
+      "step": 3361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.668248319493871e-05,
+      "loss": 2.094,
+      "step": 3362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6678529062870706e-05,
+      "loss": 2.1405,
+      "step": 3363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.667457493080269e-05,
+      "loss": 2.0633,
+      "step": 3364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.667062079873468e-05,
+      "loss": 1.9959,
+      "step": 3365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 2.1926,
+      "step": 3366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.666271253459866e-05,
+      "loss": 2.0764,
+      "step": 3367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.665875840253064e-05,
+      "loss": 2.1943,
+      "step": 3368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.665480427046263e-05,
+      "loss": 2.149,
+      "step": 3369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.665085013839463e-05,
+      "loss": 2.1536,
+      "step": 3370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.664689600632662e-05,
+      "loss": 2.2598,
+      "step": 3371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.66429418742586e-05,
+      "loss": 2.0404,
+      "step": 3372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.663898774219059e-05,
+      "loss": 2.2013,
+      "step": 3373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.663503361012258e-05,
+      "loss": 2.1127,
+      "step": 3374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.663107947805457e-05,
+      "loss": 1.869,
+      "step": 3375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6627125345986554e-05,
+      "loss": 2.1679,
+      "step": 3376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.662317121391854e-05,
+      "loss": 2.0584,
+      "step": 3377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.661921708185054e-05,
+      "loss": 1.9899,
+      "step": 3378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.661526294978253e-05,
+      "loss": 2.0224,
+      "step": 3379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.661130881771451e-05,
+      "loss": 1.8879,
+      "step": 3380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.66073546856465e-05,
+      "loss": 2.1082,
+      "step": 3381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6603400553578494e-05,
+      "loss": 1.932,
+      "step": 3382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.659944642151048e-05,
+      "loss": 1.984,
+      "step": 3383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6595492289442465e-05,
+      "loss": 2.1832,
+      "step": 3384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6591538157374453e-05,
+      "loss": 1.814,
+      "step": 3385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.658758402530645e-05,
+      "loss": 2.1577,
+      "step": 3386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.658362989323844e-05,
+      "loss": 2.0688,
+      "step": 3387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.657967576117042e-05,
+      "loss": 1.8884,
+      "step": 3388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6575721629102416e-05,
+      "loss": 2.3908,
+      "step": 3389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6571767497034404e-05,
+      "loss": 1.7788,
+      "step": 3390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.656781336496639e-05,
+      "loss": 2.0541,
+      "step": 3391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.656385923289838e-05,
+      "loss": 1.769,
+      "step": 3392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.655990510083037e-05,
+      "loss": 2.0724,
+      "step": 3393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.655595096876236e-05,
+      "loss": 2.1051,
+      "step": 3394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.655199683669435e-05,
+      "loss": 2.3283,
+      "step": 3395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.654804270462634e-05,
+      "loss": 2.0953,
+      "step": 3396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6544088572558326e-05,
+      "loss": 2.0153,
+      "step": 3397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6540134440490315e-05,
+      "loss": 2.0456,
+      "step": 3398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6536180308422304e-05,
+      "loss": 1.8836,
+      "step": 3399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.653222617635429e-05,
+      "loss": 2.1727,
+      "step": 3400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.652827204428628e-05,
+      "loss": 1.8801,
+      "step": 3401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.652431791221827e-05,
+      "loss": 1.9408,
+      "step": 3402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.652036378015026e-05,
+      "loss": 2.1082,
+      "step": 3403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.651640964808225e-05,
+      "loss": 2.0863,
+      "step": 3404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.651245551601424e-05,
+      "loss": 2.0592,
+      "step": 3405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6508501383946226e-05,
+      "loss": 1.8747,
+      "step": 3406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6504547251878214e-05,
+      "loss": 1.8193,
+      "step": 3407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.65005931198102e-05,
+      "loss": 1.9987,
+      "step": 3408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.649663898774219e-05,
+      "loss": 2.3335,
+      "step": 3409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.649268485567418e-05,
+      "loss": 1.9216,
+      "step": 3410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.648873072360617e-05,
+      "loss": 2.0924,
+      "step": 3411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.648477659153816e-05,
+      "loss": 2.1378,
+      "step": 3412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.648082245947015e-05,
+      "loss": 2.0609,
+      "step": 3413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6476868327402136e-05,
+      "loss": 2.0827,
+      "step": 3414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6472914195334125e-05,
+      "loss": 2.0813,
+      "step": 3415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6468960063266114e-05,
+      "loss": 2.2453,
+      "step": 3416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.64650059311981e-05,
+      "loss": 2.1615,
+      "step": 3417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.646105179913009e-05,
+      "loss": 2.0053,
+      "step": 3418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.645709766706208e-05,
+      "loss": 2.0702,
+      "step": 3419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.645314353499407e-05,
+      "loss": 2.1513,
+      "step": 3420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.644918940292606e-05,
+      "loss": 2.1804,
+      "step": 3421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.644523527085805e-05,
+      "loss": 2.0966,
+      "step": 3422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.644128113879004e-05,
+      "loss": 2.0618,
+      "step": 3423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6437327006722024e-05,
+      "loss": 2.0844,
+      "step": 3424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.643337287465401e-05,
+      "loss": 2.0679,
+      "step": 3425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6429418742586e-05,
+      "loss": 2.0617,
+      "step": 3426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6425464610518e-05,
+      "loss": 1.9932,
+      "step": 3427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.642151047844998e-05,
+      "loss": 2.3029,
+      "step": 3428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.641755634638197e-05,
+      "loss": 1.9262,
+      "step": 3429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.641360221431396e-05,
+      "loss": 2.0125,
+      "step": 3430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.640964808224595e-05,
+      "loss": 2.1457,
+      "step": 3431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6405693950177935e-05,
+      "loss": 2.4148,
+      "step": 3432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6401739818109924e-05,
+      "loss": 2.0275,
+      "step": 3433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.639778568604191e-05,
+      "loss": 1.8032,
+      "step": 3434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.639383155397391e-05,
+      "loss": 2.1496,
+      "step": 3435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.638987742190589e-05,
+      "loss": 1.9169,
+      "step": 3436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.638592328983788e-05,
+      "loss": 2.0842,
+      "step": 3437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6381969157769875e-05,
+      "loss": 2.1682,
+      "step": 3438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6378015025701864e-05,
+      "loss": 1.8843,
+      "step": 3439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6374060893633846e-05,
+      "loss": 2.0428,
+      "step": 3440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6370106761565835e-05,
+      "loss": 2.0179,
+      "step": 3441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.636615262949783e-05,
+      "loss": 1.8111,
+      "step": 3442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.636219849742982e-05,
+      "loss": 1.9029,
+      "step": 3443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.63582443653618e-05,
+      "loss": 1.8028,
+      "step": 3444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.635429023329379e-05,
+      "loss": 2.1927,
+      "step": 3445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6350336101225785e-05,
+      "loss": 1.9123,
+      "step": 3446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6346381969157774e-05,
+      "loss": 2.0503,
+      "step": 3447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.634242783708976e-05,
+      "loss": 2.0664,
+      "step": 3448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6338473705021745e-05,
+      "loss": 2.09,
+      "step": 3449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.633451957295374e-05,
+      "loss": 1.9438,
+      "step": 3450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.633056544088573e-05,
+      "loss": 2.306,
+      "step": 3451
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.632661130881772e-05,
+      "loss": 1.9812,
+      "step": 3452
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.63226571767497e-05,
+      "loss": 2.067,
+      "step": 3453
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6318703044681696e-05,
+      "loss": 2.0687,
+      "step": 3454
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6314748912613685e-05,
+      "loss": 1.8141,
+      "step": 3455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6310794780545674e-05,
+      "loss": 1.7841,
+      "step": 3456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.630684064847766e-05,
+      "loss": 1.7685,
+      "step": 3457
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.630288651640965e-05,
+      "loss": 2.33,
+      "step": 3458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.629893238434164e-05,
+      "loss": 2.0796,
+      "step": 3459
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.629497825227363e-05,
+      "loss": 2.0977,
+      "step": 3460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.629102412020562e-05,
+      "loss": 2.0477,
+      "step": 3461
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.628706998813761e-05,
+      "loss": 2.0285,
+      "step": 3462
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6283115856069595e-05,
+      "loss": 2.0585,
+      "step": 3463
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6279161724001584e-05,
+      "loss": 2.1372,
+      "step": 3464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.627520759193357e-05,
+      "loss": 2.323,
+      "step": 3465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.627125345986556e-05,
+      "loss": 2.0061,
+      "step": 3466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.626729932779755e-05,
+      "loss": 2.1667,
+      "step": 3467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.626334519572954e-05,
+      "loss": 1.842,
+      "step": 3468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.625939106366153e-05,
+      "loss": 2.1277,
+      "step": 3469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.625543693159352e-05,
+      "loss": 2.2567,
+      "step": 3470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6251482799525506e-05,
+      "loss": 2.1601,
+      "step": 3471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6247528667457495e-05,
+      "loss": 1.8769,
+      "step": 3472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.6243574535389484e-05,
+      "loss": 2.5754,
+      "step": 3473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.623962040332147e-05,
+      "loss": 2.001,
+      "step": 3474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.623566627125346e-05,
+      "loss": 2.1056,
+      "step": 3475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.623171213918545e-05,
+      "loss": 2.2265,
+      "step": 3476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.622775800711744e-05,
+      "loss": 2.3316,
+      "step": 3477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.622380387504943e-05,
+      "loss": 1.9103,
+      "step": 3478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.621984974298142e-05,
+      "loss": 2.2305,
+      "step": 3479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6215895610913406e-05,
+      "loss": 2.0061,
+      "step": 3480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6211941478845394e-05,
+      "loss": 1.8894,
+      "step": 3481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.620798734677738e-05,
+      "loss": 1.8512,
+      "step": 3482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.620403321470937e-05,
+      "loss": 1.9988,
+      "step": 3483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.620007908264136e-05,
+      "loss": 2.2007,
+      "step": 3484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.619612495057335e-05,
+      "loss": 1.9339,
+      "step": 3485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.619217081850534e-05,
+      "loss": 2.0798,
+      "step": 3486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.618821668643733e-05,
+      "loss": 1.9745,
+      "step": 3487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6184262554369316e-05,
+      "loss": 2.0105,
+      "step": 3488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6180308422301305e-05,
+      "loss": 2.1239,
+      "step": 3489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6176354290233294e-05,
+      "loss": 1.9041,
+      "step": 3490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.617240015816529e-05,
+      "loss": 2.063,
+      "step": 3491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.616844602609727e-05,
+      "loss": 2.0143,
+      "step": 3492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.616449189402926e-05,
+      "loss": 1.8931,
+      "step": 3493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.616053776196125e-05,
+      "loss": 1.8541,
+      "step": 3494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6156583629893245e-05,
+      "loss": 2.0609,
+      "step": 3495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.615262949782523e-05,
+      "loss": 1.8598,
+      "step": 3496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6148675365757216e-05,
+      "loss": 1.9188,
+      "step": 3497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6144721233689204e-05,
+      "loss": 2.1605,
+      "step": 3498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.61407671016212e-05,
+      "loss": 1.9465,
+      "step": 3499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.613681296955318e-05,
+      "loss": 2.1306,
+      "step": 3500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.613285883748517e-05,
+      "loss": 2.0256,
+      "step": 3501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.612890470541716e-05,
+      "loss": 2.1224,
+      "step": 3502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6124950573349155e-05,
+      "loss": 1.9785,
+      "step": 3503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6120996441281144e-05,
+      "loss": 2.0984,
+      "step": 3504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6117042309213126e-05,
+      "loss": 1.8695,
+      "step": 3505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6113088177145115e-05,
+      "loss": 2.3005,
+      "step": 3506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.610913404507711e-05,
+      "loss": 2.1798,
+      "step": 3507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.61051799130091e-05,
+      "loss": 1.9828,
+      "step": 3508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.610122578094108e-05,
+      "loss": 2.0478,
+      "step": 3509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.609727164887308e-05,
+      "loss": 2.1636,
+      "step": 3510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6093317516805066e-05,
+      "loss": 2.2521,
+      "step": 3511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6089363384737055e-05,
+      "loss": 2.1704,
+      "step": 3512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.608540925266904e-05,
+      "loss": 2.3007,
+      "step": 3513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.608145512060103e-05,
+      "loss": 2.0854,
+      "step": 3514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.607750098853302e-05,
+      "loss": 1.9551,
+      "step": 3515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.607354685646501e-05,
+      "loss": 2.0134,
+      "step": 3516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.606959272439699e-05,
+      "loss": 2.0545,
+      "step": 3517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.606563859232899e-05,
+      "loss": 1.9409,
+      "step": 3518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6061684460260977e-05,
+      "loss": 1.9449,
+      "step": 3519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6057730328192965e-05,
+      "loss": 2.1773,
+      "step": 3520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.605377619612495e-05,
+      "loss": 1.8488,
+      "step": 3521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.604982206405694e-05,
+      "loss": 2.2065,
+      "step": 3522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.604586793198893e-05,
+      "loss": 1.6995,
+      "step": 3523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.604191379992092e-05,
+      "loss": 2.0395,
+      "step": 3524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.603795966785291e-05,
+      "loss": 2.0442,
+      "step": 3525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.60340055357849e-05,
+      "loss": 1.9184,
+      "step": 3526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.603005140371689e-05,
+      "loss": 2.0636,
+      "step": 3527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6026097271648876e-05,
+      "loss": 2.1267,
+      "step": 3528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6022143139580865e-05,
+      "loss": 1.9617,
+      "step": 3529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.6018189007512854e-05,
+      "loss": 2.0228,
+      "step": 3530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.601423487544484e-05,
+      "loss": 2.2891,
+      "step": 3531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.601028074337683e-05,
+      "loss": 2.3704,
+      "step": 3532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.600632661130882e-05,
+      "loss": 2.0334,
+      "step": 3533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.600237247924081e-05,
+      "loss": 2.0588,
+      "step": 3534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.59984183471728e-05,
+      "loss": 2.0725,
+      "step": 3535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.599446421510479e-05,
+      "loss": 2.3503,
+      "step": 3536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5990510083036775e-05,
+      "loss": 1.9721,
+      "step": 3537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5986555950968764e-05,
+      "loss": 2.1629,
+      "step": 3538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.598260181890075e-05,
+      "loss": 2.2559,
+      "step": 3539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.597864768683274e-05,
+      "loss": 1.9068,
+      "step": 3540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.597469355476473e-05,
+      "loss": 2.1057,
+      "step": 3541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.597073942269672e-05,
+      "loss": 1.8516,
+      "step": 3542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.596678529062871e-05,
+      "loss": 2.269,
+      "step": 3543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.59628311585607e-05,
+      "loss": 1.8367,
+      "step": 3544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5958877026492686e-05,
+      "loss": 1.7475,
+      "step": 3545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5954922894424675e-05,
+      "loss": 1.9067,
+      "step": 3546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5950968762356664e-05,
+      "loss": 2.246,
+      "step": 3547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.594701463028865e-05,
+      "loss": 1.968,
+      "step": 3548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.594306049822064e-05,
+      "loss": 2.1508,
+      "step": 3549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.593910636615263e-05,
+      "loss": 2.0752,
+      "step": 3550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.593515223408462e-05,
+      "loss": 1.9749,
+      "step": 3551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.593119810201661e-05,
+      "loss": 2.0027,
+      "step": 3552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.59272439699486e-05,
+      "loss": 1.8954,
+      "step": 3553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5923289837880586e-05,
+      "loss": 2.3781,
+      "step": 3554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5919335705812574e-05,
+      "loss": 2.0105,
+      "step": 3555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.591538157374456e-05,
+      "loss": 1.9084,
+      "step": 3556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.591142744167655e-05,
+      "loss": 1.9613,
+      "step": 3557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.590747330960854e-05,
+      "loss": 2.056,
+      "step": 3558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.590351917754053e-05,
+      "loss": 2.0926,
+      "step": 3559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5899565045472525e-05,
+      "loss": 2.0896,
+      "step": 3560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.589561091340451e-05,
+      "loss": 1.9808,
+      "step": 3561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5891656781336496e-05,
+      "loss": 2.1166,
+      "step": 3562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.588770264926849e-05,
+      "loss": 2.1756,
+      "step": 3563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.588374851720048e-05,
+      "loss": 2.0522,
+      "step": 3564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.587979438513246e-05,
+      "loss": 2.0357,
+      "step": 3565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.587584025306445e-05,
+      "loss": 1.7929,
+      "step": 3566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.587188612099645e-05,
+      "loss": 2.0974,
+      "step": 3567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5867931988928436e-05,
+      "loss": 2.1246,
+      "step": 3568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.586397785686042e-05,
+      "loss": 1.942,
+      "step": 3569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.586002372479241e-05,
+      "loss": 2.0161,
+      "step": 3570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.58560695927244e-05,
+      "loss": 2.2266,
+      "step": 3571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.585211546065639e-05,
+      "loss": 1.9568,
+      "step": 3572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.584816132858837e-05,
+      "loss": 2.2326,
+      "step": 3573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.584420719652036e-05,
+      "loss": 2.002,
+      "step": 3574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.584025306445236e-05,
+      "loss": 2.1465,
+      "step": 3575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5836298932384346e-05,
+      "loss": 1.9629,
+      "step": 3576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.583234480031633e-05,
+      "loss": 1.8324,
+      "step": 3577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5828390668248324e-05,
+      "loss": 2.056,
+      "step": 3578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.582443653618031e-05,
+      "loss": 1.9279,
+      "step": 3579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.58204824041123e-05,
+      "loss": 2.1076,
+      "step": 3580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5816528272044284e-05,
+      "loss": 1.9454,
+      "step": 3581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.581257413997628e-05,
+      "loss": 2.2274,
+      "step": 3582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.580862000790827e-05,
+      "loss": 2.0108,
+      "step": 3583
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.580466587584026e-05,
+      "loss": 1.847,
+      "step": 3584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.580071174377224e-05,
+      "loss": 2.1506,
+      "step": 3585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5796757611704235e-05,
+      "loss": 2.1015,
+      "step": 3586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5792803479636224e-05,
+      "loss": 2.1484,
+      "step": 3587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.578884934756821e-05,
+      "loss": 2.1919,
+      "step": 3588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5784895215500194e-05,
+      "loss": 1.9058,
+      "step": 3589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.578094108343219e-05,
+      "loss": 2.0498,
+      "step": 3590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.577698695136418e-05,
+      "loss": 1.9468,
+      "step": 3591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.577303281929617e-05,
+      "loss": 2.0193,
+      "step": 3592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.576907868722815e-05,
+      "loss": 2.0725,
+      "step": 3593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5765124555160145e-05,
+      "loss": 2.1762,
+      "step": 3594
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5761170423092134e-05,
+      "loss": 2.0255,
+      "step": 3595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.575721629102412e-05,
+      "loss": 1.9559,
+      "step": 3596
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.575326215895611e-05,
+      "loss": 2.062,
+      "step": 3597
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.57493080268881e-05,
+      "loss": 1.8652,
+      "step": 3598
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.574535389482009e-05,
+      "loss": 2.1742,
+      "step": 3599
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.574139976275208e-05,
+      "loss": 2.0427,
+      "step": 3600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.573744563068407e-05,
+      "loss": 1.9261,
+      "step": 3601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5733491498616056e-05,
+      "loss": 1.8223,
+      "step": 3602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5729537366548045e-05,
+      "loss": 1.8747,
+      "step": 3603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5725583234480034e-05,
+      "loss": 2.0066,
+      "step": 3604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.572162910241202e-05,
+      "loss": 1.9896,
+      "step": 3605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.571767497034401e-05,
+      "loss": 1.9158,
+      "step": 3606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5713720838276e-05,
+      "loss": 1.9687,
+      "step": 3607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.570976670620799e-05,
+      "loss": 2.3502,
+      "step": 3608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.570581257413998e-05,
+      "loss": 2.0657,
+      "step": 3609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5701858442071967e-05,
+      "loss": 2.0614,
+      "step": 3610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5697904310003955e-05,
+      "loss": 2.1244,
+      "step": 3611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5693950177935944e-05,
+      "loss": 2.0309,
+      "step": 3612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.568999604586793e-05,
+      "loss": 1.9214,
+      "step": 3613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.568604191379992e-05,
+      "loss": 1.7206,
+      "step": 3614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.568208778173191e-05,
+      "loss": 1.9969,
+      "step": 3615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.56781336496639e-05,
+      "loss": 2.0051,
+      "step": 3616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.567417951759589e-05,
+      "loss": 1.9829,
+      "step": 3617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.567022538552788e-05,
+      "loss": 2.0684,
+      "step": 3618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5666271253459866e-05,
+      "loss": 1.8797,
+      "step": 3619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.566231712139186e-05,
+      "loss": 2.1108,
+      "step": 3620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5658362989323844e-05,
+      "loss": 1.7596,
+      "step": 3621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.565440885725583e-05,
+      "loss": 2.1735,
+      "step": 3622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.565045472518782e-05,
+      "loss": 2.1265,
+      "step": 3623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.564650059311982e-05,
+      "loss": 2.0087,
+      "step": 3624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.56425464610518e-05,
+      "loss": 2.0476,
+      "step": 3625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.563859232898379e-05,
+      "loss": 1.9135,
+      "step": 3626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.563463819691578e-05,
+      "loss": 2.0142,
+      "step": 3627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.563068406484777e-05,
+      "loss": 2.1314,
+      "step": 3628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5626729932779754e-05,
+      "loss": 1.9223,
+      "step": 3629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.562277580071174e-05,
+      "loss": 1.9248,
+      "step": 3630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.561882166864374e-05,
+      "loss": 1.8457,
+      "step": 3631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.561486753657573e-05,
+      "loss": 1.9514,
+      "step": 3632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.561091340450771e-05,
+      "loss": 2.3167,
+      "step": 3633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.56069592724397e-05,
+      "loss": 1.8882,
+      "step": 3634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5603005140371694e-05,
+      "loss": 1.8972,
+      "step": 3635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.559905100830368e-05,
+      "loss": 1.9856,
+      "step": 3636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5595096876235665e-05,
+      "loss": 1.9494,
+      "step": 3637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5591142744167654e-05,
+      "loss": 1.9002,
+      "step": 3638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.558718861209965e-05,
+      "loss": 2.2512,
+      "step": 3639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.558323448003164e-05,
+      "loss": 2.1883,
+      "step": 3640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.557928034796362e-05,
+      "loss": 1.9112,
+      "step": 3641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.557532621589561e-05,
+      "loss": 1.7717,
+      "step": 3642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5571372083827605e-05,
+      "loss": 1.8853,
+      "step": 3643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5567417951759593e-05,
+      "loss": 2.1313,
+      "step": 3644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5563463819691576e-05,
+      "loss": 1.9364,
+      "step": 3645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5559509687623564e-05,
+      "loss": 2.0509,
+      "step": 3646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.555555555555556e-05,
+      "loss": 2.0724,
+      "step": 3647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.555160142348755e-05,
+      "loss": 1.979,
+      "step": 3648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.554764729141953e-05,
+      "loss": 2.0135,
+      "step": 3649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5543693159351526e-05,
+      "loss": 1.9075,
+      "step": 3650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5539739027283515e-05,
+      "loss": 1.9859,
+      "step": 3651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5535784895215504e-05,
+      "loss": 2.2364,
+      "step": 3652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5531830763147486e-05,
+      "loss": 1.9424,
+      "step": 3653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.552787663107948e-05,
+      "loss": 2.3661,
+      "step": 3654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.552392249901147e-05,
+      "loss": 1.9158,
+      "step": 3655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.551996836694346e-05,
+      "loss": 1.8678,
+      "step": 3656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.551601423487544e-05,
+      "loss": 2.2413,
+      "step": 3657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.551206010280744e-05,
+      "loss": 2.0119,
+      "step": 3658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5508105970739426e-05,
+      "loss": 1.9947,
+      "step": 3659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5504151838671415e-05,
+      "loss": 2.1012,
+      "step": 3660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.55001977066034e-05,
+      "loss": 2.0889,
+      "step": 3661
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.549624357453539e-05,
+      "loss": 2.0608,
+      "step": 3662
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.549228944246738e-05,
+      "loss": 2.2308,
+      "step": 3663
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.548833531039937e-05,
+      "loss": 2.2981,
+      "step": 3664
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.548438117833136e-05,
+      "loss": 2.065,
+      "step": 3665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.548042704626335e-05,
+      "loss": 2.1048,
+      "step": 3666
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5476472914195337e-05,
+      "loss": 2.0143,
+      "step": 3667
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5472518782127325e-05,
+      "loss": 1.753,
+      "step": 3668
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5468564650059314e-05,
+      "loss": 2.1273,
+      "step": 3669
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.54646105179913e-05,
+      "loss": 1.9597,
+      "step": 3670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.546065638592329e-05,
+      "loss": 2.0824,
+      "step": 3671
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.545670225385528e-05,
+      "loss": 1.9892,
+      "step": 3672
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.545274812178727e-05,
+      "loss": 2.2254,
+      "step": 3673
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.544879398971926e-05,
+      "loss": 2.0521,
+      "step": 3674
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.544483985765125e-05,
+      "loss": 2.3705,
+      "step": 3675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5440885725583236e-05,
+      "loss": 1.8792,
+      "step": 3676
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5436931593515225e-05,
+      "loss": 2.0218,
+      "step": 3677
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5432977461447214e-05,
+      "loss": 1.7069,
+      "step": 3678
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.54290233293792e-05,
+      "loss": 1.8803,
+      "step": 3679
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.542506919731119e-05,
+      "loss": 2.0697,
+      "step": 3680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.542111506524318e-05,
+      "loss": 2.184,
+      "step": 3681
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.541716093317517e-05,
+      "loss": 2.0074,
+      "step": 3682
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.541320680110716e-05,
+      "loss": 2.2064,
+      "step": 3683
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.540925266903915e-05,
+      "loss": 1.9732,
+      "step": 3684
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5405298536971135e-05,
+      "loss": 2.0026,
+      "step": 3685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5401344404903124e-05,
+      "loss": 2.2659,
+      "step": 3686
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.539739027283511e-05,
+      "loss": 2.0902,
+      "step": 3687
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.539343614076711e-05,
+      "loss": 2.1647,
+      "step": 3688
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.538948200869909e-05,
+      "loss": 2.1178,
+      "step": 3689
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.538552787663108e-05,
+      "loss": 2.1335,
+      "step": 3690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.538157374456307e-05,
+      "loss": 2.2439,
+      "step": 3691
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5377619612495064e-05,
+      "loss": 2.2015,
+      "step": 3692
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5373665480427046e-05,
+      "loss": 1.9663,
+      "step": 3693
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5369711348359035e-05,
+      "loss": 2.2784,
+      "step": 3694
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5365757216291024e-05,
+      "loss": 1.8794,
+      "step": 3695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.536180308422302e-05,
+      "loss": 2.3205,
+      "step": 3696
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5357848952155e-05,
+      "loss": 2.0409,
+      "step": 3697
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.535389482008699e-05,
+      "loss": 1.9285,
+      "step": 3698
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5349940688018986e-05,
+      "loss": 2.1751,
+      "step": 3699
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5345986555950975e-05,
+      "loss": 2.109,
+      "step": 3700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.534203242388296e-05,
+      "loss": 2.0691,
+      "step": 3701
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5338078291814945e-05,
+      "loss": 2.2237,
+      "step": 3702
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.533412415974694e-05,
+      "loss": 1.9144,
+      "step": 3703
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.533017002767893e-05,
+      "loss": 2.0106,
+      "step": 3704
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.532621589561091e-05,
+      "loss": 2.1473,
+      "step": 3705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.53222617635429e-05,
+      "loss": 2.1374,
+      "step": 3706
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5318307631474896e-05,
+      "loss": 1.8199,
+      "step": 3707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5314353499406885e-05,
+      "loss": 2.0749,
+      "step": 3708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.531039936733887e-05,
+      "loss": 1.9782,
+      "step": 3709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5306445235270856e-05,
+      "loss": 2.0504,
+      "step": 3710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.530249110320285e-05,
+      "loss": 2.0664,
+      "step": 3711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.529853697113484e-05,
+      "loss": 2.1292,
+      "step": 3712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.529458283906682e-05,
+      "loss": 2.2811,
+      "step": 3713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.529062870699881e-05,
+      "loss": 2.3184,
+      "step": 3714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.528667457493081e-05,
+      "loss": 2.1,
+      "step": 3715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5282720442862796e-05,
+      "loss": 2.0697,
+      "step": 3716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.527876631079478e-05,
+      "loss": 2.1761,
+      "step": 3717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5274812178726773e-05,
+      "loss": 1.8818,
+      "step": 3718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.527085804665876e-05,
+      "loss": 2.1013,
+      "step": 3719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.526690391459075e-05,
+      "loss": 2.3087,
+      "step": 3720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.526294978252273e-05,
+      "loss": 2.0998,
+      "step": 3721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.525899565045473e-05,
+      "loss": 1.8741,
+      "step": 3722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.525504151838672e-05,
+      "loss": 1.9979,
+      "step": 3723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5251087386318706e-05,
+      "loss": 2.2358,
+      "step": 3724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.524713325425069e-05,
+      "loss": 1.8154,
+      "step": 3725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5243179122182684e-05,
+      "loss": 2.2539,
+      "step": 3726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.523922499011467e-05,
+      "loss": 1.8501,
+      "step": 3727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.523527085804666e-05,
+      "loss": 1.8777,
+      "step": 3728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5231316725978644e-05,
+      "loss": 2.2478,
+      "step": 3729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.522736259391064e-05,
+      "loss": 2.1752,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.522340846184263e-05,
+      "loss": 1.8676,
+      "step": 3731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.521945432977462e-05,
+      "loss": 1.9983,
+      "step": 3732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.52155001977066e-05,
+      "loss": 1.7659,
+      "step": 3733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5211546065638595e-05,
+      "loss": 2.0657,
+      "step": 3734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5207591933570584e-05,
+      "loss": 1.9565,
+      "step": 3735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.520363780150257e-05,
+      "loss": 1.8586,
+      "step": 3736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.519968366943456e-05,
+      "loss": 2.071,
+      "step": 3737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.519572953736655e-05,
+      "loss": 1.9053,
+      "step": 3738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.519177540529854e-05,
+      "loss": 2.2328,
+      "step": 3739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.518782127323053e-05,
+      "loss": 2.1807,
+      "step": 3740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5183867141162516e-05,
+      "loss": 1.9841,
+      "step": 3741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5179913009094505e-05,
+      "loss": 1.6429,
+      "step": 3742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5175958877026494e-05,
+      "loss": 2.2619,
+      "step": 3743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.517200474495848e-05,
+      "loss": 2.114,
+      "step": 3744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.516805061289047e-05,
+      "loss": 2.1712,
+      "step": 3745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.516409648082246e-05,
+      "loss": 2.097,
+      "step": 3746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.516014234875445e-05,
+      "loss": 2.1763,
+      "step": 3747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.515618821668644e-05,
+      "loss": 2.3172,
+      "step": 3748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.515223408461843e-05,
+      "loss": 1.9993,
+      "step": 3749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5148279952550416e-05,
+      "loss": 2.1431,
+      "step": 3750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5144325820482405e-05,
+      "loss": 2.2038,
+      "step": 3751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.51403716884144e-05,
+      "loss": 2.1305,
+      "step": 3752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.513641755634638e-05,
+      "loss": 1.6626,
+      "step": 3753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.513246342427837e-05,
+      "loss": 2.2422,
+      "step": 3754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.512850929221036e-05,
+      "loss": 1.9318,
+      "step": 3755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5124555160142356e-05,
+      "loss": 1.9062,
+      "step": 3756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.512060102807434e-05,
+      "loss": 1.7803,
+      "step": 3757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5116646896006327e-05,
+      "loss": 2.0285,
+      "step": 3758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5112692763938315e-05,
+      "loss": 2.0335,
+      "step": 3759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.510873863187031e-05,
+      "loss": 2.0447,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.510478449980229e-05,
+      "loss": 2.087,
+      "step": 3761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.510083036773428e-05,
+      "loss": 1.8267,
+      "step": 3762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.509687623566627e-05,
+      "loss": 2.0429,
+      "step": 3763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5092922103598266e-05,
+      "loss": 1.9596,
+      "step": 3764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.508896797153025e-05,
+      "loss": 2.0857,
+      "step": 3765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.508501383946224e-05,
+      "loss": 2.2163,
+      "step": 3766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5081059707394226e-05,
+      "loss": 1.8712,
+      "step": 3767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.507710557532622e-05,
+      "loss": 2.1793,
+      "step": 3768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5073151443258204e-05,
+      "loss": 2.1142,
+      "step": 3769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.506919731119019e-05,
+      "loss": 1.705,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.506524317912219e-05,
+      "loss": 1.8742,
+      "step": 3771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.506128904705418e-05,
+      "loss": 2.3477,
+      "step": 3772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.505733491498616e-05,
+      "loss": 1.9288,
+      "step": 3773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.505338078291815e-05,
+      "loss": 1.9022,
+      "step": 3774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.504942665085014e-05,
+      "loss": 2.2038,
+      "step": 3775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.504547251878213e-05,
+      "loss": 2.016,
+      "step": 3776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5041518386714114e-05,
+      "loss": 2.0644,
+      "step": 3777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.50375642546461e-05,
+      "loss": 2.3327,
+      "step": 3778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.50336101225781e-05,
+      "loss": 1.9141,
+      "step": 3779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.502965599051009e-05,
+      "loss": 1.8158,
+      "step": 3780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.502570185844207e-05,
+      "loss": 2.0074,
+      "step": 3781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.502174772637406e-05,
+      "loss": 2.3438,
+      "step": 3782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5017793594306054e-05,
+      "loss": 2.0342,
+      "step": 3783
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.501383946223804e-05,
+      "loss": 1.9322,
+      "step": 3784
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.5009885330170025e-05,
+      "loss": 1.9491,
+      "step": 3785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.500593119810202e-05,
+      "loss": 2.0596,
+      "step": 3786
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.500197706603401e-05,
+      "loss": 1.9963,
+      "step": 3787
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4998022933966e-05,
+      "loss": 2.116,
+      "step": 3788
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.499406880189798e-05,
+      "loss": 2.048,
+      "step": 3789
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4990114669829976e-05,
+      "loss": 1.8947,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4986160537761965e-05,
+      "loss": 2.0564,
+      "step": 3791
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4982206405693953e-05,
+      "loss": 1.996,
+      "step": 3792
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4978252273625935e-05,
+      "loss": 2.036,
+      "step": 3793
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.497429814155793e-05,
+      "loss": 1.921,
+      "step": 3794
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.497034400948992e-05,
+      "loss": 2.3067,
+      "step": 3795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.496638987742191e-05,
+      "loss": 1.942,
+      "step": 3796
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.49624357453539e-05,
+      "loss": 2.1993,
+      "step": 3797
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4958481613285886e-05,
+      "loss": 2.0517,
+      "step": 3798
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4954527481217875e-05,
+      "loss": 2.0261,
+      "step": 3799
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4950573349149864e-05,
+      "loss": 2.0646,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.494661921708185e-05,
+      "loss": 2.0206,
+      "step": 3801
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.494266508501384e-05,
+      "loss": 2.0616,
+      "step": 3802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.493871095294583e-05,
+      "loss": 1.7991,
+      "step": 3803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.493475682087782e-05,
+      "loss": 2.2173,
+      "step": 3804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.493080268880981e-05,
+      "loss": 2.038,
+      "step": 3805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.49268485567418e-05,
+      "loss": 1.898,
+      "step": 3806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4922894424673786e-05,
+      "loss": 2.0866,
+      "step": 3807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4918940292605775e-05,
+      "loss": 2.086,
+      "step": 3808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4914986160537763e-05,
+      "loss": 2.0325,
+      "step": 3809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.491103202846975e-05,
+      "loss": 2.0865,
+      "step": 3810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.490707789640174e-05,
+      "loss": 2.0027,
+      "step": 3811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.490312376433373e-05,
+      "loss": 1.7496,
+      "step": 3812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.489916963226572e-05,
+      "loss": 1.8633,
+      "step": 3813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.489521550019771e-05,
+      "loss": 1.8123,
+      "step": 3814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4891261368129696e-05,
+      "loss": 2.1064,
+      "step": 3815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4887307236061685e-05,
+      "loss": 1.9541,
+      "step": 3816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4883353103993674e-05,
+      "loss": 1.9189,
+      "step": 3817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.487939897192566e-05,
+      "loss": 1.9952,
+      "step": 3818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.487544483985765e-05,
+      "loss": 2.1947,
+      "step": 3819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.487149070778964e-05,
+      "loss": 2.1123,
+      "step": 3820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.486753657572163e-05,
+      "loss": 1.8847,
+      "step": 3821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.486358244365362e-05,
+      "loss": 1.951,
+      "step": 3822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.485962831158561e-05,
+      "loss": 2.0396,
+      "step": 3823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.48556741795176e-05,
+      "loss": 2.1545,
+      "step": 3824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4851720047449585e-05,
+      "loss": 2.095,
+      "step": 3825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4847765915381574e-05,
+      "loss": 2.1415,
+      "step": 3826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.484381178331356e-05,
+      "loss": 1.8688,
+      "step": 3827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.483985765124556e-05,
+      "loss": 2.1068,
+      "step": 3828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.483590351917754e-05,
+      "loss": 1.8533,
+      "step": 3829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.483194938710953e-05,
+      "loss": 2.3491,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.482799525504152e-05,
+      "loss": 2.0807,
+      "step": 3831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.482404112297351e-05,
+      "loss": 2.0944,
+      "step": 3832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4820086990905495e-05,
+      "loss": 1.9838,
+      "step": 3833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4816132858837484e-05,
+      "loss": 2.1166,
+      "step": 3834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.481217872676947e-05,
+      "loss": 2.0344,
+      "step": 3835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.480822459470147e-05,
+      "loss": 2.2045,
+      "step": 3836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.480427046263345e-05,
+      "loss": 2.0672,
+      "step": 3837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.480031633056544e-05,
+      "loss": 1.9602,
+      "step": 3838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4796362198497435e-05,
+      "loss": 2.2149,
+      "step": 3839
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4792408066429424e-05,
+      "loss": 1.7418,
+      "step": 3840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4788453934361406e-05,
+      "loss": 2.0032,
+      "step": 3841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4784499802293395e-05,
+      "loss": 2.0119,
+      "step": 3842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.478054567022539e-05,
+      "loss": 2.1395,
+      "step": 3843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.477659153815738e-05,
+      "loss": 2.1027,
+      "step": 3844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.477263740608936e-05,
+      "loss": 2.0781,
+      "step": 3845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.476868327402135e-05,
+      "loss": 2.0179,
+      "step": 3846
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4764729141953346e-05,
+      "loss": 2.2873,
+      "step": 3847
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4760775009885335e-05,
+      "loss": 2.0784,
+      "step": 3848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4756820877817317e-05,
+      "loss": 1.8262,
+      "step": 3849
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4752866745749305e-05,
+      "loss": 1.996,
+      "step": 3850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.47489126136813e-05,
+      "loss": 2.1457,
+      "step": 3851
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.474495848161329e-05,
+      "loss": 1.91,
+      "step": 3852
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.474100434954528e-05,
+      "loss": 2.1482,
+      "step": 3853
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.473705021747726e-05,
+      "loss": 2.1762,
+      "step": 3854
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4733096085409256e-05,
+      "loss": 2.3477,
+      "step": 3855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.4729141953341245e-05,
+      "loss": 2.0,
+      "step": 3856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4725187821273234e-05,
+      "loss": 2.1196,
+      "step": 3857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.472123368920522e-05,
+      "loss": 1.9926,
+      "step": 3858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.471727955713721e-05,
+      "loss": 2.0336,
+      "step": 3859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.47133254250692e-05,
+      "loss": 2.1045,
+      "step": 3860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.470937129300119e-05,
+      "loss": 1.9762,
+      "step": 3861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.470541716093318e-05,
+      "loss": 2.1392,
+      "step": 3862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.470146302886517e-05,
+      "loss": 1.9346,
+      "step": 3863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4697508896797156e-05,
+      "loss": 2.1789,
+      "step": 3864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4693554764729145e-05,
+      "loss": 2.0747,
+      "step": 3865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4689600632661133e-05,
+      "loss": 1.84,
+      "step": 3866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.468564650059312e-05,
+      "loss": 2.1329,
+      "step": 3867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.468169236852511e-05,
+      "loss": 1.9566,
+      "step": 3868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.46777382364571e-05,
+      "loss": 1.9884,
+      "step": 3869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.467378410438909e-05,
+      "loss": 2.0503,
+      "step": 3870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.466982997232108e-05,
+      "loss": 1.9554,
+      "step": 3871
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4665875840253066e-05,
+      "loss": 2.0828,
+      "step": 3872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4661921708185055e-05,
+      "loss": 2.1837,
+      "step": 3873
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4657967576117044e-05,
+      "loss": 2.3821,
+      "step": 3874
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.465401344404903e-05,
+      "loss": 2.0172,
+      "step": 3875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.465005931198102e-05,
+      "loss": 2.2299,
+      "step": 3876
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.464610517991301e-05,
+      "loss": 1.7653,
+      "step": 3877
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4642151047845e-05,
+      "loss": 1.9693,
+      "step": 3878
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.463819691577699e-05,
+      "loss": 2.015,
+      "step": 3879
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.463424278370898e-05,
+      "loss": 1.9099,
+      "step": 3880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4630288651640966e-05,
+      "loss": 2.3009,
+      "step": 3881
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4626334519572955e-05,
+      "loss": 2.1475,
+      "step": 3882
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4622380387504943e-05,
+      "loss": 2.0257,
+      "step": 3883
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.461842625543693e-05,
+      "loss": 1.9504,
+      "step": 3884
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.461447212336892e-05,
+      "loss": 2.2464,
+      "step": 3885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.461051799130091e-05,
+      "loss": 2.3137,
+      "step": 3886
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.46065638592329e-05,
+      "loss": 1.9377,
+      "step": 3887
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.460260972716489e-05,
+      "loss": 1.7864,
+      "step": 3888
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4598655595096876e-05,
+      "loss": 1.9042,
+      "step": 3889
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4594701463028865e-05,
+      "loss": 2.0882,
+      "step": 3890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4590747330960854e-05,
+      "loss": 1.919,
+      "step": 3891
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.458679319889285e-05,
+      "loss": 2.0682,
+      "step": 3892
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.458283906682483e-05,
+      "loss": 2.1269,
+      "step": 3893
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.457888493475682e-05,
+      "loss": 2.362,
+      "step": 3894
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.457493080268881e-05,
+      "loss": 1.7047,
+      "step": 3895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4570976670620805e-05,
+      "loss": 1.8507,
+      "step": 3896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.456702253855279e-05,
+      "loss": 1.9421,
+      "step": 3897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4563068406484776e-05,
+      "loss": 2.1648,
+      "step": 3898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4559114274416765e-05,
+      "loss": 2.078,
+      "step": 3899
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.455516014234876e-05,
+      "loss": 2.042,
+      "step": 3900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.455120601028074e-05,
+      "loss": 2.0389,
+      "step": 3901
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.454725187821273e-05,
+      "loss": 1.9916,
+      "step": 3902
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.454329774614472e-05,
+      "loss": 1.9637,
+      "step": 3903
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4539343614076716e-05,
+      "loss": 2.0159,
+      "step": 3904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.45353894820087e-05,
+      "loss": 2.1464,
+      "step": 3905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4531435349940686e-05,
+      "loss": 1.9344,
+      "step": 3906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4527481217872675e-05,
+      "loss": 1.8836,
+      "step": 3907
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.452352708580467e-05,
+      "loss": 1.8197,
+      "step": 3908
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.451957295373665e-05,
+      "loss": 2.0404,
+      "step": 3909
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.451561882166864e-05,
+      "loss": 1.9874,
+      "step": 3910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.451166468960064e-05,
+      "loss": 2.03,
+      "step": 3911
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4507710557532626e-05,
+      "loss": 2.1381,
+      "step": 3912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4503756425464615e-05,
+      "loss": 1.8549,
+      "step": 3913
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.44998022933966e-05,
+      "loss": 2.1143,
+      "step": 3914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.449584816132859e-05,
+      "loss": 2.1512,
+      "step": 3915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.449189402926058e-05,
+      "loss": 1.7974,
+      "step": 3916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.448793989719257e-05,
+      "loss": 1.904,
+      "step": 3917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.448398576512455e-05,
+      "loss": 2.165,
+      "step": 3918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.448003163305655e-05,
+      "loss": 1.9615,
+      "step": 3919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.447607750098854e-05,
+      "loss": 2.3935,
+      "step": 3920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4472123368920526e-05,
+      "loss": 2.0251,
+      "step": 3921
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.446816923685251e-05,
+      "loss": 1.9063,
+      "step": 3922
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.44642151047845e-05,
+      "loss": 1.8691,
+      "step": 3923
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.446026097271649e-05,
+      "loss": 1.9986,
+      "step": 3924
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.445630684064848e-05,
+      "loss": 1.9866,
+      "step": 3925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.445235270858047e-05,
+      "loss": 1.8935,
+      "step": 3926
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.444839857651246e-05,
+      "loss": 1.9407,
+      "step": 3927
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.444444444444445e-05,
+      "loss": 2.1656,
+      "step": 3928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4440490312376436e-05,
+      "loss": 2.1453,
+      "step": 3929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4436536180308425e-05,
+      "loss": 2.0141,
+      "step": 3930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4432582048240414e-05,
+      "loss": 1.9706,
+      "step": 3931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.44286279161724e-05,
+      "loss": 2.2088,
+      "step": 3932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.442467378410439e-05,
+      "loss": 1.8544,
+      "step": 3933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.442071965203638e-05,
+      "loss": 2.1296,
+      "step": 3934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.441676551996837e-05,
+      "loss": 2.2633,
+      "step": 3935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.441281138790036e-05,
+      "loss": 2.2138,
+      "step": 3936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.440885725583235e-05,
+      "loss": 2.2295,
+      "step": 3937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4404903123764336e-05,
+      "loss": 2.1391,
+      "step": 3938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4400948991696325e-05,
+      "loss": 1.6918,
+      "step": 3939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.439699485962831e-05,
+      "loss": 2.1635,
+      "step": 3940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.43930407275603e-05,
+      "loss": 1.8448,
+      "step": 3941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.438908659549229e-05,
+      "loss": 1.8806,
+      "step": 3942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.438513246342428e-05,
+      "loss": 1.9654,
+      "step": 3943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.438117833135627e-05,
+      "loss": 2.0827,
+      "step": 3944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.437722419928826e-05,
+      "loss": 2.1119,
+      "step": 3945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4373270067220246e-05,
+      "loss": 2.0254,
+      "step": 3946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4369315935152235e-05,
+      "loss": 2.0176,
+      "step": 3947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4365361803084224e-05,
+      "loss": 2.1274,
+      "step": 3948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.436140767101621e-05,
+      "loss": 1.8697,
+      "step": 3949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.43574535389482e-05,
+      "loss": 1.9995,
+      "step": 3950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.435349940688019e-05,
+      "loss": 1.9497,
+      "step": 3951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.434954527481218e-05,
+      "loss": 1.6673,
+      "step": 3952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.434559114274417e-05,
+      "loss": 1.8404,
+      "step": 3953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.434163701067616e-05,
+      "loss": 2.0354,
+      "step": 3954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4337682878608146e-05,
+      "loss": 2.2186,
+      "step": 3955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4333728746540135e-05,
+      "loss": 2.0519,
+      "step": 3956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4329774614472123e-05,
+      "loss": 1.8253,
+      "step": 3957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.432582048240411e-05,
+      "loss": 2.0639,
+      "step": 3958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.43218663503361e-05,
+      "loss": 1.9274,
+      "step": 3959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.43179122182681e-05,
+      "loss": 2.1727,
+      "step": 3960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.431395808620008e-05,
+      "loss": 2.0577,
+      "step": 3961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.431000395413207e-05,
+      "loss": 2.3084,
+      "step": 3962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4306049822064056e-05,
+      "loss": 2.1335,
+      "step": 3963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.430209568999605e-05,
+      "loss": 1.898,
+      "step": 3964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4298141557928034e-05,
+      "loss": 2.0012,
+      "step": 3965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.429418742586002e-05,
+      "loss": 1.9749,
+      "step": 3966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.429023329379201e-05,
+      "loss": 2.108,
+      "step": 3967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.428627916172401e-05,
+      "loss": 2.0883,
+      "step": 3968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4282325029655996e-05,
+      "loss": 2.0128,
+      "step": 3969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.427837089758798e-05,
+      "loss": 1.8096,
+      "step": 3970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.427441676551997e-05,
+      "loss": 2.0419,
+      "step": 3971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.427046263345196e-05,
+      "loss": 1.9564,
+      "step": 3972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.426650850138395e-05,
+      "loss": 1.9808,
+      "step": 3973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4262554369315933e-05,
+      "loss": 1.914,
+      "step": 3974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.425860023724792e-05,
+      "loss": 1.7962,
+      "step": 3975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.425464610517992e-05,
+      "loss": 2.1286,
+      "step": 3976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.425069197311191e-05,
+      "loss": 2.0337,
+      "step": 3977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.424673784104389e-05,
+      "loss": 2.0244,
+      "step": 3978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4242783708975884e-05,
+      "loss": 2.0225,
+      "step": 3979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.423882957690787e-05,
+      "loss": 1.8741,
+      "step": 3980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.423487544483986e-05,
+      "loss": 2.0801,
+      "step": 3981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4230921312771844e-05,
+      "loss": 1.8523,
+      "step": 3982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.422696718070384e-05,
+      "loss": 2.1732,
+      "step": 3983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.422301304863583e-05,
+      "loss": 2.1646,
+      "step": 3984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.421905891656782e-05,
+      "loss": 1.9943,
+      "step": 3985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.42151047844998e-05,
+      "loss": 2.4087,
+      "step": 3986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4211150652431795e-05,
+      "loss": 2.2876,
+      "step": 3987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4207196520363784e-05,
+      "loss": 2.2841,
+      "step": 3988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.420324238829577e-05,
+      "loss": 1.8316,
+      "step": 3989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4199288256227755e-05,
+      "loss": 1.9497,
+      "step": 3990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.419533412415975e-05,
+      "loss": 1.7279,
+      "step": 3991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.419137999209174e-05,
+      "loss": 2.022,
+      "step": 3992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.418742586002373e-05,
+      "loss": 2.0804,
+      "step": 3993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.418347172795571e-05,
+      "loss": 2.0318,
+      "step": 3994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4179517595887706e-05,
+      "loss": 1.8802,
+      "step": 3995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4175563463819694e-05,
+      "loss": 2.1571,
+      "step": 3996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.417160933175168e-05,
+      "loss": 1.9455,
+      "step": 3997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.416765519968367e-05,
+      "loss": 1.69,
+      "step": 3998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.416370106761566e-05,
+      "loss": 2.0041,
+      "step": 3999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.415974693554765e-05,
+      "loss": 2.0999,
+      "step": 4000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.415579280347964e-05,
+      "loss": 2.1006,
+      "step": 4001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.415183867141163e-05,
+      "loss": 2.1962,
+      "step": 4002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4147884539343616e-05,
+      "loss": 1.9885,
+      "step": 4003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4143930407275605e-05,
+      "loss": 2.1682,
+      "step": 4004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4139976275207594e-05,
+      "loss": 1.8843,
+      "step": 4005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.413602214313958e-05,
+      "loss": 2.0919,
+      "step": 4006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.413206801107157e-05,
+      "loss": 2.2221,
+      "step": 4007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.412811387900356e-05,
+      "loss": 2.027,
+      "step": 4008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.412415974693555e-05,
+      "loss": 1.9979,
+      "step": 4009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.412020561486754e-05,
+      "loss": 2.1272,
+      "step": 4010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.411625148279953e-05,
+      "loss": 2.0978,
+      "step": 4011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4112297350731516e-05,
+      "loss": 1.9311,
+      "step": 4012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4108343218663505e-05,
+      "loss": 1.9965,
+      "step": 4013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.410438908659549e-05,
+      "loss": 2.1819,
+      "step": 4014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.410043495452748e-05,
+      "loss": 2.1067,
+      "step": 4015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.409648082245947e-05,
+      "loss": 2.0524,
+      "step": 4016
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.409252669039146e-05,
+      "loss": 1.9749,
+      "step": 4017
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.408857255832345e-05,
+      "loss": 1.9135,
+      "step": 4018
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.408461842625544e-05,
+      "loss": 1.9893,
+      "step": 4019
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4080664294187426e-05,
+      "loss": 2.0885,
+      "step": 4020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4076710162119415e-05,
+      "loss": 1.9414,
+      "step": 4021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4072756030051404e-05,
+      "loss": 1.9297,
+      "step": 4022
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.406880189798339e-05,
+      "loss": 1.8998,
+      "step": 4023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.406484776591538e-05,
+      "loss": 1.8994,
+      "step": 4024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.406089363384738e-05,
+      "loss": 2.1033,
+      "step": 4025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.405693950177936e-05,
+      "loss": 1.9218,
+      "step": 4026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.405298536971135e-05,
+      "loss": 2.1704,
+      "step": 4027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.404903123764334e-05,
+      "loss": 1.9711,
+      "step": 4028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.404507710557533e-05,
+      "loss": 1.9031,
+      "step": 4029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4041122973507315e-05,
+      "loss": 1.8627,
+      "step": 4030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4037168841439303e-05,
+      "loss": 1.8727,
+      "step": 4031
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.40332147093713e-05,
+      "loss": 2.0121,
+      "step": 4032
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.402926057730329e-05,
+      "loss": 1.9568,
+      "step": 4033
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.402530644523527e-05,
+      "loss": 2.1151,
+      "step": 4034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.402135231316726e-05,
+      "loss": 2.1714,
+      "step": 4035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4017398181099254e-05,
+      "loss": 2.2088,
+      "step": 4036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.401344404903124e-05,
+      "loss": 2.0776,
+      "step": 4037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4009489916963225e-05,
+      "loss": 2.0965,
+      "step": 4038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.4005535784895214e-05,
+      "loss": 1.8488,
+      "step": 4039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.400158165282721e-05,
+      "loss": 2.4447,
+      "step": 4040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.39976275207592e-05,
+      "loss": 2.0501,
+      "step": 4041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.399367338869118e-05,
+      "loss": 1.9617,
+      "step": 4042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.398971925662317e-05,
+      "loss": 2.004,
+      "step": 4043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3985765124555165e-05,
+      "loss": 2.238,
+      "step": 4044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3981810992487154e-05,
+      "loss": 1.7975,
+      "step": 4045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3977856860419136e-05,
+      "loss": 1.7549,
+      "step": 4046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.397390272835113e-05,
+      "loss": 2.1044,
+      "step": 4047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.396994859628312e-05,
+      "loss": 1.9987,
+      "step": 4048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.396599446421511e-05,
+      "loss": 1.8596,
+      "step": 4049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.396204033214709e-05,
+      "loss": 2.1262,
+      "step": 4050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.395808620007909e-05,
+      "loss": 2.1222,
+      "step": 4051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3954132068011076e-05,
+      "loss": 1.9164,
+      "step": 4052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3950177935943064e-05,
+      "loss": 2.2699,
+      "step": 4053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3946223803875046e-05,
+      "loss": 1.743,
+      "step": 4054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.394226967180704e-05,
+      "loss": 2.0324,
+      "step": 4055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.393831553973903e-05,
+      "loss": 2.0375,
+      "step": 4056
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.393436140767102e-05,
+      "loss": 2.3555,
+      "step": 4057
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3930407275603e-05,
+      "loss": 1.9317,
+      "step": 4058
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3926453143535e-05,
+      "loss": 2.1661,
+      "step": 4059
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3922499011466986e-05,
+      "loss": 2.1822,
+      "step": 4060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3918544879398975e-05,
+      "loss": 2.0073,
+      "step": 4061
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.391459074733096e-05,
+      "loss": 1.8792,
+      "step": 4062
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.391063661526295e-05,
+      "loss": 1.9486,
+      "step": 4063
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.390668248319494e-05,
+      "loss": 2.0828,
+      "step": 4064
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.390272835112693e-05,
+      "loss": 2.2902,
+      "step": 4065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.389877421905892e-05,
+      "loss": 2.0178,
+      "step": 4066
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.389482008699091e-05,
+      "loss": 1.9421,
+      "step": 4067
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.38908659549229e-05,
+      "loss": 2.0269,
+      "step": 4068
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3886911822854886e-05,
+      "loss": 1.9563,
+      "step": 4069
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3882957690786874e-05,
+      "loss": 2.0997,
+      "step": 4070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.387900355871886e-05,
+      "loss": 2.052,
+      "step": 4071
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.387504942665085e-05,
+      "loss": 2.2262,
+      "step": 4072
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.387109529458284e-05,
+      "loss": 1.8539,
+      "step": 4073
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.386714116251483e-05,
+      "loss": 2.0774,
+      "step": 4074
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.386318703044682e-05,
+      "loss": 2.2665,
+      "step": 4075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.385923289837881e-05,
+      "loss": 2.0019,
+      "step": 4076
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3855278766310796e-05,
+      "loss": 1.8582,
+      "step": 4077
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3851324634242785e-05,
+      "loss": 2.1194,
+      "step": 4078
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3847370502174774e-05,
+      "loss": 2.0217,
+      "step": 4079
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.384341637010676e-05,
+      "loss": 2.002,
+      "step": 4080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.383946223803875e-05,
+      "loss": 1.9904,
+      "step": 4081
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.383550810597074e-05,
+      "loss": 2.3254,
+      "step": 4082
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.383155397390273e-05,
+      "loss": 1.9437,
+      "step": 4083
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.382759984183472e-05,
+      "loss": 2.1931,
+      "step": 4084
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3823645709766714e-05,
+      "loss": 2.0011,
+      "step": 4085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3819691577698696e-05,
+      "loss": 1.8579,
+      "step": 4086
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3815737445630684e-05,
+      "loss": 2.0656,
+      "step": 4087
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.381178331356267e-05,
+      "loss": 2.0029,
+      "step": 4088
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.380782918149467e-05,
+      "loss": 2.0174,
+      "step": 4089
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.380387504942665e-05,
+      "loss": 1.7889,
+      "step": 4090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.379992091735864e-05,
+      "loss": 1.9062,
+      "step": 4091
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.379596678529063e-05,
+      "loss": 2.0487,
+      "step": 4092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3792012653222624e-05,
+      "loss": 1.9655,
+      "step": 4093
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3788058521154606e-05,
+      "loss": 1.7027,
+      "step": 4094
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3784104389086595e-05,
+      "loss": 1.9369,
+      "step": 4095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3780150257018584e-05,
+      "loss": 1.9378,
+      "step": 4096
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.377619612495058e-05,
+      "loss": 1.8384,
+      "step": 4097
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.377224199288256e-05,
+      "loss": 2.115,
+      "step": 4098
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.376828786081455e-05,
+      "loss": 1.7579,
+      "step": 4099
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3764333728746546e-05,
+      "loss": 2.3061,
+      "step": 4100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3760379596678535e-05,
+      "loss": 2.1423,
+      "step": 4101
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.375642546461052e-05,
+      "loss": 2.2058,
+      "step": 4102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3752471332542506e-05,
+      "loss": 1.9508,
+      "step": 4103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.37485172004745e-05,
+      "loss": 2.0298,
+      "step": 4104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.374456306840649e-05,
+      "loss": 2.0189,
+      "step": 4105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.374060893633847e-05,
+      "loss": 1.9011,
+      "step": 4106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.373665480427046e-05,
+      "loss": 2.0132,
+      "step": 4107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.373270067220246e-05,
+      "loss": 2.3111,
+      "step": 4108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3728746540134445e-05,
+      "loss": 2.1872,
+      "step": 4109
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.372479240806643e-05,
+      "loss": 2.0373,
+      "step": 4110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3720838275998416e-05,
+      "loss": 2.0601,
+      "step": 4111
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.371688414393041e-05,
+      "loss": 1.8419,
+      "step": 4112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.37129300118624e-05,
+      "loss": 2.0721,
+      "step": 4113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.370897587979438e-05,
+      "loss": 2.0576,
+      "step": 4114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.370502174772637e-05,
+      "loss": 1.8423,
+      "step": 4115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.370106761565837e-05,
+      "loss": 2.2407,
+      "step": 4116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3697113483590356e-05,
+      "loss": 2.2606,
+      "step": 4117
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.369315935152234e-05,
+      "loss": 2.0216,
+      "step": 4118
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3689205219454334e-05,
+      "loss": 2.3207,
+      "step": 4119
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.368525108738632e-05,
+      "loss": 2.1217,
+      "step": 4120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.368129695531831e-05,
+      "loss": 2.1832,
+      "step": 4121
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3677342823250293e-05,
+      "loss": 1.909,
+      "step": 4122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.367338869118229e-05,
+      "loss": 2.0652,
+      "step": 4123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.366943455911428e-05,
+      "loss": 1.9996,
+      "step": 4124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.366548042704627e-05,
+      "loss": 2.1194,
+      "step": 4125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.366152629497825e-05,
+      "loss": 2.1047,
+      "step": 4126
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3657572162910244e-05,
+      "loss": 2.2154,
+      "step": 4127
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.365361803084223e-05,
+      "loss": 2.3038,
+      "step": 4128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.364966389877422e-05,
+      "loss": 2.1431,
+      "step": 4129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3645709766706204e-05,
+      "loss": 2.0276,
+      "step": 4130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.36417556346382e-05,
+      "loss": 1.9115,
+      "step": 4131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.363780150257019e-05,
+      "loss": 1.9431,
+      "step": 4132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.363384737050218e-05,
+      "loss": 2.0804,
+      "step": 4133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3629893238434166e-05,
+      "loss": 2.3919,
+      "step": 4134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3625939106366155e-05,
+      "loss": 2.0614,
+      "step": 4135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3621984974298144e-05,
+      "loss": 2.1271,
+      "step": 4136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.361803084223013e-05,
+      "loss": 2.1759,
+      "step": 4137
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.361407671016212e-05,
+      "loss": 2.1259,
+      "step": 4138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.361012257809411e-05,
+      "loss": 1.9329,
+      "step": 4139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.36061684460261e-05,
+      "loss": 2.0592,
+      "step": 4140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.360221431395809e-05,
+      "loss": 1.8039,
+      "step": 4141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.359826018189008e-05,
+      "loss": 2.027,
+      "step": 4142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3594306049822066e-05,
+      "loss": 2.0834,
+      "step": 4143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3590351917754054e-05,
+      "loss": 1.9724,
+      "step": 4144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.358639778568604e-05,
+      "loss": 2.0885,
+      "step": 4145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.358244365361803e-05,
+      "loss": 2.1441,
+      "step": 4146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.357848952155002e-05,
+      "loss": 2.1333,
+      "step": 4147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.357453538948201e-05,
+      "loss": 2.3094,
+      "step": 4148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3570581257414e-05,
+      "loss": 1.9051,
+      "step": 4149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.356662712534599e-05,
+      "loss": 2.0482,
+      "step": 4150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3562672993277976e-05,
+      "loss": 2.0326,
+      "step": 4151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3558718861209965e-05,
+      "loss": 2.3577,
+      "step": 4152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.355476472914196e-05,
+      "loss": 2.2255,
+      "step": 4153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.355081059707394e-05,
+      "loss": 1.9243,
+      "step": 4154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.354685646500593e-05,
+      "loss": 1.9506,
+      "step": 4155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.354290233293792e-05,
+      "loss": 1.9875,
+      "step": 4156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3538948200869916e-05,
+      "loss": 2.0481,
+      "step": 4157
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.35349940688019e-05,
+      "loss": 2.0572,
+      "step": 4158
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.353103993673389e-05,
+      "loss": 1.9659,
+      "step": 4159
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3527085804665876e-05,
+      "loss": 1.8228,
+      "step": 4160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.352313167259787e-05,
+      "loss": 1.8328,
+      "step": 4161
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.351917754052985e-05,
+      "loss": 2.1347,
+      "step": 4162
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.351522340846184e-05,
+      "loss": 1.7551,
+      "step": 4163
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.351126927639383e-05,
+      "loss": 1.8973,
+      "step": 4164
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3507315144325827e-05,
+      "loss": 2.0427,
+      "step": 4165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.350336101225781e-05,
+      "loss": 1.8026,
+      "step": 4166
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.34994068801898e-05,
+      "loss": 1.9166,
+      "step": 4167
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3495452748121786e-05,
+      "loss": 2.4578,
+      "step": 4168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.349149861605378e-05,
+      "loss": 1.9706,
+      "step": 4169
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3487544483985764e-05,
+      "loss": 2.0401,
+      "step": 4170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.348359035191775e-05,
+      "loss": 2.35,
+      "step": 4171
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.347963621984975e-05,
+      "loss": 2.0474,
+      "step": 4172
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.347568208778174e-05,
+      "loss": 2.2376,
+      "step": 4173
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.347172795571372e-05,
+      "loss": 2.0656,
+      "step": 4174
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.346777382364571e-05,
+      "loss": 2.2882,
+      "step": 4175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3463819691577704e-05,
+      "loss": 1.825,
+      "step": 4176
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.345986555950969e-05,
+      "loss": 2.1406,
+      "step": 4177
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3455911427441675e-05,
+      "loss": 2.1262,
+      "step": 4178
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.345195729537366e-05,
+      "loss": 1.7868,
+      "step": 4179
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.344800316330566e-05,
+      "loss": 2.1386,
+      "step": 4180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.344404903123765e-05,
+      "loss": 2.2309,
+      "step": 4181
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.344009489916963e-05,
+      "loss": 2.0601,
+      "step": 4182
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.343614076710162e-05,
+      "loss": 1.818,
+      "step": 4183
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3432186635033614e-05,
+      "loss": 2.0628,
+      "step": 4184
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.34282325029656e-05,
+      "loss": 2.2398,
+      "step": 4185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3424278370897585e-05,
+      "loss": 2.1182,
+      "step": 4186
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.342032423882958e-05,
+      "loss": 2.1415,
+      "step": 4187
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.341637010676157e-05,
+      "loss": 1.7301,
+      "step": 4188
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.341241597469356e-05,
+      "loss": 1.8569,
+      "step": 4189
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.340846184262554e-05,
+      "loss": 2.1748,
+      "step": 4190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3404507710557536e-05,
+      "loss": 2.245,
+      "step": 4191
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3400553578489525e-05,
+      "loss": 2.1391,
+      "step": 4192
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3396599446421514e-05,
+      "loss": 2.0238,
+      "step": 4193
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3392645314353496e-05,
+      "loss": 1.8658,
+      "step": 4194
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.338869118228549e-05,
+      "loss": 2.0074,
+      "step": 4195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.338473705021748e-05,
+      "loss": 2.0698,
+      "step": 4196
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.338078291814947e-05,
+      "loss": 2.0483,
+      "step": 4197
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.337682878608145e-05,
+      "loss": 2.0211,
+      "step": 4198
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.337287465401345e-05,
+      "loss": 2.1467,
+      "step": 4199
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3368920521945435e-05,
+      "loss": 2.1257,
+      "step": 4200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3364966389877424e-05,
+      "loss": 2.0033,
+      "step": 4201
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3361012257809406e-05,
+      "loss": 1.9198,
+      "step": 4202
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.33570581257414e-05,
+      "loss": 2.1362,
+      "step": 4203
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.335310399367339e-05,
+      "loss": 2.0793,
+      "step": 4204
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.334914986160538e-05,
+      "loss": 2.0551,
+      "step": 4205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.334519572953737e-05,
+      "loss": 1.9891,
+      "step": 4206
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.334124159746936e-05,
+      "loss": 2.0452,
+      "step": 4207
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3337287465401346e-05,
+      "loss": 1.927,
+      "step": 4208
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 1.8537,
+      "step": 4209
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3329379201265324e-05,
+      "loss": 2.1295,
+      "step": 4210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.332542506919731e-05,
+      "loss": 2.0217,
+      "step": 4211
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.33214709371293e-05,
+      "loss": 2.0044,
+      "step": 4212
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.331751680506129e-05,
+      "loss": 1.9199,
+      "step": 4213
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.331356267299328e-05,
+      "loss": 1.9574,
+      "step": 4214
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.330960854092527e-05,
+      "loss": 1.9504,
+      "step": 4215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.330565440885726e-05,
+      "loss": 1.9862,
+      "step": 4216
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3301700276789246e-05,
+      "loss": 2.0268,
+      "step": 4217
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3297746144721234e-05,
+      "loss": 2.0239,
+      "step": 4218
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.329379201265322e-05,
+      "loss": 1.7589,
+      "step": 4219
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.328983788058521e-05,
+      "loss": 1.9047,
+      "step": 4220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.328588374851721e-05,
+      "loss": 2.0746,
+      "step": 4221
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.328192961644919e-05,
+      "loss": 2.0963,
+      "step": 4222
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.327797548438118e-05,
+      "loss": 1.8941,
+      "step": 4223
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.327402135231317e-05,
+      "loss": 1.8409,
+      "step": 4224
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.327006722024516e-05,
+      "loss": 2.034,
+      "step": 4225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3266113088177145e-05,
+      "loss": 1.8975,
+      "step": 4226
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3262158956109134e-05,
+      "loss": 2.2358,
+      "step": 4227
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.325820482404112e-05,
+      "loss": 2.1077,
+      "step": 4228
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.325425069197312e-05,
+      "loss": 2.0531,
+      "step": 4229
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.32502965599051e-05,
+      "loss": 2.1474,
+      "step": 4230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.324634242783709e-05,
+      "loss": 2.2213,
+      "step": 4231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.324238829576908e-05,
+      "loss": 2.0872,
+      "step": 4232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3238434163701074e-05,
+      "loss": 2.0326,
+      "step": 4233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3234480031633056e-05,
+      "loss": 2.3445,
+      "step": 4234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3230525899565044e-05,
+      "loss": 1.9596,
+      "step": 4235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.322657176749703e-05,
+      "loss": 2.0376,
+      "step": 4236
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.322261763542903e-05,
+      "loss": 2.2076,
+      "step": 4237
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.321866350336101e-05,
+      "loss": 2.2223,
+      "step": 4238
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3214709371293e-05,
+      "loss": 1.9674,
+      "step": 4239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3210755239224995e-05,
+      "loss": 2.0856,
+      "step": 4240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3206801107156984e-05,
+      "loss": 2.132,
+      "step": 4241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3202846975088966e-05,
+      "loss": 2.1775,
+      "step": 4242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3198892843020955e-05,
+      "loss": 1.8097,
+      "step": 4243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.319493871095295e-05,
+      "loss": 1.9256,
+      "step": 4244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.319098457888494e-05,
+      "loss": 2.0477,
+      "step": 4245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.318703044681692e-05,
+      "loss": 1.9993,
+      "step": 4246
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.318307631474891e-05,
+      "loss": 2.1974,
+      "step": 4247
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3179122182680906e-05,
+      "loss": 1.9774,
+      "step": 4248
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3175168050612895e-05,
+      "loss": 2.1092,
+      "step": 4249
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.317121391854488e-05,
+      "loss": 1.9743,
+      "step": 4250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3167259786476866e-05,
+      "loss": 1.7506,
+      "step": 4251
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.316330565440886e-05,
+      "loss": 1.8009,
+      "step": 4252
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.315935152234085e-05,
+      "loss": 2.2986,
+      "step": 4253
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.315539739027283e-05,
+      "loss": 2.0681,
+      "step": 4254
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.315144325820482e-05,
+      "loss": 2.0622,
+      "step": 4255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3147489126136817e-05,
+      "loss": 2.0299,
+      "step": 4256
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3143534994068805e-05,
+      "loss": 1.8256,
+      "step": 4257
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.313958086200079e-05,
+      "loss": 1.7823,
+      "step": 4258
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.313562672993278e-05,
+      "loss": 1.9408,
+      "step": 4259
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.313167259786477e-05,
+      "loss": 1.9051,
+      "step": 4260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.312771846579676e-05,
+      "loss": 2.1155,
+      "step": 4261
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.312376433372875e-05,
+      "loss": 1.9861,
+      "step": 4262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.311981020166074e-05,
+      "loss": 1.9641,
+      "step": 4263
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.311585606959273e-05,
+      "loss": 1.9735,
+      "step": 4264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3111901937524716e-05,
+      "loss": 1.8026,
+      "step": 4265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3107947805456705e-05,
+      "loss": 2.3601,
+      "step": 4266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3103993673388694e-05,
+      "loss": 1.8255,
+      "step": 4267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.310003954132068e-05,
+      "loss": 1.9883,
+      "step": 4268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.309608540925267e-05,
+      "loss": 1.8643,
+      "step": 4269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.309213127718466e-05,
+      "loss": 2.0196,
+      "step": 4270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.308817714511665e-05,
+      "loss": 2.3025,
+      "step": 4271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.308422301304864e-05,
+      "loss": 2.0802,
+      "step": 4272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.308026888098063e-05,
+      "loss": 1.9639,
+      "step": 4273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3076314748912615e-05,
+      "loss": 1.9538,
+      "step": 4274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3072360616844604e-05,
+      "loss": 2.0649,
+      "step": 4275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.306840648477659e-05,
+      "loss": 1.9636,
+      "step": 4276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.306445235270858e-05,
+      "loss": 2.032,
+      "step": 4277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.306049822064057e-05,
+      "loss": 2.0538,
+      "step": 4278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.305654408857256e-05,
+      "loss": 1.8767,
+      "step": 4279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.305258995650455e-05,
+      "loss": 1.9725,
+      "step": 4280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.304863582443654e-05,
+      "loss": 2.3237,
+      "step": 4281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3044681692368526e-05,
+      "loss": 2.0078,
+      "step": 4282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3040727560300515e-05,
+      "loss": 1.8709,
+      "step": 4283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3036773428232504e-05,
+      "loss": 2.0015,
+      "step": 4284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.303281929616449e-05,
+      "loss": 2.1077,
+      "step": 4285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.302886516409648e-05,
+      "loss": 2.0607,
+      "step": 4286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.302491103202847e-05,
+      "loss": 1.9953,
+      "step": 4287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.302095689996046e-05,
+      "loss": 2.1317,
+      "step": 4288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.301700276789245e-05,
+      "loss": 1.8732,
+      "step": 4289
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.301304863582444e-05,
+      "loss": 1.9478,
+      "step": 4290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3009094503756426e-05,
+      "loss": 2.0653,
+      "step": 4291
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3005140371688414e-05,
+      "loss": 2.039,
+      "step": 4292
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.300118623962041e-05,
+      "loss": 2.1236,
+      "step": 4293
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.299723210755239e-05,
+      "loss": 1.9792,
+      "step": 4294
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.299327797548438e-05,
+      "loss": 1.9594,
+      "step": 4295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.298932384341637e-05,
+      "loss": 2.1699,
+      "step": 4296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2985369711348365e-05,
+      "loss": 1.8587,
+      "step": 4297
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.298141557928035e-05,
+      "loss": 1.9603,
+      "step": 4298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2977461447212336e-05,
+      "loss": 1.9412,
+      "step": 4299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2973507315144325e-05,
+      "loss": 1.9527,
+      "step": 4300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.296955318307632e-05,
+      "loss": 1.7393,
+      "step": 4301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.29655990510083e-05,
+      "loss": 2.0142,
+      "step": 4302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.296164491894029e-05,
+      "loss": 2.0301,
+      "step": 4303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.295769078687228e-05,
+      "loss": 1.9118,
+      "step": 4304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2953736654804276e-05,
+      "loss": 1.8799,
+      "step": 4305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.294978252273626e-05,
+      "loss": 1.934,
+      "step": 4306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.294582839066825e-05,
+      "loss": 1.7409,
+      "step": 4307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.294187425860024e-05,
+      "loss": 1.9979,
+      "step": 4308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.293792012653223e-05,
+      "loss": 2.031,
+      "step": 4309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.293396599446421e-05,
+      "loss": 1.931,
+      "step": 4310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.29300118623962e-05,
+      "loss": 2.0878,
+      "step": 4311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.29260577303282e-05,
+      "loss": 2.2383,
+      "step": 4312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2922103598260186e-05,
+      "loss": 2.1769,
+      "step": 4313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.291814946619217e-05,
+      "loss": 2.1713,
+      "step": 4314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.291419533412416e-05,
+      "loss": 1.8428,
+      "step": 4315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.291024120205615e-05,
+      "loss": 2.0254,
+      "step": 4316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.290628706998814e-05,
+      "loss": 2.109,
+      "step": 4317
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.290233293792013e-05,
+      "loss": 2.0954,
+      "step": 4318
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.289837880585211e-05,
+      "loss": 1.9602,
+      "step": 4319
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.289442467378411e-05,
+      "loss": 1.9885,
+      "step": 4320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.28904705417161e-05,
+      "loss": 2.049,
+      "step": 4321
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2886516409648086e-05,
+      "loss": 1.9886,
+      "step": 4322
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.288256227758007e-05,
+      "loss": 1.8443,
+      "step": 4323
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2878608145512064e-05,
+      "loss": 1.8696,
+      "step": 4324
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.287465401344405e-05,
+      "loss": 2.1707,
+      "step": 4325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.287069988137604e-05,
+      "loss": 2.0238,
+      "step": 4326
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.286674574930803e-05,
+      "loss": 2.2111,
+      "step": 4327
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.286279161724002e-05,
+      "loss": 2.1123,
+      "step": 4328
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.285883748517201e-05,
+      "loss": 1.9029,
+      "step": 4329
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2854883353103997e-05,
+      "loss": 2.0114,
+      "step": 4330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2850929221035985e-05,
+      "loss": 1.9059,
+      "step": 4331
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2846975088967974e-05,
+      "loss": 1.937,
+      "step": 4332
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.284302095689996e-05,
+      "loss": 1.9494,
+      "step": 4333
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.283906682483195e-05,
+      "loss": 1.8426,
+      "step": 4334
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.283511269276394e-05,
+      "loss": 2.3214,
+      "step": 4335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.283115856069593e-05,
+      "loss": 1.9646,
+      "step": 4336
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.282720442862792e-05,
+      "loss": 1.9379,
+      "step": 4337
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.282325029655991e-05,
+      "loss": 1.945,
+      "step": 4338
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2819296164491896e-05,
+      "loss": 1.9224,
+      "step": 4339
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2815342032423885e-05,
+      "loss": 1.8745,
+      "step": 4340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2811387900355874e-05,
+      "loss": 2.1978,
+      "step": 4341
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.280743376828786e-05,
+      "loss": 2.2507,
+      "step": 4342
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.280347963621985e-05,
+      "loss": 2.0468,
+      "step": 4343
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.279952550415184e-05,
+      "loss": 2.0447,
+      "step": 4344
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.279557137208383e-05,
+      "loss": 2.0145,
+      "step": 4345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.279161724001582e-05,
+      "loss": 2.1463,
+      "step": 4346
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2787663107947807e-05,
+      "loss": 1.9733,
+      "step": 4347
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2783708975879795e-05,
+      "loss": 2.4429,
+      "step": 4348
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2779754843811784e-05,
+      "loss": 2.293,
+      "step": 4349
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.277580071174377e-05,
+      "loss": 1.9925,
+      "step": 4350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.277184657967576e-05,
+      "loss": 1.9935,
+      "step": 4351
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.276789244760775e-05,
+      "loss": 1.8758,
+      "step": 4352
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.276393831553974e-05,
+      "loss": 1.9869,
+      "step": 4353
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.275998418347173e-05,
+      "loss": 1.9965,
+      "step": 4354
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.275603005140372e-05,
+      "loss": 2.0217,
+      "step": 4355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2752075919335706e-05,
+      "loss": 2.1091,
+      "step": 4356
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2748121787267695e-05,
+      "loss": 2.0024,
+      "step": 4357
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.2744167655199684e-05,
+      "loss": 2.0673,
+      "step": 4358
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.274021352313167e-05,
+      "loss": 1.9717,
+      "step": 4359
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.273625939106366e-05,
+      "loss": 2.2189,
+      "step": 4360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.273230525899566e-05,
+      "loss": 2.0236,
+      "step": 4361
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.272835112692764e-05,
+      "loss": 2.0871,
+      "step": 4362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.272439699485963e-05,
+      "loss": 2.2215,
+      "step": 4363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.272044286279162e-05,
+      "loss": 1.9701,
+      "step": 4364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.271648873072361e-05,
+      "loss": 2.2435,
+      "step": 4365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2712534598655594e-05,
+      "loss": 2.0016,
+      "step": 4366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.270858046658758e-05,
+      "loss": 1.8827,
+      "step": 4367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.270462633451957e-05,
+      "loss": 2.1483,
+      "step": 4368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.270067220245157e-05,
+      "loss": 2.2521,
+      "step": 4369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.269671807038355e-05,
+      "loss": 2.029,
+      "step": 4370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.269276393831554e-05,
+      "loss": 2.0374,
+      "step": 4371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.268880980624753e-05,
+      "loss": 1.8171,
+      "step": 4372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.268485567417952e-05,
+      "loss": 1.8894,
+      "step": 4373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.268090154211151e-05,
+      "loss": 2.0576,
+      "step": 4374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2676947410043494e-05,
+      "loss": 2.2331,
+      "step": 4375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.267299327797548e-05,
+      "loss": 2.0313,
+      "step": 4376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.266903914590748e-05,
+      "loss": 1.9467,
+      "step": 4377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.266508501383947e-05,
+      "loss": 2.072,
+      "step": 4378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.266113088177145e-05,
+      "loss": 1.9846,
+      "step": 4379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2657176749703445e-05,
+      "loss": 2.2094,
+      "step": 4380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2653222617635433e-05,
+      "loss": 2.0039,
+      "step": 4381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.264926848556742e-05,
+      "loss": 2.0859,
+      "step": 4382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2645314353499404e-05,
+      "loss": 2.0221,
+      "step": 4383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.26413602214314e-05,
+      "loss": 1.8952,
+      "step": 4384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.263740608936339e-05,
+      "loss": 2.0686,
+      "step": 4385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.263345195729538e-05,
+      "loss": 1.9201,
+      "step": 4386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.262949782522736e-05,
+      "loss": 1.9964,
+      "step": 4387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2625543693159355e-05,
+      "loss": 2.1674,
+      "step": 4388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2621589561091344e-05,
+      "loss": 1.9027,
+      "step": 4389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.261763542902333e-05,
+      "loss": 2.0955,
+      "step": 4390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2613681296955315e-05,
+      "loss": 2.1356,
+      "step": 4391
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.260972716488731e-05,
+      "loss": 1.7774,
+      "step": 4392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.26057730328193e-05,
+      "loss": 2.2674,
+      "step": 4393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.260181890075129e-05,
+      "loss": 2.0179,
+      "step": 4394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.259786476868328e-05,
+      "loss": 2.0401,
+      "step": 4395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2593910636615266e-05,
+      "loss": 2.1352,
+      "step": 4396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2589956504547255e-05,
+      "loss": 2.244,
+      "step": 4397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2586002372479244e-05,
+      "loss": 1.9231,
+      "step": 4398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.258204824041123e-05,
+      "loss": 2.1446,
+      "step": 4399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.257809410834322e-05,
+      "loss": 2.0358,
+      "step": 4400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.257413997627521e-05,
+      "loss": 1.8632,
+      "step": 4401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.25701858442072e-05,
+      "loss": 2.3651,
+      "step": 4402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.256623171213919e-05,
+      "loss": 2.0069,
+      "step": 4403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2562277580071177e-05,
+      "loss": 2.0912,
+      "step": 4404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2558323448003165e-05,
+      "loss": 2.0635,
+      "step": 4405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2554369315935154e-05,
+      "loss": 2.1209,
+      "step": 4406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.255041518386714e-05,
+      "loss": 2.3599,
+      "step": 4407
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.254646105179913e-05,
+      "loss": 2.0234,
+      "step": 4408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.254250691973112e-05,
+      "loss": 1.7354,
+      "step": 4409
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.253855278766311e-05,
+      "loss": 2.0319,
+      "step": 4410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.25345986555951e-05,
+      "loss": 1.9753,
+      "step": 4411
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.253064452352709e-05,
+      "loss": 2.1244,
+      "step": 4412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2526690391459076e-05,
+      "loss": 2.1334,
+      "step": 4413
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2522736259391065e-05,
+      "loss": 2.1958,
+      "step": 4414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2518782127323054e-05,
+      "loss": 1.8594,
+      "step": 4415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.251482799525504e-05,
+      "loss": 2.0424,
+      "step": 4416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.251087386318703e-05,
+      "loss": 2.2749,
+      "step": 4417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.250691973111902e-05,
+      "loss": 1.9676,
+      "step": 4418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.250296559905101e-05,
+      "loss": 1.8894,
+      "step": 4419
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2499011466983e-05,
+      "loss": 1.9913,
+      "step": 4420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2495057334914987e-05,
+      "loss": 1.9442,
+      "step": 4421
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2491103202846975e-05,
+      "loss": 2.1988,
+      "step": 4422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2487149070778964e-05,
+      "loss": 2.0941,
+      "step": 4423
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.248319493871095e-05,
+      "loss": 1.9293,
+      "step": 4424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.247924080664294e-05,
+      "loss": 2.1052,
+      "step": 4425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.247528667457493e-05,
+      "loss": 1.9778,
+      "step": 4426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.247133254250692e-05,
+      "loss": 2.0,
+      "step": 4427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.246737841043891e-05,
+      "loss": 1.9781,
+      "step": 4428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.24634242783709e-05,
+      "loss": 1.7589,
+      "step": 4429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2459470146302886e-05,
+      "loss": 2.0632,
+      "step": 4430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2455516014234875e-05,
+      "loss": 2.0076,
+      "step": 4431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2451561882166864e-05,
+      "loss": 2.0383,
+      "step": 4432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.244760775009886e-05,
+      "loss": 1.9268,
+      "step": 4433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.244365361803085e-05,
+      "loss": 2.0508,
+      "step": 4434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.243969948596283e-05,
+      "loss": 1.8999,
+      "step": 4435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.243574535389482e-05,
+      "loss": 2.0727,
+      "step": 4436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2431791221826815e-05,
+      "loss": 1.9097,
+      "step": 4437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2427837089758803e-05,
+      "loss": 1.9591,
+      "step": 4438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2423882957690785e-05,
+      "loss": 2.2701,
+      "step": 4439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2419928825622774e-05,
+      "loss": 2.062,
+      "step": 4440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.241597469355477e-05,
+      "loss": 1.9796,
+      "step": 4441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.241202056148676e-05,
+      "loss": 2.1445,
+      "step": 4442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.240806642941874e-05,
+      "loss": 2.0844,
+      "step": 4443
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.240411229735073e-05,
+      "loss": 1.9408,
+      "step": 4444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2400158165282725e-05,
+      "loss": 1.9655,
+      "step": 4445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2396204033214714e-05,
+      "loss": 2.0908,
+      "step": 4446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2392249901146696e-05,
+      "loss": 2.123,
+      "step": 4447
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.238829576907869e-05,
+      "loss": 1.8338,
+      "step": 4448
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.238434163701068e-05,
+      "loss": 1.984,
+      "step": 4449
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.238038750494267e-05,
+      "loss": 2.0096,
+      "step": 4450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.237643337287465e-05,
+      "loss": 1.8399,
+      "step": 4451
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.237247924080665e-05,
+      "loss": 2.2715,
+      "step": 4452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2368525108738636e-05,
+      "loss": 2.1607,
+      "step": 4453
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2364570976670625e-05,
+      "loss": 1.7799,
+      "step": 4454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.236061684460261e-05,
+      "loss": 2.1634,
+      "step": 4455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.23566627125346e-05,
+      "loss": 1.8719,
+      "step": 4456
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.235270858046659e-05,
+      "loss": 2.1446,
+      "step": 4457
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.234875444839858e-05,
+      "loss": 2.1698,
+      "step": 4458
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.234480031633056e-05,
+      "loss": 1.9799,
+      "step": 4459
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.234084618426256e-05,
+      "loss": 1.8275,
+      "step": 4460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2336892052194546e-05,
+      "loss": 1.9165,
+      "step": 4461
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2332937920126535e-05,
+      "loss": 2.1276,
+      "step": 4462
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.232898378805852e-05,
+      "loss": 1.9669,
+      "step": 4463
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.232502965599051e-05,
+      "loss": 2.1483,
+      "step": 4464
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.23210755239225e-05,
+      "loss": 2.0765,
+      "step": 4465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.231712139185449e-05,
+      "loss": 1.9477,
+      "step": 4466
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.231316725978648e-05,
+      "loss": 2.1839,
+      "step": 4467
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.230921312771847e-05,
+      "loss": 2.1068,
+      "step": 4468
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.230525899565046e-05,
+      "loss": 1.9718,
+      "step": 4469
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2301304863582446e-05,
+      "loss": 2.1011,
+      "step": 4470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2297350731514435e-05,
+      "loss": 1.7943,
+      "step": 4471
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2293396599446424e-05,
+      "loss": 2.0735,
+      "step": 4472
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.228944246737841e-05,
+      "loss": 1.8311,
+      "step": 4473
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.22854883353104e-05,
+      "loss": 2.0628,
+      "step": 4474
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.228153420324239e-05,
+      "loss": 2.129,
+      "step": 4475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.227758007117438e-05,
+      "loss": 1.8689,
+      "step": 4476
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.227362593910637e-05,
+      "loss": 2.003,
+      "step": 4477
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2269671807038356e-05,
+      "loss": 2.1704,
+      "step": 4478
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2265717674970345e-05,
+      "loss": 2.1047,
+      "step": 4479
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2261763542902334e-05,
+      "loss": 2.1324,
+      "step": 4480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.225780941083432e-05,
+      "loss": 2.019,
+      "step": 4481
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.225385527876631e-05,
+      "loss": 2.1459,
+      "step": 4482
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.22499011466983e-05,
+      "loss": 1.7516,
+      "step": 4483
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.224594701463029e-05,
+      "loss": 1.9673,
+      "step": 4484
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.224199288256228e-05,
+      "loss": 2.0892,
+      "step": 4485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.223803875049427e-05,
+      "loss": 1.8823,
+      "step": 4486
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2234084618426256e-05,
+      "loss": 1.8447,
+      "step": 4487
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2230130486358245e-05,
+      "loss": 2.087,
+      "step": 4488
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2226176354290234e-05,
+      "loss": 1.9571,
+      "step": 4489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.222222222222223e-05,
+      "loss": 1.7059,
+      "step": 4490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.221826809015421e-05,
+      "loss": 1.9968,
+      "step": 4491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.22143139580862e-05,
+      "loss": 2.3204,
+      "step": 4492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.221035982601819e-05,
+      "loss": 2.1572,
+      "step": 4493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2206405693950184e-05,
+      "loss": 1.8306,
+      "step": 4494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2202451561882167e-05,
+      "loss": 1.9241,
+      "step": 4495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2198497429814155e-05,
+      "loss": 1.9644,
+      "step": 4496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2194543297746144e-05,
+      "loss": 1.9398,
+      "step": 4497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.219058916567814e-05,
+      "loss": 2.1981,
+      "step": 4498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.218663503361012e-05,
+      "loss": 1.9849,
+      "step": 4499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.218268090154211e-05,
+      "loss": 2.094,
+      "step": 4500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2178726769474106e-05,
+      "loss": 2.0725,
+      "step": 4501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2174772637406095e-05,
+      "loss": 1.8951,
+      "step": 4502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.217081850533808e-05,
+      "loss": 2.1876,
+      "step": 4503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2166864373270066e-05,
+      "loss": 1.9962,
+      "step": 4504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.216291024120206e-05,
+      "loss": 1.829,
+      "step": 4505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.215895610913405e-05,
+      "loss": 1.9326,
+      "step": 4506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.215500197706603e-05,
+      "loss": 2.0525,
+      "step": 4507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.215104784499802e-05,
+      "loss": 1.7857,
+      "step": 4508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.214709371293002e-05,
+      "loss": 2.0974,
+      "step": 4509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2143139580862006e-05,
+      "loss": 1.8839,
+      "step": 4510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.213918544879399e-05,
+      "loss": 2.014,
+      "step": 4511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2135231316725977e-05,
+      "loss": 2.2725,
+      "step": 4512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.213127718465797e-05,
+      "loss": 1.8599,
+      "step": 4513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.212732305258996e-05,
+      "loss": 2.1379,
+      "step": 4514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.212336892052194e-05,
+      "loss": 2.0196,
+      "step": 4515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.211941478845393e-05,
+      "loss": 1.827,
+      "step": 4516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.211546065638593e-05,
+      "loss": 1.7271,
+      "step": 4517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2111506524317916e-05,
+      "loss": 1.8294,
+      "step": 4518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.21075523922499e-05,
+      "loss": 1.9652,
+      "step": 4519
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2103598260181894e-05,
+      "loss": 2.0994,
+      "step": 4520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.209964412811388e-05,
+      "loss": 2.0447,
+      "step": 4521
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.209568999604587e-05,
+      "loss": 2.1417,
+      "step": 4522
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2091735863977854e-05,
+      "loss": 1.9636,
+      "step": 4523
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.208778173190985e-05,
+      "loss": 1.9291,
+      "step": 4524
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.208382759984184e-05,
+      "loss": 2.02,
+      "step": 4525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.207987346777383e-05,
+      "loss": 1.8489,
+      "step": 4526
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.207591933570581e-05,
+      "loss": 1.9902,
+      "step": 4527
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2071965203637805e-05,
+      "loss": 2.0339,
+      "step": 4528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2068011071569793e-05,
+      "loss": 2.2742,
+      "step": 4529
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.206405693950178e-05,
+      "loss": 1.8637,
+      "step": 4530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2060102807433764e-05,
+      "loss": 2.3416,
+      "step": 4531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.205614867536576e-05,
+      "loss": 2.1974,
+      "step": 4532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.205219454329775e-05,
+      "loss": 1.9581,
+      "step": 4533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.204824041122974e-05,
+      "loss": 1.7395,
+      "step": 4534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2044286279161726e-05,
+      "loss": 2.0379,
+      "step": 4535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2040332147093715e-05,
+      "loss": 2.0792,
+      "step": 4536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2036378015025704e-05,
+      "loss": 1.8669,
+      "step": 4537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.203242388295769e-05,
+      "loss": 2.0876,
+      "step": 4538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.202846975088968e-05,
+      "loss": 2.105,
+      "step": 4539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.202451561882167e-05,
+      "loss": 1.7755,
+      "step": 4540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.202056148675366e-05,
+      "loss": 1.9408,
+      "step": 4541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.201660735468565e-05,
+      "loss": 2.2659,
+      "step": 4542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.201265322261764e-05,
+      "loss": 2.1111,
+      "step": 4543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2008699090549626e-05,
+      "loss": 2.0944,
+      "step": 4544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2004744958481615e-05,
+      "loss": 2.1669,
+      "step": 4545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2000790826413603e-05,
+      "loss": 1.9247,
+      "step": 4546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.199683669434559e-05,
+      "loss": 1.9583,
+      "step": 4547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.199288256227758e-05,
+      "loss": 1.9845,
+      "step": 4548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.198892843020957e-05,
+      "loss": 2.2529,
+      "step": 4549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.198497429814156e-05,
+      "loss": 2.0056,
+      "step": 4550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.198102016607355e-05,
+      "loss": 1.8431,
+      "step": 4551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1977066034005536e-05,
+      "loss": 2.0114,
+      "step": 4552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1973111901937525e-05,
+      "loss": 2.1977,
+      "step": 4553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.196915776986952e-05,
+      "loss": 2.1194,
+      "step": 4554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.19652036378015e-05,
+      "loss": 2.1239,
+      "step": 4555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.196124950573349e-05,
+      "loss": 1.9255,
+      "step": 4556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.195729537366548e-05,
+      "loss": 2.1887,
+      "step": 4557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1953341241597476e-05,
+      "loss": 1.9513,
+      "step": 4558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.194938710952946e-05,
+      "loss": 2.1179,
+      "step": 4559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.194543297746145e-05,
+      "loss": 2.1008,
+      "step": 4560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1941478845393436e-05,
+      "loss": 1.8787,
+      "step": 4561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.193752471332543e-05,
+      "loss": 1.7546,
+      "step": 4562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1933570581257414e-05,
+      "loss": 2.009,
+      "step": 4563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.19296164491894e-05,
+      "loss": 2.208,
+      "step": 4564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.192566231712139e-05,
+      "loss": 2.0567,
+      "step": 4565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.192170818505339e-05,
+      "loss": 1.9339,
+      "step": 4566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.191775405298537e-05,
+      "loss": 1.9669,
+      "step": 4567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.191379992091736e-05,
+      "loss": 2.0022,
+      "step": 4568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.190984578884935e-05,
+      "loss": 1.6928,
+      "step": 4569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.190589165678134e-05,
+      "loss": 1.9496,
+      "step": 4570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1901937524713324e-05,
+      "loss": 1.9926,
+      "step": 4571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.189798339264531e-05,
+      "loss": 2.0975,
+      "step": 4572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.189402926057731e-05,
+      "loss": 2.0414,
+      "step": 4573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.18900751285093e-05,
+      "loss": 2.0306,
+      "step": 4574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.188612099644128e-05,
+      "loss": 1.9304,
+      "step": 4575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.188216686437327e-05,
+      "loss": 1.8833,
+      "step": 4576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1878212732305264e-05,
+      "loss": 2.0689,
+      "step": 4577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.187425860023725e-05,
+      "loss": 2.3124,
+      "step": 4578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1870304468169235e-05,
+      "loss": 2.0714,
+      "step": 4579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1866350336101224e-05,
+      "loss": 1.8755,
+      "step": 4580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.186239620403322e-05,
+      "loss": 2.0911,
+      "step": 4581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.185844207196521e-05,
+      "loss": 1.8491,
+      "step": 4582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.185448793989719e-05,
+      "loss": 2.1252,
+      "step": 4583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.185053380782918e-05,
+      "loss": 1.9244,
+      "step": 4584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1846579675761175e-05,
+      "loss": 1.9894,
+      "step": 4585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.184262554369316e-05,
+      "loss": 2.1332,
+      "step": 4586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1838671411625145e-05,
+      "loss": 1.8668,
+      "step": 4587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.183471727955714e-05,
+      "loss": 2.113,
+      "step": 4588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.183076314748913e-05,
+      "loss": 2.053,
+      "step": 4589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.182680901542112e-05,
+      "loss": 2.1966,
+      "step": 4590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.18228548833531e-05,
+      "loss": 1.9996,
+      "step": 4591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1818900751285096e-05,
+      "loss": 1.822,
+      "step": 4592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1814946619217085e-05,
+      "loss": 2.283,
+      "step": 4593
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1810992487149074e-05,
+      "loss": 2.1312,
+      "step": 4594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1807038355081056e-05,
+      "loss": 2.407,
+      "step": 4595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.180308422301305e-05,
+      "loss": 2.3689,
+      "step": 4596
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.179913009094504e-05,
+      "loss": 1.9059,
+      "step": 4597
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.179517595887703e-05,
+      "loss": 2.1592,
+      "step": 4598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.179122182680901e-05,
+      "loss": 1.7268,
+      "step": 4599
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.178726769474101e-05,
+      "loss": 2.0325,
+      "step": 4600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1783313562672996e-05,
+      "loss": 1.9448,
+      "step": 4601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1779359430604985e-05,
+      "loss": 2.247,
+      "step": 4602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.177540529853697e-05,
+      "loss": 1.9671,
+      "step": 4603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.177145116646896e-05,
+      "loss": 2.2269,
+      "step": 4604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.176749703440095e-05,
+      "loss": 1.7747,
+      "step": 4605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.176354290233294e-05,
+      "loss": 1.8005,
+      "step": 4606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.175958877026493e-05,
+      "loss": 1.8965,
+      "step": 4607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.175563463819692e-05,
+      "loss": 1.8801,
+      "step": 4608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1751680506128906e-05,
+      "loss": 1.9622,
+      "step": 4609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1747726374060895e-05,
+      "loss": 1.8047,
+      "step": 4610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.1743772241992884e-05,
+      "loss": 2.0041,
+      "step": 4611
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.173981810992487e-05,
+      "loss": 2.0789,
+      "step": 4612
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.173586397785686e-05,
+      "loss": 2.2871,
+      "step": 4613
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.173190984578885e-05,
+      "loss": 1.7835,
+      "step": 4614
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.172795571372084e-05,
+      "loss": 2.0436,
+      "step": 4615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.172400158165283e-05,
+      "loss": 1.9839,
+      "step": 4616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.172004744958482e-05,
+      "loss": 2.1597,
+      "step": 4617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1716093317516806e-05,
+      "loss": 1.8263,
+      "step": 4618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1712139185448795e-05,
+      "loss": 2.1598,
+      "step": 4619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1708185053380783e-05,
+      "loss": 2.0948,
+      "step": 4620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.170423092131277e-05,
+      "loss": 2.0153,
+      "step": 4621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.170027678924477e-05,
+      "loss": 1.7894,
+      "step": 4622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.169632265717675e-05,
+      "loss": 1.7505,
+      "step": 4623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.169236852510874e-05,
+      "loss": 2.1175,
+      "step": 4624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.168841439304073e-05,
+      "loss": 1.8325,
+      "step": 4625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.168446026097272e-05,
+      "loss": 2.0613,
+      "step": 4626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1680506128904705e-05,
+      "loss": 1.9123,
+      "step": 4627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1676551996836694e-05,
+      "loss": 1.7848,
+      "step": 4628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.167259786476868e-05,
+      "loss": 1.8164,
+      "step": 4629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.166864373270068e-05,
+      "loss": 2.0519,
+      "step": 4630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.166468960063266e-05,
+      "loss": 1.9448,
+      "step": 4631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.166073546856465e-05,
+      "loss": 2.1838,
+      "step": 4632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.165678133649664e-05,
+      "loss": 2.0859,
+      "step": 4633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1652827204428634e-05,
+      "loss": 1.9367,
+      "step": 4634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1648873072360616e-05,
+      "loss": 1.9229,
+      "step": 4635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1644918940292605e-05,
+      "loss": 2.0288,
+      "step": 4636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1640964808224594e-05,
+      "loss": 1.8866,
+      "step": 4637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.163701067615659e-05,
+      "loss": 2.1515,
+      "step": 4638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.163305654408857e-05,
+      "loss": 1.9184,
+      "step": 4639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.162910241202056e-05,
+      "loss": 1.976,
+      "step": 4640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1625148279952556e-05,
+      "loss": 2.0681,
+      "step": 4641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1621194147884544e-05,
+      "loss": 2.0039,
+      "step": 4642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1617240015816526e-05,
+      "loss": 1.9464,
+      "step": 4643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1613285883748515e-05,
+      "loss": 1.9734,
+      "step": 4644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.160933175168051e-05,
+      "loss": 2.1664,
+      "step": 4645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.16053776196125e-05,
+      "loss": 1.7674,
+      "step": 4646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.160142348754448e-05,
+      "loss": 1.8398,
+      "step": 4647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.159746935547647e-05,
+      "loss": 1.8517,
+      "step": 4648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1593515223408466e-05,
+      "loss": 1.9298,
+      "step": 4649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1589561091340455e-05,
+      "loss": 2.3045,
+      "step": 4650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.158560695927244e-05,
+      "loss": 1.9838,
+      "step": 4651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1581652827204426e-05,
+      "loss": 2.5003,
+      "step": 4652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.157769869513642e-05,
+      "loss": 2.0296,
+      "step": 4653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.157374456306841e-05,
+      "loss": 2.1712,
+      "step": 4654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.156979043100039e-05,
+      "loss": 2.0537,
+      "step": 4655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.156583629893239e-05,
+      "loss": 2.0404,
+      "step": 4656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.156188216686438e-05,
+      "loss": 2.1342,
+      "step": 4657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1557928034796366e-05,
+      "loss": 2.0955,
+      "step": 4658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.155397390272835e-05,
+      "loss": 1.9414,
+      "step": 4659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.155001977066034e-05,
+      "loss": 2.0508,
+      "step": 4660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.154606563859233e-05,
+      "loss": 1.9981,
+      "step": 4661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.154211150652432e-05,
+      "loss": 1.9428,
+      "step": 4662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.15381573744563e-05,
+      "loss": 1.9275,
+      "step": 4663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.15342032423883e-05,
+      "loss": 1.934,
+      "step": 4664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.153024911032029e-05,
+      "loss": 2.0778,
+      "step": 4665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1526294978252276e-05,
+      "loss": 2.202,
+      "step": 4666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1522340846184265e-05,
+      "loss": 1.9581,
+      "step": 4667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1518386714116254e-05,
+      "loss": 2.026,
+      "step": 4668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.151443258204824e-05,
+      "loss": 2.0925,
+      "step": 4669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.151047844998023e-05,
+      "loss": 1.9559,
+      "step": 4670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.150652431791222e-05,
+      "loss": 2.0753,
+      "step": 4671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.150257018584421e-05,
+      "loss": 2.0374,
+      "step": 4672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.14986160537762e-05,
+      "loss": 1.8356,
+      "step": 4673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.149466192170819e-05,
+      "loss": 1.7456,
+      "step": 4674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1490707789640176e-05,
+      "loss": 1.9466,
+      "step": 4675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1486753657572165e-05,
+      "loss": 1.9685,
+      "step": 4676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.148279952550415e-05,
+      "loss": 1.7067,
+      "step": 4677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.147884539343614e-05,
+      "loss": 1.6798,
+      "step": 4678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.147489126136813e-05,
+      "loss": 2.0885,
+      "step": 4679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.147093712930012e-05,
+      "loss": 1.8607,
+      "step": 4680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.146698299723211e-05,
+      "loss": 2.1443,
+      "step": 4681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.14630288651641e-05,
+      "loss": 1.8022,
+      "step": 4682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1459074733096086e-05,
+      "loss": 1.9058,
+      "step": 4683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1455120601028075e-05,
+      "loss": 2.1013,
+      "step": 4684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1451166468960064e-05,
+      "loss": 2.2573,
+      "step": 4685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.144721233689205e-05,
+      "loss": 1.8444,
+      "step": 4686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.144325820482404e-05,
+      "loss": 2.1925,
+      "step": 4687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.143930407275603e-05,
+      "loss": 1.9793,
+      "step": 4688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.143534994068802e-05,
+      "loss": 2.0538,
+      "step": 4689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.143139580862001e-05,
+      "loss": 2.1085,
+      "step": 4690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1427441676552e-05,
+      "loss": 1.9581,
+      "step": 4691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1423487544483986e-05,
+      "loss": 1.6978,
+      "step": 4692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1419533412415975e-05,
+      "loss": 1.9523,
+      "step": 4693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.141557928034797e-05,
+      "loss": 2.0808,
+      "step": 4694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.141162514827995e-05,
+      "loss": 2.2042,
+      "step": 4695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.140767101621194e-05,
+      "loss": 1.7488,
+      "step": 4696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.140371688414393e-05,
+      "loss": 2.145,
+      "step": 4697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1399762752075926e-05,
+      "loss": 2.0842,
+      "step": 4698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.139580862000791e-05,
+      "loss": 1.8347,
+      "step": 4699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1391854487939896e-05,
+      "loss": 1.8188,
+      "step": 4700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1387900355871885e-05,
+      "loss": 1.8901,
+      "step": 4701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.138394622380388e-05,
+      "loss": 2.1622,
+      "step": 4702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.137999209173586e-05,
+      "loss": 2.3726,
+      "step": 4703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.137603795966785e-05,
+      "loss": 1.7202,
+      "step": 4704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.137208382759984e-05,
+      "loss": 2.0838,
+      "step": 4705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1368129695531836e-05,
+      "loss": 1.8138,
+      "step": 4706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.136417556346382e-05,
+      "loss": 2.037,
+      "step": 4707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.136022143139581e-05,
+      "loss": 1.8394,
+      "step": 4708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.13562672993278e-05,
+      "loss": 1.8353,
+      "step": 4709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.135231316725979e-05,
+      "loss": 2.0253,
+      "step": 4710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1348359035191773e-05,
+      "loss": 2.1248,
+      "step": 4711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.134440490312376e-05,
+      "loss": 2.0025,
+      "step": 4712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.134045077105576e-05,
+      "loss": 1.9312,
+      "step": 4713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.133649663898775e-05,
+      "loss": 1.8836,
+      "step": 4714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.133254250691973e-05,
+      "loss": 2.0947,
+      "step": 4715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.132858837485172e-05,
+      "loss": 2.0555,
+      "step": 4716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.132463424278371e-05,
+      "loss": 1.9514,
+      "step": 4717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.13206801107157e-05,
+      "loss": 2.003,
+      "step": 4718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1316725978647684e-05,
+      "loss": 1.944,
+      "step": 4719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.131277184657967e-05,
+      "loss": 1.944,
+      "step": 4720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.130881771451167e-05,
+      "loss": 1.8686,
+      "step": 4721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.130486358244366e-05,
+      "loss": 1.9145,
+      "step": 4722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1300909450375646e-05,
+      "loss": 2.0323,
+      "step": 4723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.129695531830763e-05,
+      "loss": 2.0279,
+      "step": 4724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1293001186239624e-05,
+      "loss": 1.8916,
+      "step": 4725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.128904705417161e-05,
+      "loss": 2.0614,
+      "step": 4726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.12850929221036e-05,
+      "loss": 1.9292,
+      "step": 4727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.128113879003559e-05,
+      "loss": 1.8624,
+      "step": 4728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.127718465796758e-05,
+      "loss": 2.0938,
+      "step": 4729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.127323052589957e-05,
+      "loss": 2.039,
+      "step": 4730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.126927639383156e-05,
+      "loss": 1.8741,
+      "step": 4731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1265322261763546e-05,
+      "loss": 1.7609,
+      "step": 4732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1261368129695534e-05,
+      "loss": 1.9467,
+      "step": 4733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.125741399762752e-05,
+      "loss": 1.8221,
+      "step": 4734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.125345986555951e-05,
+      "loss": 1.7741,
+      "step": 4735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.12495057334915e-05,
+      "loss": 2.3742,
+      "step": 4736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.124555160142349e-05,
+      "loss": 2.0481,
+      "step": 4737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.124159746935548e-05,
+      "loss": 2.141,
+      "step": 4738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.123764333728747e-05,
+      "loss": 2.0402,
+      "step": 4739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1233689205219456e-05,
+      "loss": 2.0102,
+      "step": 4740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1229735073151445e-05,
+      "loss": 1.8797,
+      "step": 4741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.1225780941083434e-05,
+      "loss": 2.2764,
+      "step": 4742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.122182680901542e-05,
+      "loss": 1.7953,
+      "step": 4743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.121787267694741e-05,
+      "loss": 2.1574,
+      "step": 4744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.12139185448794e-05,
+      "loss": 1.8246,
+      "step": 4745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.120996441281139e-05,
+      "loss": 2.0829,
+      "step": 4746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.120601028074338e-05,
+      "loss": 2.3577,
+      "step": 4747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.120205614867537e-05,
+      "loss": 1.8842,
+      "step": 4748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1198102016607356e-05,
+      "loss": 2.1651,
+      "step": 4749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1194147884539345e-05,
+      "loss": 1.7566,
+      "step": 4750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.119019375247133e-05,
+      "loss": 1.9143,
+      "step": 4751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.118623962040332e-05,
+      "loss": 1.9695,
+      "step": 4752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.118228548833531e-05,
+      "loss": 2.0213,
+      "step": 4753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.11783313562673e-05,
+      "loss": 2.1988,
+      "step": 4754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.117437722419929e-05,
+      "loss": 1.8056,
+      "step": 4755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.117042309213128e-05,
+      "loss": 2.2735,
+      "step": 4756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1166468960063266e-05,
+      "loss": 2.2005,
+      "step": 4757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1162514827995255e-05,
+      "loss": 1.9808,
+      "step": 4758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1158560695927244e-05,
+      "loss": 1.9279,
+      "step": 4759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.115460656385923e-05,
+      "loss": 1.8282,
+      "step": 4760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.115065243179122e-05,
+      "loss": 1.9098,
+      "step": 4761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.114669829972322e-05,
+      "loss": 1.8679,
+      "step": 4762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.11427441676552e-05,
+      "loss": 1.9722,
+      "step": 4763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.113879003558719e-05,
+      "loss": 2.1789,
+      "step": 4764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.113483590351918e-05,
+      "loss": 2.1316,
+      "step": 4765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.113088177145117e-05,
+      "loss": 2.0106,
+      "step": 4766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1126927639383155e-05,
+      "loss": 1.7718,
+      "step": 4767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1122973507315143e-05,
+      "loss": 2.261,
+      "step": 4768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.111901937524713e-05,
+      "loss": 2.1073,
+      "step": 4769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.111506524317913e-05,
+      "loss": 2.2489,
+      "step": 4770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 1.8264,
+      "step": 4771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.11071569790431e-05,
+      "loss": 1.9843,
+      "step": 4772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.110320284697509e-05,
+      "loss": 2.1349,
+      "step": 4773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.109924871490708e-05,
+      "loss": 1.8995,
+      "step": 4774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1095294582839065e-05,
+      "loss": 2.3396,
+      "step": 4775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1091340450771054e-05,
+      "loss": 1.9627,
+      "step": 4776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.108738631870304e-05,
+      "loss": 2.1501,
+      "step": 4777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.108343218663504e-05,
+      "loss": 2.0186,
+      "step": 4778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.107947805456702e-05,
+      "loss": 2.2149,
+      "step": 4779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.107552392249901e-05,
+      "loss": 2.1322,
+      "step": 4780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1071569790431005e-05,
+      "loss": 1.919,
+      "step": 4781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1067615658362994e-05,
+      "loss": 1.9875,
+      "step": 4782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.106366152629498e-05,
+      "loss": 2.1128,
+      "step": 4783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1059707394226965e-05,
+      "loss": 1.8097,
+      "step": 4784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.105575326215896e-05,
+      "loss": 2.0565,
+      "step": 4785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.105179913009095e-05,
+      "loss": 1.895,
+      "step": 4786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.104784499802294e-05,
+      "loss": 1.9569,
+      "step": 4787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.104389086595492e-05,
+      "loss": 2.1042,
+      "step": 4788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1039936733886916e-05,
+      "loss": 2.0734,
+      "step": 4789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1035982601818904e-05,
+      "loss": 1.9985,
+      "step": 4790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.103202846975089e-05,
+      "loss": 2.1059,
+      "step": 4791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1028074337682875e-05,
+      "loss": 2.0424,
+      "step": 4792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.102412020561487e-05,
+      "loss": 1.9036,
+      "step": 4793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.102016607354686e-05,
+      "loss": 2.1897,
+      "step": 4794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.101621194147885e-05,
+      "loss": 2.0631,
+      "step": 4795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.101225780941084e-05,
+      "loss": 2.1791,
+      "step": 4796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1008303677342826e-05,
+      "loss": 1.9292,
+      "step": 4797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1004349545274815e-05,
+      "loss": 2.0872,
+      "step": 4798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.1000395413206804e-05,
+      "loss": 2.1298,
+      "step": 4799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.099644128113879e-05,
+      "loss": 2.0518,
+      "step": 4800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.099248714907078e-05,
+      "loss": 2.1634,
+      "step": 4801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.098853301700277e-05,
+      "loss": 2.241,
+      "step": 4802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.098457888493476e-05,
+      "loss": 2.1551,
+      "step": 4803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.098062475286675e-05,
+      "loss": 2.09,
+      "step": 4804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.097667062079874e-05,
+      "loss": 2.126,
+      "step": 4805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0972716488730726e-05,
+      "loss": 2.1021,
+      "step": 4806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0968762356662714e-05,
+      "loss": 1.7835,
+      "step": 4807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.09648082245947e-05,
+      "loss": 1.8282,
+      "step": 4808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.096085409252669e-05,
+      "loss": 1.9237,
+      "step": 4809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.095689996045868e-05,
+      "loss": 1.8201,
+      "step": 4810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.095294582839067e-05,
+      "loss": 1.7997,
+      "step": 4811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.094899169632266e-05,
+      "loss": 2.1301,
+      "step": 4812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.094503756425465e-05,
+      "loss": 2.1906,
+      "step": 4813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0941083432186636e-05,
+      "loss": 2.0313,
+      "step": 4814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0937129300118625e-05,
+      "loss": 2.0238,
+      "step": 4815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0933175168050614e-05,
+      "loss": 2.1202,
+      "step": 4816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.09292210359826e-05,
+      "loss": 2.0015,
+      "step": 4817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.092526690391459e-05,
+      "loss": 1.9008,
+      "step": 4818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.092131277184658e-05,
+      "loss": 1.8365,
+      "step": 4819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.091735863977857e-05,
+      "loss": 2.0885,
+      "step": 4820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.091340450771056e-05,
+      "loss": 1.8773,
+      "step": 4821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.090945037564255e-05,
+      "loss": 1.7741,
+      "step": 4822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0905496243574536e-05,
+      "loss": 2.2583,
+      "step": 4823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0901542111506524e-05,
+      "loss": 2.1701,
+      "step": 4824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.089758797943851e-05,
+      "loss": 2.0257,
+      "step": 4825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.08936338473705e-05,
+      "loss": 1.9258,
+      "step": 4826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.088967971530249e-05,
+      "loss": 2.0462,
+      "step": 4827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.088572558323448e-05,
+      "loss": 1.8998,
+      "step": 4828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.088177145116647e-05,
+      "loss": 2.1998,
+      "step": 4829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0877817319098464e-05,
+      "loss": 1.9061,
+      "step": 4830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0873863187030446e-05,
+      "loss": 2.066,
+      "step": 4831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0869909054962435e-05,
+      "loss": 2.2506,
+      "step": 4832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0865954922894424e-05,
+      "loss": 2.2868,
+      "step": 4833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.086200079082642e-05,
+      "loss": 2.0447,
+      "step": 4834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.08580466587584e-05,
+      "loss": 1.8057,
+      "step": 4835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.085409252669039e-05,
+      "loss": 2.0717,
+      "step": 4836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.085013839462238e-05,
+      "loss": 2.0241,
+      "step": 4837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0846184262554375e-05,
+      "loss": 1.9685,
+      "step": 4838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0842230130486364e-05,
+      "loss": 2.0117,
+      "step": 4839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0838275998418346e-05,
+      "loss": 1.9908,
+      "step": 4840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0834321866350335e-05,
+      "loss": 2.4673,
+      "step": 4841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.083036773428233e-05,
+      "loss": 2.1172,
+      "step": 4842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.082641360221432e-05,
+      "loss": 2.0883,
+      "step": 4843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.08224594701463e-05,
+      "loss": 2.0609,
+      "step": 4844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.081850533807829e-05,
+      "loss": 2.179,
+      "step": 4845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0814551206010285e-05,
+      "loss": 2.0201,
+      "step": 4846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0810597073942274e-05,
+      "loss": 1.7869,
+      "step": 4847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0806642941874256e-05,
+      "loss": 2.0589,
+      "step": 4848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.080268880980625e-05,
+      "loss": 1.9974,
+      "step": 4849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.079873467773824e-05,
+      "loss": 2.1507,
+      "step": 4850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.079478054567023e-05,
+      "loss": 1.8366,
+      "step": 4851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.079082641360221e-05,
+      "loss": 1.9563,
+      "step": 4852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.078687228153421e-05,
+      "loss": 2.3498,
+      "step": 4853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0782918149466196e-05,
+      "loss": 2.0165,
+      "step": 4854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0778964017398185e-05,
+      "loss": 2.0144,
+      "step": 4855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.077500988533017e-05,
+      "loss": 2.013,
+      "step": 4856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.077105575326216e-05,
+      "loss": 1.9482,
+      "step": 4857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.076710162119415e-05,
+      "loss": 2.1046,
+      "step": 4858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.076314748912614e-05,
+      "loss": 1.9983,
+      "step": 4859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.075919335705812e-05,
+      "loss": 2.0233,
+      "step": 4860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.075523922499012e-05,
+      "loss": 2.1194,
+      "step": 4861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.075128509292211e-05,
+      "loss": 1.716,
+      "step": 4862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.0747330960854096e-05,
+      "loss": 1.8871,
+      "step": 4863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.074337682878608e-05,
+      "loss": 1.7086,
+      "step": 4864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.073942269671807e-05,
+      "loss": 1.9501,
+      "step": 4865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.073546856465006e-05,
+      "loss": 2.191,
+      "step": 4866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.073151443258205e-05,
+      "loss": 1.8475,
+      "step": 4867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.072756030051404e-05,
+      "loss": 2.1311,
+      "step": 4868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.072360616844603e-05,
+      "loss": 1.9691,
+      "step": 4869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.071965203637802e-05,
+      "loss": 2.0776,
+      "step": 4870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0715697904310006e-05,
+      "loss": 2.0025,
+      "step": 4871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0711743772241995e-05,
+      "loss": 2.186,
+      "step": 4872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0707789640173984e-05,
+      "loss": 1.9995,
+      "step": 4873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.070383550810597e-05,
+      "loss": 2.0516,
+      "step": 4874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.069988137603796e-05,
+      "loss": 2.2312,
+      "step": 4875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.069592724396995e-05,
+      "loss": 1.9887,
+      "step": 4876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.069197311190194e-05,
+      "loss": 1.784,
+      "step": 4877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.068801897983393e-05,
+      "loss": 1.8358,
+      "step": 4878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.068406484776592e-05,
+      "loss": 2.0428,
+      "step": 4879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0680110715697906e-05,
+      "loss": 1.8357,
+      "step": 4880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0676156583629894e-05,
+      "loss": 2.2469,
+      "step": 4881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.067220245156188e-05,
+      "loss": 1.796,
+      "step": 4882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.066824831949387e-05,
+      "loss": 1.9076,
+      "step": 4883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.066429418742586e-05,
+      "loss": 2.0306,
+      "step": 4884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.066034005535785e-05,
+      "loss": 1.9142,
+      "step": 4885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.065638592328984e-05,
+      "loss": 1.7646,
+      "step": 4886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.065243179122183e-05,
+      "loss": 1.8537,
+      "step": 4887
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0648477659153816e-05,
+      "loss": 1.8888,
+      "step": 4888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0644523527085805e-05,
+      "loss": 2.0254,
+      "step": 4889
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0640569395017794e-05,
+      "loss": 2.1618,
+      "step": 4890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.063661526294978e-05,
+      "loss": 1.7095,
+      "step": 4891
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.063266113088177e-05,
+      "loss": 1.9936,
+      "step": 4892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.062870699881376e-05,
+      "loss": 2.0562,
+      "step": 4893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.062475286674575e-05,
+      "loss": 1.9323,
+      "step": 4894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0620798734677745e-05,
+      "loss": 1.8477,
+      "step": 4895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.061684460260973e-05,
+      "loss": 1.956,
+      "step": 4896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0612890470541716e-05,
+      "loss": 2.1613,
+      "step": 4897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0608936338473704e-05,
+      "loss": 2.1664,
+      "step": 4898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.06049822064057e-05,
+      "loss": 1.9564,
+      "step": 4899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.060102807433768e-05,
+      "loss": 1.9506,
+      "step": 4900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.059707394226967e-05,
+      "loss": 2.2409,
+      "step": 4901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0593119810201667e-05,
+      "loss": 1.8978,
+      "step": 4902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0589165678133655e-05,
+      "loss": 1.8247,
+      "step": 4903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.058521154606564e-05,
+      "loss": 2.072,
+      "step": 4904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0581257413997626e-05,
+      "loss": 2.3633,
+      "step": 4905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.057730328192962e-05,
+      "loss": 1.9675,
+      "step": 4906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.057334914986161e-05,
+      "loss": 2.2059,
+      "step": 4907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.056939501779359e-05,
+      "loss": 1.8779,
+      "step": 4908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.056544088572558e-05,
+      "loss": 1.9069,
+      "step": 4909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.056148675365758e-05,
+      "loss": 2.0635,
+      "step": 4910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0557532621589566e-05,
+      "loss": 1.9637,
+      "step": 4911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.055357848952155e-05,
+      "loss": 1.8143,
+      "step": 4912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.054962435745354e-05,
+      "loss": 2.0478,
+      "step": 4913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.054567022538553e-05,
+      "loss": 2.0066,
+      "step": 4914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.054171609331752e-05,
+      "loss": 2.0684,
+      "step": 4915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.05377619612495e-05,
+      "loss": 1.9842,
+      "step": 4916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.05338078291815e-05,
+      "loss": 1.7602,
+      "step": 4917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.052985369711349e-05,
+      "loss": 1.9327,
+      "step": 4918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0525899565045477e-05,
+      "loss": 2.0925,
+      "step": 4919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.052194543297746e-05,
+      "loss": 2.0448,
+      "step": 4920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0517991300909454e-05,
+      "loss": 2.274,
+      "step": 4921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0514037168841443e-05,
+      "loss": 2.0305,
+      "step": 4922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.051008303677343e-05,
+      "loss": 2.2015,
+      "step": 4923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0506128904705417e-05,
+      "loss": 2.05,
+      "step": 4924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.050217477263741e-05,
+      "loss": 2.0109,
+      "step": 4925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.04982206405694e-05,
+      "loss": 2.3169,
+      "step": 4926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0494266508501384e-05,
+      "loss": 2.1247,
+      "step": 4927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0490312376433373e-05,
+      "loss": 1.9086,
+      "step": 4928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0486358244365365e-05,
+      "loss": 1.9436,
+      "step": 4929
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0482404112297354e-05,
+      "loss": 2.0004,
+      "step": 4930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.047844998022934e-05,
+      "loss": 1.9866,
+      "step": 4931
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0474495848161328e-05,
+      "loss": 2.008,
+      "step": 4932
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.047054171609332e-05,
+      "loss": 1.9759,
+      "step": 4933
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.046658758402531e-05,
+      "loss": 2.3376,
+      "step": 4934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0462633451957294e-05,
+      "loss": 2.0222,
+      "step": 4935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0458679319889287e-05,
+      "loss": 1.9714,
+      "step": 4936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0454725187821275e-05,
+      "loss": 1.9217,
+      "step": 4937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0450771055753264e-05,
+      "loss": 1.9394,
+      "step": 4938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.044681692368525e-05,
+      "loss": 1.7323,
+      "step": 4939
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0442862791617245e-05,
+      "loss": 1.8927,
+      "step": 4940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.043890865954923e-05,
+      "loss": 1.9685,
+      "step": 4941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.043495452748122e-05,
+      "loss": 2.2442,
+      "step": 4942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0431000395413205e-05,
+      "loss": 1.8228,
+      "step": 4943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.04270462633452e-05,
+      "loss": 2.0655,
+      "step": 4944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0423092131277186e-05,
+      "loss": 2.0415,
+      "step": 4945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0419137999209175e-05,
+      "loss": 2.0363,
+      "step": 4946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.041518386714116e-05,
+      "loss": 2.0405,
+      "step": 4947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0411229735073156e-05,
+      "loss": 2.0192,
+      "step": 4948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.040727560300514e-05,
+      "loss": 1.8553,
+      "step": 4949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.040332147093713e-05,
+      "loss": 2.2021,
+      "step": 4950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0399367338869116e-05,
+      "loss": 1.9941,
+      "step": 4951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.039541320680111e-05,
+      "loss": 2.043,
+      "step": 4952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0391459074733097e-05,
+      "loss": 2.0671,
+      "step": 4953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0387504942665086e-05,
+      "loss": 1.7241,
+      "step": 4954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0383550810597078e-05,
+      "loss": 2.1111,
+      "step": 4955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0379596678529067e-05,
+      "loss": 2.2593,
+      "step": 4956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0375642546461052e-05,
+      "loss": 1.9295,
+      "step": 4957
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.037168841439304e-05,
+      "loss": 2.0468,
+      "step": 4958
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0367734282325033e-05,
+      "loss": 1.711,
+      "step": 4959
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0363780150257022e-05,
+      "loss": 1.8166,
+      "step": 4960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0359826018189007e-05,
+      "loss": 2.1666,
+      "step": 4961
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0355871886120996e-05,
+      "loss": 1.8445,
+      "step": 4962
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.035191775405299e-05,
+      "loss": 2.0082,
+      "step": 4963
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0347963621984977e-05,
+      "loss": 2.0186,
+      "step": 4964
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0344009489916963e-05,
+      "loss": 2.0399,
+      "step": 4965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.034005535784895e-05,
+      "loss": 2.2457,
+      "step": 4966
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0336101225780944e-05,
+      "loss": 2.0107,
+      "step": 4967
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0332147093712933e-05,
+      "loss": 2.1394,
+      "step": 4968
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0328192961644918e-05,
+      "loss": 2.0549,
+      "step": 4969
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0324238829576914e-05,
+      "loss": 1.8766,
+      "step": 4970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.03202846975089e-05,
+      "loss": 2.0446,
+      "step": 4971
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0316330565440888e-05,
+      "loss": 2.0226,
+      "step": 4972
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0312376433372873e-05,
+      "loss": 1.8654,
+      "step": 4973
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.030842230130487e-05,
+      "loss": 2.0751,
+      "step": 4974
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0304468169236854e-05,
+      "loss": 1.988,
+      "step": 4975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0300514037168843e-05,
+      "loss": 2.0271,
+      "step": 4976
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.029655990510083e-05,
+      "loss": 1.8535,
+      "step": 4977
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0292605773032824e-05,
+      "loss": 1.7249,
+      "step": 4978
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.028865164096481e-05,
+      "loss": 2.1808,
+      "step": 4979
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.02846975088968e-05,
+      "loss": 1.9382,
+      "step": 4980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0280743376828784e-05,
+      "loss": 1.8857,
+      "step": 4981
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.027678924476078e-05,
+      "loss": 1.8249,
+      "step": 4982
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0272835112692765e-05,
+      "loss": 1.998,
+      "step": 4983
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0268880980624754e-05,
+      "loss": 1.9693,
+      "step": 4984
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.026492684855674e-05,
+      "loss": 1.7142,
+      "step": 4985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0260972716488735e-05,
+      "loss": 1.9409,
+      "step": 4986
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.025701858442072e-05,
+      "loss": 2.0512,
+      "step": 4987
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.025306445235271e-05,
+      "loss": 1.7723,
+      "step": 4988
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.02491103202847e-05,
+      "loss": 2.0877,
+      "step": 4989
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.024515618821669e-05,
+      "loss": 2.0309,
+      "step": 4990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0241202056148676e-05,
+      "loss": 2.105,
+      "step": 4991
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0237247924080664e-05,
+      "loss": 2.0814,
+      "step": 4992
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0233293792012657e-05,
+      "loss": 1.8238,
+      "step": 4993
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.0229339659944645e-05,
+      "loss": 2.2079,
+      "step": 4994
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.022538552787663e-05,
+      "loss": 1.8607,
+      "step": 4995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.022143139580862e-05,
+      "loss": 1.8794,
+      "step": 4996
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0217477263740612e-05,
+      "loss": 1.9824,
+      "step": 4997
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.02135231316726e-05,
+      "loss": 1.9389,
+      "step": 4998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0209568999604586e-05,
+      "loss": 1.8998,
+      "step": 4999
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0205614867536575e-05,
+      "loss": 1.949,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0201660735468567e-05,
+      "loss": 1.872,
+      "step": 5001
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0197706603400556e-05,
+      "loss": 2.0044,
+      "step": 5002
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.019375247133254e-05,
+      "loss": 2.0103,
+      "step": 5003
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0189798339264537e-05,
+      "loss": 1.9452,
+      "step": 5004
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0185844207196523e-05,
+      "loss": 1.9433,
+      "step": 5005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.018189007512851e-05,
+      "loss": 2.0127,
+      "step": 5006
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0177935943060497e-05,
+      "loss": 2.1318,
+      "step": 5007
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0173981810992492e-05,
+      "loss": 2.1753,
+      "step": 5008
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0170027678924478e-05,
+      "loss": 1.9788,
+      "step": 5009
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0166073546856467e-05,
+      "loss": 2.0592,
+      "step": 5010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0162119414788452e-05,
+      "loss": 1.9355,
+      "step": 5011
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0158165282720448e-05,
+      "loss": 2.1026,
+      "step": 5012
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0154211150652433e-05,
+      "loss": 2.0857,
+      "step": 5013
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0150257018584422e-05,
+      "loss": 2.0468,
+      "step": 5014
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0146302886516407e-05,
+      "loss": 2.1486,
+      "step": 5015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0142348754448403e-05,
+      "loss": 1.8337,
+      "step": 5016
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.013839462238039e-05,
+      "loss": 1.9421,
+      "step": 5017
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0134440490312377e-05,
+      "loss": 2.2903,
+      "step": 5018
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0130486358244363e-05,
+      "loss": 1.8809,
+      "step": 5019
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0126532226176358e-05,
+      "loss": 2.0755,
+      "step": 5020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0122578094108344e-05,
+      "loss": 2.3614,
+      "step": 5021
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0118623962040333e-05,
+      "loss": 2.1207,
+      "step": 5022
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0114669829972325e-05,
+      "loss": 1.7691,
+      "step": 5023
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0110715697904314e-05,
+      "loss": 2.18,
+      "step": 5024
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.01067615658363e-05,
+      "loss": 1.8862,
+      "step": 5025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0102807433768288e-05,
+      "loss": 1.974,
+      "step": 5026
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.009885330170028e-05,
+      "loss": 1.996,
+      "step": 5027
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.009489916963227e-05,
+      "loss": 1.879,
+      "step": 5028
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0090945037564254e-05,
+      "loss": 1.9911,
+      "step": 5029
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0086990905496243e-05,
+      "loss": 1.8776,
+      "step": 5030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0083036773428235e-05,
+      "loss": 1.994,
+      "step": 5031
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0079082641360224e-05,
+      "loss": 1.8714,
+      "step": 5032
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.007512850929221e-05,
+      "loss": 1.9641,
+      "step": 5033
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.00711743772242e-05,
+      "loss": 1.9495,
+      "step": 5034
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.006722024515619e-05,
+      "loss": 2.1041,
+      "step": 5035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.006326611308818e-05,
+      "loss": 1.9473,
+      "step": 5036
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0059311981020165e-05,
+      "loss": 2.1907,
+      "step": 5037
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0055357848952154e-05,
+      "loss": 1.8579,
+      "step": 5038
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0051403716884146e-05,
+      "loss": 1.9947,
+      "step": 5039
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0047449584816135e-05,
+      "loss": 2.0725,
+      "step": 5040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.004349545274812e-05,
+      "loss": 1.9089,
+      "step": 5041
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0039541320680116e-05,
+      "loss": 2.0242,
+      "step": 5042
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.00355871886121e-05,
+      "loss": 1.9693,
+      "step": 5043
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.003163305654409e-05,
+      "loss": 1.7881,
+      "step": 5044
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0027678924476076e-05,
+      "loss": 1.8832,
+      "step": 5045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.002372479240807e-05,
+      "loss": 1.8725,
+      "step": 5046
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0019770660340057e-05,
+      "loss": 2.0934,
+      "step": 5047
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0015816528272045e-05,
+      "loss": 2.0616,
+      "step": 5048
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.001186239620403e-05,
+      "loss": 1.9247,
+      "step": 5049
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0007908264136026e-05,
+      "loss": 1.9162,
+      "step": 5050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.0003954132068012e-05,
+      "loss": 1.7976,
+      "step": 5051
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3e-05,
+      "loss": 1.9831,
+      "step": 5052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9996045867931986e-05,
+      "loss": 2.0816,
+      "step": 5053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9992091735863982e-05,
+      "loss": 1.8335,
+      "step": 5054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9988137603795967e-05,
+      "loss": 2.156,
+      "step": 5055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9984183471727956e-05,
+      "loss": 1.7645,
+      "step": 5056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9980229339659948e-05,
+      "loss": 1.9018,
+      "step": 5057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9976275207591937e-05,
+      "loss": 1.7592,
+      "step": 5058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9972321075523923e-05,
+      "loss": 2.0751,
+      "step": 5059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.996836694345591e-05,
+      "loss": 1.9586,
+      "step": 5060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9964412811387904e-05,
+      "loss": 1.9169,
+      "step": 5061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9960458679319892e-05,
+      "loss": 1.7486,
+      "step": 5062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9956504547251878e-05,
+      "loss": 2.1661,
+      "step": 5063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9952550415183867e-05,
+      "loss": 2.1383,
+      "step": 5064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.994859628311586e-05,
+      "loss": 1.9064,
+      "step": 5065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9944642151047848e-05,
+      "loss": 1.8826,
+      "step": 5066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9940688018979833e-05,
+      "loss": 2.0966,
+      "step": 5067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9936733886911822e-05,
+      "loss": 1.9243,
+      "step": 5068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9932779754843814e-05,
+      "loss": 1.7916,
+      "step": 5069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9928825622775803e-05,
+      "loss": 2.2197,
+      "step": 5070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.992487149070779e-05,
+      "loss": 2.2128,
+      "step": 5071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9920917358639777e-05,
+      "loss": 2.063,
+      "step": 5072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.991696322657177e-05,
+      "loss": 2.0947,
+      "step": 5073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.991300909450376e-05,
+      "loss": 1.915,
+      "step": 5074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9909054962435744e-05,
+      "loss": 1.9127,
+      "step": 5075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.990510083036774e-05,
+      "loss": 1.9821,
+      "step": 5076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9901146698299725e-05,
+      "loss": 1.9609,
+      "step": 5077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9897192566231714e-05,
+      "loss": 1.8525,
+      "step": 5078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.98932384341637e-05,
+      "loss": 2.0907,
+      "step": 5079
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9889284302095695e-05,
+      "loss": 1.8805,
+      "step": 5080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.988533017002768e-05,
+      "loss": 2.0925,
+      "step": 5081
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.988137603795967e-05,
+      "loss": 2.2128,
+      "step": 5082
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9877421905891654e-05,
+      "loss": 2.0009,
+      "step": 5083
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.987346777382365e-05,
+      "loss": 1.8222,
+      "step": 5084
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9869513641755635e-05,
+      "loss": 2.3381,
+      "step": 5085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9865559509687624e-05,
+      "loss": 1.9506,
+      "step": 5086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9861605377619613e-05,
+      "loss": 2.115,
+      "step": 5087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9857651245551605e-05,
+      "loss": 1.8886,
+      "step": 5088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.985369711348359e-05,
+      "loss": 2.3209,
+      "step": 5089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.984974298141558e-05,
+      "loss": 2.1335,
+      "step": 5090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9845788849347572e-05,
+      "loss": 1.9559,
+      "step": 5091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.984183471727956e-05,
+      "loss": 2.0011,
+      "step": 5092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9837880585211546e-05,
+      "loss": 2.1087,
+      "step": 5093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9833926453143535e-05,
+      "loss": 1.8475,
+      "step": 5094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9829972321075527e-05,
+      "loss": 2.1458,
+      "step": 5095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9826018189007516e-05,
+      "loss": 1.8549,
+      "step": 5096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.98220640569395e-05,
+      "loss": 1.8608,
+      "step": 5097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.981810992487149e-05,
+      "loss": 2.0832,
+      "step": 5098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9814155792803482e-05,
+      "loss": 1.897,
+      "step": 5099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.981020166073547e-05,
+      "loss": 2.3173,
+      "step": 5100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9806247528667457e-05,
+      "loss": 1.9626,
+      "step": 5101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9802293396599445e-05,
+      "loss": 1.8139,
+      "step": 5102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9798339264531438e-05,
+      "loss": 2.0141,
+      "step": 5103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9794385132463427e-05,
+      "loss": 1.6852,
+      "step": 5104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9790431000395412e-05,
+      "loss": 1.9891,
+      "step": 5105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.97864768683274e-05,
+      "loss": 1.9127,
+      "step": 5106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9782522736259393e-05,
+      "loss": 1.9321,
+      "step": 5107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9778568604191382e-05,
+      "loss": 2.1083,
+      "step": 5108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9774614472123367e-05,
+      "loss": 2.2366,
+      "step": 5109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9770660340055363e-05,
+      "loss": 1.9547,
+      "step": 5110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.976670620798735e-05,
+      "loss": 1.7884,
+      "step": 5111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9762752075919337e-05,
+      "loss": 1.9893,
+      "step": 5112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9758797943851323e-05,
+      "loss": 2.0517,
+      "step": 5113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9754843811783318e-05,
+      "loss": 2.1724,
+      "step": 5114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9750889679715304e-05,
+      "loss": 2.1979,
+      "step": 5115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9746935547647292e-05,
+      "loss": 1.9119,
+      "step": 5116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.974298141557928e-05,
+      "loss": 2.0259,
+      "step": 5117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9739027283511274e-05,
+      "loss": 2.1175,
+      "step": 5118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.973507315144326e-05,
+      "loss": 1.9454,
+      "step": 5119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9731119019375248e-05,
+      "loss": 2.1881,
+      "step": 5120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9727164887307237e-05,
+      "loss": 2.0363,
+      "step": 5121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.972321075523923e-05,
+      "loss": 1.8654,
+      "step": 5122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9719256623171214e-05,
+      "loss": 2.1407,
+      "step": 5123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9715302491103203e-05,
+      "loss": 2.0674,
+      "step": 5124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9711348359035192e-05,
+      "loss": 2.0045,
+      "step": 5125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9707394226967184e-05,
+      "loss": 2.0853,
+      "step": 5126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.970344009489917e-05,
+      "loss": 1.9637,
+      "step": 5127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.969948596283116e-05,
+      "loss": 2.1006,
+      "step": 5128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.969553183076315e-05,
+      "loss": 2.2031,
+      "step": 5129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.969157769869514e-05,
+      "loss": 1.8336,
+      "step": 5130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9687623566627125e-05,
+      "loss": 1.8711,
+      "step": 5131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9683669434559114e-05,
+      "loss": 2.3045,
+      "step": 5132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9679715302491106e-05,
+      "loss": 2.1472,
+      "step": 5133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9675761170423095e-05,
+      "loss": 2.0498,
+      "step": 5134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.967180703835508e-05,
+      "loss": 1.9438,
+      "step": 5135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.966785290628707e-05,
+      "loss": 1.884,
+      "step": 5136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.966389877421906e-05,
+      "loss": 2.052,
+      "step": 5137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.965994464215105e-05,
+      "loss": 1.8991,
+      "step": 5138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9655990510083035e-05,
+      "loss": 2.0386,
+      "step": 5139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9652036378015024e-05,
+      "loss": 1.9124,
+      "step": 5140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9648082245947017e-05,
+      "loss": 1.9736,
+      "step": 5141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9644128113879005e-05,
+      "loss": 2.111,
+      "step": 5142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.964017398181099e-05,
+      "loss": 2.1416,
+      "step": 5143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9636219849742986e-05,
+      "loss": 2.1435,
+      "step": 5144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9632265717674972e-05,
+      "loss": 2.0478,
+      "step": 5145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.962831158560696e-05,
+      "loss": 2.0259,
+      "step": 5146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.962435745353895e-05,
+      "loss": 2.004,
+      "step": 5147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9620403321470942e-05,
+      "loss": 2.0716,
+      "step": 5148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9616449189402927e-05,
+      "loss": 2.174,
+      "step": 5149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9612495057334916e-05,
+      "loss": 2.0826,
+      "step": 5150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9608540925266905e-05,
+      "loss": 1.9956,
+      "step": 5151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9604586793198897e-05,
+      "loss": 2.0515,
+      "step": 5152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9600632661130882e-05,
+      "loss": 2.1802,
+      "step": 5153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.959667852906287e-05,
+      "loss": 1.8386,
+      "step": 5154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.959272439699486e-05,
+      "loss": 2.1035,
+      "step": 5155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9588770264926852e-05,
+      "loss": 1.7437,
+      "step": 5156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9584816132858838e-05,
+      "loss": 1.9773,
+      "step": 5157
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9580862000790827e-05,
+      "loss": 2.2154,
+      "step": 5158
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9576907868722815e-05,
+      "loss": 1.8586,
+      "step": 5159
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9572953736654808e-05,
+      "loss": 2.18,
+      "step": 5160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9568999604586793e-05,
+      "loss": 1.7578,
+      "step": 5161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9565045472518782e-05,
+      "loss": 1.7206,
+      "step": 5162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9561091340450774e-05,
+      "loss": 1.9883,
+      "step": 5163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9557137208382763e-05,
+      "loss": 2.09,
+      "step": 5164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.955318307631475e-05,
+      "loss": 2.2537,
+      "step": 5165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9549228944246737e-05,
+      "loss": 1.8439,
+      "step": 5166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.954527481217873e-05,
+      "loss": 2.1145,
+      "step": 5167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9541320680110718e-05,
+      "loss": 1.9653,
+      "step": 5168
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9537366548042704e-05,
+      "loss": 1.8322,
+      "step": 5169
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9533412415974693e-05,
+      "loss": 2.0614,
+      "step": 5170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9529458283906685e-05,
+      "loss": 1.947,
+      "step": 5171
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9525504151838674e-05,
+      "loss": 1.8725,
+      "step": 5172
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9521550019770662e-05,
+      "loss": 1.903,
+      "step": 5173
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9517595887702648e-05,
+      "loss": 2.1351,
+      "step": 5174
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.951364175563464e-05,
+      "loss": 2.1297,
+      "step": 5175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.950968762356663e-05,
+      "loss": 1.9637,
+      "step": 5176
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9505733491498618e-05,
+      "loss": 1.6858,
+      "step": 5177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.950177935943061e-05,
+      "loss": 2.205,
+      "step": 5178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9497825227362595e-05,
+      "loss": 2.1054,
+      "step": 5179
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9493871095294584e-05,
+      "loss": 2.3262,
+      "step": 5180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9489916963226573e-05,
+      "loss": 2.0832,
+      "step": 5181
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9485962831158565e-05,
+      "loss": 2.0227,
+      "step": 5182
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.948200869909055e-05,
+      "loss": 1.8263,
+      "step": 5183
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.947805456702254e-05,
+      "loss": 2.0351,
+      "step": 5184
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9474100434954528e-05,
+      "loss": 2.1885,
+      "step": 5185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.947014630288652e-05,
+      "loss": 1.9928,
+      "step": 5186
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9466192170818506e-05,
+      "loss": 1.9326,
+      "step": 5187
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9462238038750495e-05,
+      "loss": 2.0264,
+      "step": 5188
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9458283906682484e-05,
+      "loss": 1.9177,
+      "step": 5189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9454329774614476e-05,
+      "loss": 1.9655,
+      "step": 5190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.945037564254646e-05,
+      "loss": 1.8335,
+      "step": 5191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.944642151047845e-05,
+      "loss": 2.006,
+      "step": 5192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.944246737841044e-05,
+      "loss": 2.1225,
+      "step": 5193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.943851324634243e-05,
+      "loss": 2.117,
+      "step": 5194
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9434559114274417e-05,
+      "loss": 2.1289,
+      "step": 5195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9430604982206405e-05,
+      "loss": 2.1775,
+      "step": 5196
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9426650850138398e-05,
+      "loss": 1.7685,
+      "step": 5197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9422696718070386e-05,
+      "loss": 1.869,
+      "step": 5198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9418742586002372e-05,
+      "loss": 1.8214,
+      "step": 5199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.941478845393436e-05,
+      "loss": 1.8505,
+      "step": 5200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9410834321866353e-05,
+      "loss": 2.1814,
+      "step": 5201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9406880189798342e-05,
+      "loss": 1.9806,
+      "step": 5202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.940292605773033e-05,
+      "loss": 1.8709,
+      "step": 5203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9398971925662316e-05,
+      "loss": 1.9897,
+      "step": 5204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9395017793594308e-05,
+      "loss": 2.0646,
+      "step": 5205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9391063661526297e-05,
+      "loss": 1.8813,
+      "step": 5206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9387109529458286e-05,
+      "loss": 2.1691,
+      "step": 5207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.938315539739027e-05,
+      "loss": 2.0275,
+      "step": 5208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9379201265322264e-05,
+      "loss": 2.0643,
+      "step": 5209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9375247133254252e-05,
+      "loss": 1.9371,
+      "step": 5210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.937129300118624e-05,
+      "loss": 1.885,
+      "step": 5211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9367338869118227e-05,
+      "loss": 2.0112,
+      "step": 5212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.936338473705022e-05,
+      "loss": 1.911,
+      "step": 5213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9359430604982208e-05,
+      "loss": 1.7758,
+      "step": 5214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9355476472914196e-05,
+      "loss": 1.9526,
+      "step": 5215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.935152234084619e-05,
+      "loss": 2.2421,
+      "step": 5216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9347568208778174e-05,
+      "loss": 1.997,
+      "step": 5217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9343614076710163e-05,
+      "loss": 2.2591,
+      "step": 5218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9339659944642152e-05,
+      "loss": 2.0654,
+      "step": 5219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9335705812574144e-05,
+      "loss": 2.2344,
+      "step": 5220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.933175168050613e-05,
+      "loss": 2.0332,
+      "step": 5221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9327797548438118e-05,
+      "loss": 1.9733,
+      "step": 5222
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9323843416370107e-05,
+      "loss": 2.2015,
+      "step": 5223
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.93198892843021e-05,
+      "loss": 2.1379,
+      "step": 5224
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9315935152234085e-05,
+      "loss": 1.8927,
+      "step": 5225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9311981020166074e-05,
+      "loss": 2.06,
+      "step": 5226
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9308026888098062e-05,
+      "loss": 1.869,
+      "step": 5227
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9304072756030055e-05,
+      "loss": 1.9437,
+      "step": 5228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9300118623962043e-05,
+      "loss": 1.9124,
+      "step": 5229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.929616449189403e-05,
+      "loss": 2.1496,
+      "step": 5230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.929221035982602e-05,
+      "loss": 1.8261,
+      "step": 5231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.928825622775801e-05,
+      "loss": 2.0569,
+      "step": 5232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.928430209569e-05,
+      "loss": 1.9165,
+      "step": 5233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9280347963621984e-05,
+      "loss": 1.8949,
+      "step": 5234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9276393831553976e-05,
+      "loss": 1.8715,
+      "step": 5235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9272439699485965e-05,
+      "loss": 1.7521,
+      "step": 5236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9268485567417954e-05,
+      "loss": 2.1726,
+      "step": 5237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.926453143534994e-05,
+      "loss": 2.0266,
+      "step": 5238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9260577303281932e-05,
+      "loss": 2.0771,
+      "step": 5239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.925662317121392e-05,
+      "loss": 1.978,
+      "step": 5240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.925266903914591e-05,
+      "loss": 1.8455,
+      "step": 5241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9248714907077895e-05,
+      "loss": 2.0503,
+      "step": 5242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9244760775009887e-05,
+      "loss": 2.0115,
+      "step": 5243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9240806642941876e-05,
+      "loss": 2.0403,
+      "step": 5244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9236852510873865e-05,
+      "loss": 2.2543,
+      "step": 5245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.923289837880585e-05,
+      "loss": 1.931,
+      "step": 5246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9228944246737842e-05,
+      "loss": 2.2168,
+      "step": 5247
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.922499011466983e-05,
+      "loss": 1.9395,
+      "step": 5248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.922103598260182e-05,
+      "loss": 2.0159,
+      "step": 5249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9217081850533812e-05,
+      "loss": 2.0263,
+      "step": 5250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9213127718465798e-05,
+      "loss": 2.1551,
+      "step": 5251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9209173586397786e-05,
+      "loss": 1.8752,
+      "step": 5252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9205219454329775e-05,
+      "loss": 1.9824,
+      "step": 5253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9201265322261768e-05,
+      "loss": 1.7477,
+      "step": 5254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9197311190193753e-05,
+      "loss": 1.8026,
+      "step": 5255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9193357058125742e-05,
+      "loss": 1.9787,
+      "step": 5256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.918940292605773e-05,
+      "loss": 2.0426,
+      "step": 5257
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9185448793989723e-05,
+      "loss": 1.8909,
+      "step": 5258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.918149466192171e-05,
+      "loss": 1.8872,
+      "step": 5259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9177540529853697e-05,
+      "loss": 1.7181,
+      "step": 5260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9173586397785686e-05,
+      "loss": 1.8888,
+      "step": 5261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9169632265717678e-05,
+      "loss": 2.0263,
+      "step": 5262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9165678133649667e-05,
+      "loss": 1.7436,
+      "step": 5263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9161724001581652e-05,
+      "loss": 2.1861,
+      "step": 5264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9157769869513645e-05,
+      "loss": 1.9449,
+      "step": 5265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9153815737445633e-05,
+      "loss": 1.857,
+      "step": 5266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9149861605377622e-05,
+      "loss": 1.8498,
+      "step": 5267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9145907473309608e-05,
+      "loss": 2.2984,
+      "step": 5268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.91419533412416e-05,
+      "loss": 2.2305,
+      "step": 5269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.913799920917359e-05,
+      "loss": 2.037,
+      "step": 5270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9134045077105578e-05,
+      "loss": 1.7803,
+      "step": 5271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9130090945037563e-05,
+      "loss": 1.998,
+      "step": 5272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9126136812969555e-05,
+      "loss": 2.013,
+      "step": 5273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9122182680901544e-05,
+      "loss": 2.1305,
+      "step": 5274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9118228548833533e-05,
+      "loss": 1.73,
+      "step": 5275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.911427441676552e-05,
+      "loss": 1.9122,
+      "step": 5276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.911032028469751e-05,
+      "loss": 2.0145,
+      "step": 5277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.91063661526295e-05,
+      "loss": 2.0206,
+      "step": 5278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9102412020561488e-05,
+      "loss": 1.7323,
+      "step": 5279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9098457888493474e-05,
+      "loss": 2.0383,
+      "step": 5280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9094503756425466e-05,
+      "loss": 1.9521,
+      "step": 5281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9090549624357455e-05,
+      "loss": 2.0462,
+      "step": 5282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9086595492289444e-05,
+      "loss": 2.2423,
+      "step": 5283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9082641360221436e-05,
+      "loss": 2.1723,
+      "step": 5284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.907868722815342e-05,
+      "loss": 2.2664,
+      "step": 5285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.907473309608541e-05,
+      "loss": 2.0255,
+      "step": 5286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.90707789640174e-05,
+      "loss": 1.9929,
+      "step": 5287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.906682483194939e-05,
+      "loss": 1.8832,
+      "step": 5288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.906287069988138e-05,
+      "loss": 2.3516,
+      "step": 5289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9058916567813365e-05,
+      "loss": 2.1636,
+      "step": 5290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9054962435745354e-05,
+      "loss": 2.0558,
+      "step": 5291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9051008303677346e-05,
+      "loss": 2.0521,
+      "step": 5292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9047054171609335e-05,
+      "loss": 1.8066,
+      "step": 5293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.904310003954132e-05,
+      "loss": 2.2767,
+      "step": 5294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.903914590747331e-05,
+      "loss": 2.0879,
+      "step": 5295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.90351917754053e-05,
+      "loss": 2.1181,
+      "step": 5296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.903123764333729e-05,
+      "loss": 2.1845,
+      "step": 5297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9027283511269276e-05,
+      "loss": 2.0188,
+      "step": 5298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9023329379201268e-05,
+      "loss": 2.1098,
+      "step": 5299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9019375247133257e-05,
+      "loss": 1.9414,
+      "step": 5300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9015421115065246e-05,
+      "loss": 1.8343,
+      "step": 5301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.901146698299723e-05,
+      "loss": 1.8271,
+      "step": 5302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9007512850929223e-05,
+      "loss": 2.1471,
+      "step": 5303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9003558718861212e-05,
+      "loss": 1.9507,
+      "step": 5304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.89996045867932e-05,
+      "loss": 2.2322,
+      "step": 5305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8995650454725187e-05,
+      "loss": 2.0386,
+      "step": 5306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.899169632265718e-05,
+      "loss": 2.2654,
+      "step": 5307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8987742190589168e-05,
+      "loss": 2.0157,
+      "step": 5308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8983788058521156e-05,
+      "loss": 1.9207,
+      "step": 5309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8979833926453142e-05,
+      "loss": 1.881,
+      "step": 5310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8975879794385134e-05,
+      "loss": 1.8754,
+      "step": 5311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8971925662317123e-05,
+      "loss": 1.9979,
+      "step": 5312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8967971530249112e-05,
+      "loss": 1.9824,
+      "step": 5313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8964017398181097e-05,
+      "loss": 2.1819,
+      "step": 5314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8960063266113093e-05,
+      "loss": 2.0729,
+      "step": 5315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8956109134045078e-05,
+      "loss": 2.1361,
+      "step": 5316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8952155001977067e-05,
+      "loss": 1.8778,
+      "step": 5317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.894820086990906e-05,
+      "loss": 2.1762,
+      "step": 5318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8944246737841048e-05,
+      "loss": 2.0437,
+      "step": 5319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8940292605773033e-05,
+      "loss": 1.9345,
+      "step": 5320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8936338473705022e-05,
+      "loss": 2.1981,
+      "step": 5321
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8932384341637015e-05,
+      "loss": 1.9746,
+      "step": 5322
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8928430209569003e-05,
+      "loss": 2.1029,
+      "step": 5323
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.892447607750099e-05,
+      "loss": 2.066,
+      "step": 5324
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8920521945432978e-05,
+      "loss": 2.1167,
+      "step": 5325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.891656781336497e-05,
+      "loss": 1.7819,
+      "step": 5326
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.891261368129696e-05,
+      "loss": 1.9994,
+      "step": 5327
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8908659549228944e-05,
+      "loss": 1.889,
+      "step": 5328
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8904705417160933e-05,
+      "loss": 1.9272,
+      "step": 5329
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8900751285092925e-05,
+      "loss": 1.8625,
+      "step": 5330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8896797153024914e-05,
+      "loss": 2.0017,
+      "step": 5331
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.88928430209569e-05,
+      "loss": 1.7081,
+      "step": 5332
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8888888888888888e-05,
+      "loss": 1.8525,
+      "step": 5333
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.888493475682088e-05,
+      "loss": 1.8418,
+      "step": 5334
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.888098062475287e-05,
+      "loss": 1.8551,
+      "step": 5335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8877026492684855e-05,
+      "loss": 2.047,
+      "step": 5336
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8873072360616847e-05,
+      "loss": 2.0034,
+      "step": 5337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8869118228548836e-05,
+      "loss": 1.9132,
+      "step": 5338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8865164096480825e-05,
+      "loss": 1.9807,
+      "step": 5339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.886120996441281e-05,
+      "loss": 1.9377,
+      "step": 5340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8857255832344802e-05,
+      "loss": 1.892,
+      "step": 5341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.885330170027679e-05,
+      "loss": 1.8712,
+      "step": 5342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.884934756820878e-05,
+      "loss": 2.0722,
+      "step": 5343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8845393436140765e-05,
+      "loss": 1.7207,
+      "step": 5344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.884143930407276e-05,
+      "loss": 1.9601,
+      "step": 5345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8837485172004746e-05,
+      "loss": 2.0061,
+      "step": 5346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8833531039936735e-05,
+      "loss": 1.9396,
+      "step": 5347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.882957690786872e-05,
+      "loss": 2.3028,
+      "step": 5348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8825622775800716e-05,
+      "loss": 1.8899,
+      "step": 5349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.88216686437327e-05,
+      "loss": 1.9744,
+      "step": 5350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.881771451166469e-05,
+      "loss": 1.8413,
+      "step": 5351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8813760379596683e-05,
+      "loss": 1.8563,
+      "step": 5352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.880980624752867e-05,
+      "loss": 1.9759,
+      "step": 5353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8805852115460657e-05,
+      "loss": 1.9623,
+      "step": 5354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8801897983392646e-05,
+      "loss": 1.9205,
+      "step": 5355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8797943851324638e-05,
+      "loss": 1.8404,
+      "step": 5356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8793989719256627e-05,
+      "loss": 1.9518,
+      "step": 5357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8790035587188612e-05,
+      "loss": 1.8634,
+      "step": 5358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.87860814551206e-05,
+      "loss": 1.985,
+      "step": 5359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8782127323052593e-05,
+      "loss": 1.8105,
+      "step": 5360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8778173190984582e-05,
+      "loss": 1.986,
+      "step": 5361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8774219058916568e-05,
+      "loss": 2.1416,
+      "step": 5362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8770264926848556e-05,
+      "loss": 2.1417,
+      "step": 5363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.876631079478055e-05,
+      "loss": 2.0914,
+      "step": 5364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8762356662712537e-05,
+      "loss": 2.0017,
+      "step": 5365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8758402530644523e-05,
+      "loss": 1.954,
+      "step": 5366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8754448398576512e-05,
+      "loss": 1.8379,
+      "step": 5367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8750494266508504e-05,
+      "loss": 1.8126,
+      "step": 5368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8746540134440493e-05,
+      "loss": 2.0922,
+      "step": 5369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8742586002372478e-05,
+      "loss": 2.1122,
+      "step": 5370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.873863187030447e-05,
+      "loss": 1.9295,
+      "step": 5371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.873467773823646e-05,
+      "loss": 2.2045,
+      "step": 5372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8730723606168448e-05,
+      "loss": 2.09,
+      "step": 5373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8726769474100434e-05,
+      "loss": 2.2272,
+      "step": 5374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.872281534203243e-05,
+      "loss": 1.9019,
+      "step": 5375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8718861209964415e-05,
+      "loss": 2.0221,
+      "step": 5376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8714907077896403e-05,
+      "loss": 2.1611,
+      "step": 5377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.871095294582839e-05,
+      "loss": 2.2269,
+      "step": 5378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8706998813760384e-05,
+      "loss": 2.2362,
+      "step": 5379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.870304468169237e-05,
+      "loss": 2.3016,
+      "step": 5380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.869909054962436e-05,
+      "loss": 1.8109,
+      "step": 5381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8695136417556344e-05,
+      "loss": 2.2913,
+      "step": 5382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.869118228548834e-05,
+      "loss": 2.4047,
+      "step": 5383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8687228153420325e-05,
+      "loss": 1.7052,
+      "step": 5384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8683274021352314e-05,
+      "loss": 1.6909,
+      "step": 5385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8679319889284306e-05,
+      "loss": 2.0821,
+      "step": 5386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8675365757216295e-05,
+      "loss": 1.8251,
+      "step": 5387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.867141162514828e-05,
+      "loss": 1.9121,
+      "step": 5388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.866745749308027e-05,
+      "loss": 1.8896,
+      "step": 5389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.866350336101226e-05,
+      "loss": 1.8852,
+      "step": 5390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.865954922894425e-05,
+      "loss": 2.0996,
+      "step": 5391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8655595096876236e-05,
+      "loss": 1.9725,
+      "step": 5392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8651640964808225e-05,
+      "loss": 2.21,
+      "step": 5393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8647686832740217e-05,
+      "loss": 1.9221,
+      "step": 5394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8643732700672206e-05,
+      "loss": 2.0988,
+      "step": 5395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.863977856860419e-05,
+      "loss": 2.3169,
+      "step": 5396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.863582443653618e-05,
+      "loss": 1.9213,
+      "step": 5397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8631870304468172e-05,
+      "loss": 2.1241,
+      "step": 5398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.862791617240016e-05,
+      "loss": 2.0017,
+      "step": 5399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8623962040332146e-05,
+      "loss": 1.6955,
+      "step": 5400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8620007908264135e-05,
+      "loss": 1.8787,
+      "step": 5401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8616053776196127e-05,
+      "loss": 1.8252,
+      "step": 5402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8612099644128116e-05,
+      "loss": 2.1739,
+      "step": 5403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8608145512060102e-05,
+      "loss": 1.8467,
+      "step": 5404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8604191379992097e-05,
+      "loss": 1.9435,
+      "step": 5405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8600237247924083e-05,
+      "loss": 2.0966,
+      "step": 5406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.859628311585607e-05,
+      "loss": 1.8352,
+      "step": 5407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8592328983788057e-05,
+      "loss": 1.946,
+      "step": 5408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8588374851720053e-05,
+      "loss": 2.0241,
+      "step": 5409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8584420719652038e-05,
+      "loss": 2.0038,
+      "step": 5410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8580466587584027e-05,
+      "loss": 2.0208,
+      "step": 5411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8576512455516012e-05,
+      "loss": 2.0493,
+      "step": 5412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8572558323448008e-05,
+      "loss": 1.9153,
+      "step": 5413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8568604191379993e-05,
+      "loss": 1.8276,
+      "step": 5414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8564650059311982e-05,
+      "loss": 2.0446,
+      "step": 5415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8560695927243968e-05,
+      "loss": 2.0644,
+      "step": 5416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8556741795175963e-05,
+      "loss": 1.8367,
+      "step": 5417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.855278766310795e-05,
+      "loss": 1.8402,
+      "step": 5418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8548833531039938e-05,
+      "loss": 2.0278,
+      "step": 5419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8544879398971923e-05,
+      "loss": 1.7659,
+      "step": 5420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.854092526690392e-05,
+      "loss": 1.9937,
+      "step": 5421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8536971134835904e-05,
+      "loss": 2.1649,
+      "step": 5422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8533017002767893e-05,
+      "loss": 2.0125,
+      "step": 5423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8529062870699885e-05,
+      "loss": 1.9228,
+      "step": 5424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8525108738631874e-05,
+      "loss": 2.1534,
+      "step": 5425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.852115460656386e-05,
+      "loss": 1.8997,
+      "step": 5426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8517200474495848e-05,
+      "loss": 2.072,
+      "step": 5427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.851324634242784e-05,
+      "loss": 1.9941,
+      "step": 5428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.850929221035983e-05,
+      "loss": 1.929,
+      "step": 5429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8505338078291815e-05,
+      "loss": 1.9074,
+      "step": 5430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8501383946223803e-05,
+      "loss": 2.1038,
+      "step": 5431
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8497429814155796e-05,
+      "loss": 1.9237,
+      "step": 5432
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8493475682087784e-05,
+      "loss": 2.1192,
+      "step": 5433
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.848952155001977e-05,
+      "loss": 2.0945,
+      "step": 5434
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.848556741795176e-05,
+      "loss": 2.116,
+      "step": 5435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.848161328588375e-05,
+      "loss": 2.1392,
+      "step": 5436
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.847765915381574e-05,
+      "loss": 1.9068,
+      "step": 5437
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8473705021747725e-05,
+      "loss": 1.9913,
+      "step": 5438
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.846975088967972e-05,
+      "loss": 2.1959,
+      "step": 5439
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8465796757611706e-05,
+      "loss": 2.122,
+      "step": 5440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8461842625543695e-05,
+      "loss": 2.1244,
+      "step": 5441
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.845788849347568e-05,
+      "loss": 1.9335,
+      "step": 5442
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8453934361407676e-05,
+      "loss": 2.1019,
+      "step": 5443
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.844998022933966e-05,
+      "loss": 1.9099,
+      "step": 5444
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.844602609727165e-05,
+      "loss": 2.0409,
+      "step": 5445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8442071965203636e-05,
+      "loss": 2.0361,
+      "step": 5446
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.843811783313563e-05,
+      "loss": 2.1329,
+      "step": 5447
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8434163701067617e-05,
+      "loss": 2.1143,
+      "step": 5448
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8430209568999606e-05,
+      "loss": 2.1324,
+      "step": 5449
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.842625543693159e-05,
+      "loss": 1.7138,
+      "step": 5450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8422301304863587e-05,
+      "loss": 1.906,
+      "step": 5451
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8418347172795572e-05,
+      "loss": 2.2179,
+      "step": 5452
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.841439304072756e-05,
+      "loss": 1.9242,
+      "step": 5453
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8410438908659546e-05,
+      "loss": 2.0766,
+      "step": 5454
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8406484776591542e-05,
+      "loss": 1.9859,
+      "step": 5455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8402530644523528e-05,
+      "loss": 2.0058,
+      "step": 5456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8398576512455516e-05,
+      "loss": 2.3795,
+      "step": 5457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.839462238038751e-05,
+      "loss": 2.0303,
+      "step": 5458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8390668248319497e-05,
+      "loss": 2.0419,
+      "step": 5459
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8386714116251483e-05,
+      "loss": 2.2501,
+      "step": 5460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.838275998418347e-05,
+      "loss": 1.9805,
+      "step": 5461
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8378805852115464e-05,
+      "loss": 2.0389,
+      "step": 5462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8374851720047453e-05,
+      "loss": 1.97,
+      "step": 5463
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8370897587979438e-05,
+      "loss": 2.1109,
+      "step": 5464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8366943455911427e-05,
+      "loss": 1.8257,
+      "step": 5465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.836298932384342e-05,
+      "loss": 1.7881,
+      "step": 5466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8359035191775408e-05,
+      "loss": 1.9684,
+      "step": 5467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8355081059707393e-05,
+      "loss": 1.9325,
+      "step": 5468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8351126927639382e-05,
+      "loss": 1.8699,
+      "step": 5469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8347172795571374e-05,
+      "loss": 1.9923,
+      "step": 5470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8343218663503363e-05,
+      "loss": 1.8203,
+      "step": 5471
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.833926453143535e-05,
+      "loss": 2.1332,
+      "step": 5472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8335310399367344e-05,
+      "loss": 2.0237,
+      "step": 5473
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.833135626729933e-05,
+      "loss": 1.9588,
+      "step": 5474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.832740213523132e-05,
+      "loss": 1.9808,
+      "step": 5475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8323448003163304e-05,
+      "loss": 2.0617,
+      "step": 5476
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.83194938710953e-05,
+      "loss": 2.1607,
+      "step": 5477
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8315539739027285e-05,
+      "loss": 1.9256,
+      "step": 5478
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8311585606959274e-05,
+      "loss": 1.9566,
+      "step": 5479
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.830763147489126e-05,
+      "loss": 2.0087,
+      "step": 5480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8303677342823255e-05,
+      "loss": 1.9791,
+      "step": 5481
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.829972321075524e-05,
+      "loss": 1.9799,
+      "step": 5482
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.829576907868723e-05,
+      "loss": 2.0051,
+      "step": 5483
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8291814946619215e-05,
+      "loss": 1.9568,
+      "step": 5484
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.828786081455121e-05,
+      "loss": 2.3305,
+      "step": 5485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8283906682483196e-05,
+      "loss": 1.9726,
+      "step": 5486
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8279952550415185e-05,
+      "loss": 2.0733,
+      "step": 5487
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.827599841834717e-05,
+      "loss": 1.8063,
+      "step": 5488
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8272044286279166e-05,
+      "loss": 2.1236,
+      "step": 5489
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.826809015421115e-05,
+      "loss": 1.8435,
+      "step": 5490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.826413602214314e-05,
+      "loss": 1.9095,
+      "step": 5491
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8260181890075132e-05,
+      "loss": 1.9554,
+      "step": 5492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.825622775800712e-05,
+      "loss": 1.9545,
+      "step": 5493
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8252273625939106e-05,
+      "loss": 2.0402,
+      "step": 5494
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8248319493871095e-05,
+      "loss": 1.7991,
+      "step": 5495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8244365361803087e-05,
+      "loss": 2.0507,
+      "step": 5496
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8240411229735076e-05,
+      "loss": 2.1322,
+      "step": 5497
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.823645709766706e-05,
+      "loss": 2.0763,
+      "step": 5498
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.823250296559905e-05,
+      "loss": 1.9561,
+      "step": 5499
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8228548833531043e-05,
+      "loss": 2.0832,
+      "step": 5500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.822459470146303e-05,
+      "loss": 2.1439,
+      "step": 5501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8220640569395017e-05,
+      "loss": 1.7772,
+      "step": 5502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8216686437327006e-05,
+      "loss": 2.0466,
+      "step": 5503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8212732305258998e-05,
+      "loss": 2.0595,
+      "step": 5504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8208778173190987e-05,
+      "loss": 1.8846,
+      "step": 5505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8204824041122972e-05,
+      "loss": 2.2003,
+      "step": 5506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.820086990905496e-05,
+      "loss": 1.9438,
+      "step": 5507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8196915776986953e-05,
+      "loss": 1.9862,
+      "step": 5508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8192961644918942e-05,
+      "loss": 2.034,
+      "step": 5509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8189007512850928e-05,
+      "loss": 2.1058,
+      "step": 5510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8185053380782923e-05,
+      "loss": 2.254,
+      "step": 5511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.818109924871491e-05,
+      "loss": 1.9501,
+      "step": 5512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8177145116646897e-05,
+      "loss": 2.1251,
+      "step": 5513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8173190984578883e-05,
+      "loss": 2.1268,
+      "step": 5514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.816923685251088e-05,
+      "loss": 1.8645,
+      "step": 5515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8165282720442864e-05,
+      "loss": 2.1201,
+      "step": 5516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8161328588374853e-05,
+      "loss": 1.7817,
+      "step": 5517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8157374456306838e-05,
+      "loss": 2.0281,
+      "step": 5518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8153420324238834e-05,
+      "loss": 1.9578,
+      "step": 5519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.814946619217082e-05,
+      "loss": 1.7666,
+      "step": 5520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8145512060102808e-05,
+      "loss": 1.7312,
+      "step": 5521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8141557928034797e-05,
+      "loss": 1.9749,
+      "step": 5522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.813760379596679e-05,
+      "loss": 2.0237,
+      "step": 5523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8133649663898775e-05,
+      "loss": 1.8597,
+      "step": 5524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8129695531830763e-05,
+      "loss": 2.0628,
+      "step": 5525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8125741399762756e-05,
+      "loss": 1.9961,
+      "step": 5526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8121787267694744e-05,
+      "loss": 1.9475,
+      "step": 5527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.811783313562673e-05,
+      "loss": 2.3209,
+      "step": 5528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.811387900355872e-05,
+      "loss": 2.0244,
+      "step": 5529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.810992487149071e-05,
+      "loss": 1.9383,
+      "step": 5530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.81059707394227e-05,
+      "loss": 1.9923,
+      "step": 5531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8102016607354685e-05,
+      "loss": 1.9194,
+      "step": 5532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8098062475286674e-05,
+      "loss": 1.9843,
+      "step": 5533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8094108343218666e-05,
+      "loss": 2.0941,
+      "step": 5534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8090154211150655e-05,
+      "loss": 2.0958,
+      "step": 5535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.808620007908264e-05,
+      "loss": 2.302,
+      "step": 5536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.808224594701463e-05,
+      "loss": 1.8276,
+      "step": 5537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.807829181494662e-05,
+      "loss": 2.0487,
+      "step": 5538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.807433768287861e-05,
+      "loss": 2.0653,
+      "step": 5539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8070383550810596e-05,
+      "loss": 1.7574,
+      "step": 5540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8066429418742585e-05,
+      "loss": 1.8615,
+      "step": 5541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8062475286674577e-05,
+      "loss": 1.7745,
+      "step": 5542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8058521154606566e-05,
+      "loss": 1.8941,
+      "step": 5543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.805456702253855e-05,
+      "loss": 1.9206,
+      "step": 5544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8050612890470547e-05,
+      "loss": 1.7819,
+      "step": 5545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8046658758402532e-05,
+      "loss": 2.1377,
+      "step": 5546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.804270462633452e-05,
+      "loss": 1.8679,
+      "step": 5547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8038750494266506e-05,
+      "loss": 2.0815,
+      "step": 5548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8034796362198502e-05,
+      "loss": 1.7594,
+      "step": 5549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8030842230130487e-05,
+      "loss": 2.1794,
+      "step": 5550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8026888098062476e-05,
+      "loss": 1.9896,
+      "step": 5551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8022933965994465e-05,
+      "loss": 1.9399,
+      "step": 5552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8018979833926457e-05,
+      "loss": 2.1341,
+      "step": 5553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8015025701858443e-05,
+      "loss": 1.8272,
+      "step": 5554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.801107156979043e-05,
+      "loss": 1.9877,
+      "step": 5555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.800711743772242e-05,
+      "loss": 2.0662,
+      "step": 5556
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8003163305654413e-05,
+      "loss": 2.1106,
+      "step": 5557
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7999209173586398e-05,
+      "loss": 2.0155,
+      "step": 5558
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7995255041518387e-05,
+      "loss": 2.2442,
+      "step": 5559
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.799130090945038e-05,
+      "loss": 2.0332,
+      "step": 5560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7987346777382368e-05,
+      "loss": 1.8913,
+      "step": 5561
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7983392645314353e-05,
+      "loss": 1.9297,
+      "step": 5562
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7979438513246342e-05,
+      "loss": 1.8509,
+      "step": 5563
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7975484381178334e-05,
+      "loss": 2.0469,
+      "step": 5564
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7971530249110323e-05,
+      "loss": 2.0579,
+      "step": 5565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.796757611704231e-05,
+      "loss": 2.1064,
+      "step": 5566
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7963621984974297e-05,
+      "loss": 1.9764,
+      "step": 5567
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.795966785290629e-05,
+      "loss": 1.8646,
+      "step": 5568
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.795571372083828e-05,
+      "loss": 1.9191,
+      "step": 5569
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7951759588770264e-05,
+      "loss": 2.0193,
+      "step": 5570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7947805456702253e-05,
+      "loss": 1.8893,
+      "step": 5571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7943851324634245e-05,
+      "loss": 1.7512,
+      "step": 5572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7939897192566234e-05,
+      "loss": 1.971,
+      "step": 5573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.793594306049822e-05,
+      "loss": 1.918,
+      "step": 5574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7931988928430208e-05,
+      "loss": 2.1814,
+      "step": 5575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.79280347963622e-05,
+      "loss": 2.0462,
+      "step": 5576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.792408066429419e-05,
+      "loss": 1.936,
+      "step": 5577
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7920126532226175e-05,
+      "loss": 1.8955,
+      "step": 5578
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.791617240015817e-05,
+      "loss": 2.0989,
+      "step": 5579
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7912218268090156e-05,
+      "loss": 1.9009,
+      "step": 5580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7908264136022144e-05,
+      "loss": 2.1092,
+      "step": 5581
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7904310003954133e-05,
+      "loss": 1.8135,
+      "step": 5582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7900355871886125e-05,
+      "loss": 2.0736,
+      "step": 5583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.789640173981811e-05,
+      "loss": 1.8099,
+      "step": 5584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.78924476077501e-05,
+      "loss": 2.1227,
+      "step": 5585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.788849347568209e-05,
+      "loss": 1.9326,
+      "step": 5586
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.788453934361408e-05,
+      "loss": 1.9788,
+      "step": 5587
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7880585211546066e-05,
+      "loss": 2.1847,
+      "step": 5588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7876631079478055e-05,
+      "loss": 1.8617,
+      "step": 5589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7872676947410044e-05,
+      "loss": 2.0072,
+      "step": 5590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7868722815342036e-05,
+      "loss": 1.894,
+      "step": 5591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.786476868327402e-05,
+      "loss": 1.9342,
+      "step": 5592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.786081455120601e-05,
+      "loss": 1.8417,
+      "step": 5593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7856860419138e-05,
+      "loss": 1.8037,
+      "step": 5594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.785290628706999e-05,
+      "loss": 2.0919,
+      "step": 5595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7848952155001977e-05,
+      "loss": 2.0078,
+      "step": 5596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7844998022933966e-05,
+      "loss": 1.9163,
+      "step": 5597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7841043890865958e-05,
+      "loss": 1.7866,
+      "step": 5598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7837089758797947e-05,
+      "loss": 1.8376,
+      "step": 5599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7833135626729932e-05,
+      "loss": 1.9466,
+      "step": 5600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.782918149466192e-05,
+      "loss": 1.9262,
+      "step": 5601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7825227362593913e-05,
+      "loss": 2.0028,
+      "step": 5602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7821273230525902e-05,
+      "loss": 2.1011,
+      "step": 5603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7817319098457887e-05,
+      "loss": 1.9911,
+      "step": 5604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7813364966389876e-05,
+      "loss": 2.2947,
+      "step": 5605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.780941083432187e-05,
+      "loss": 1.8773,
+      "step": 5606
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7805456702253857e-05,
+      "loss": 2.0517,
+      "step": 5607
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7801502570185846e-05,
+      "loss": 2.0226,
+      "step": 5608
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.779754843811783e-05,
+      "loss": 2.1313,
+      "step": 5609
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7793594306049824e-05,
+      "loss": 2.0923,
+      "step": 5610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7789640173981813e-05,
+      "loss": 1.6218,
+      "step": 5611
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.77856860419138e-05,
+      "loss": 1.9138,
+      "step": 5612
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7781731909845794e-05,
+      "loss": 2.2233,
+      "step": 5613
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 2.0138,
+      "step": 5614
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7773823645709768e-05,
+      "loss": 1.9811,
+      "step": 5615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7769869513641757e-05,
+      "loss": 1.8747,
+      "step": 5616
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.776591538157375e-05,
+      "loss": 1.9391,
+      "step": 5617
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7761961249505734e-05,
+      "loss": 1.983,
+      "step": 5618
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7758007117437723e-05,
+      "loss": 1.8794,
+      "step": 5619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7754052985369712e-05,
+      "loss": 1.7149,
+      "step": 5620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7750098853301704e-05,
+      "loss": 1.8525,
+      "step": 5621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.774614472123369e-05,
+      "loss": 2.1166,
+      "step": 5622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.774219058916568e-05,
+      "loss": 2.1214,
+      "step": 5623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7738236457097667e-05,
+      "loss": 1.9102,
+      "step": 5624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.773428232502966e-05,
+      "loss": 2.1703,
+      "step": 5625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7730328192961645e-05,
+      "loss": 1.9288,
+      "step": 5626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.7726374060893634e-05,
+      "loss": 1.9588,
+      "step": 5627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7722419928825623e-05,
+      "loss": 1.8745,
+      "step": 5628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7718465796757615e-05,
+      "loss": 1.876,
+      "step": 5629
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.77145116646896e-05,
+      "loss": 1.9656,
+      "step": 5630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.771055753262159e-05,
+      "loss": 1.8263,
+      "step": 5631
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.770660340055358e-05,
+      "loss": 2.267,
+      "step": 5632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.770264926848557e-05,
+      "loss": 1.8397,
+      "step": 5633
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7698695136417556e-05,
+      "loss": 1.8832,
+      "step": 5634
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7694741004349544e-05,
+      "loss": 2.0548,
+      "step": 5635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7690786872281537e-05,
+      "loss": 1.8337,
+      "step": 5636
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7686832740213526e-05,
+      "loss": 2.1107,
+      "step": 5637
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7682878608145514e-05,
+      "loss": 2.0762,
+      "step": 5638
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.76789244760775e-05,
+      "loss": 2.2916,
+      "step": 5639
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7674970344009492e-05,
+      "loss": 1.9296,
+      "step": 5640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.767101621194148e-05,
+      "loss": 1.9981,
+      "step": 5641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.766706207987347e-05,
+      "loss": 1.9488,
+      "step": 5642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7663107947805455e-05,
+      "loss": 2.0371,
+      "step": 5643
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7659153815737447e-05,
+      "loss": 2.1268,
+      "step": 5644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7655199683669436e-05,
+      "loss": 2.0213,
+      "step": 5645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7651245551601425e-05,
+      "loss": 2.1124,
+      "step": 5646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7647291419533417e-05,
+      "loss": 2.117,
+      "step": 5647
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7643337287465403e-05,
+      "loss": 1.9753,
+      "step": 5648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.763938315539739e-05,
+      "loss": 1.9011,
+      "step": 5649
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.763542902332938e-05,
+      "loss": 2.1686,
+      "step": 5650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7631474891261372e-05,
+      "loss": 1.9947,
+      "step": 5651
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7627520759193358e-05,
+      "loss": 2.0417,
+      "step": 5652
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7623566627125347e-05,
+      "loss": 2.1013,
+      "step": 5653
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7619612495057336e-05,
+      "loss": 1.9126,
+      "step": 5654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7615658362989328e-05,
+      "loss": 2.0239,
+      "step": 5655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7611704230921313e-05,
+      "loss": 1.9934,
+      "step": 5656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7607750098853302e-05,
+      "loss": 1.8007,
+      "step": 5657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.760379596678529e-05,
+      "loss": 1.9957,
+      "step": 5658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7599841834717283e-05,
+      "loss": 2.1029,
+      "step": 5659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.759588770264927e-05,
+      "loss": 2.1055,
+      "step": 5660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7591933570581257e-05,
+      "loss": 1.9771,
+      "step": 5661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7587979438513246e-05,
+      "loss": 1.7592,
+      "step": 5662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.758402530644524e-05,
+      "loss": 1.8836,
+      "step": 5663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7580071174377227e-05,
+      "loss": 2.1538,
+      "step": 5664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7576117042309213e-05,
+      "loss": 1.9857,
+      "step": 5665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7572162910241205e-05,
+      "loss": 1.9374,
+      "step": 5666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7568208778173194e-05,
+      "loss": 2.1819,
+      "step": 5667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7564254646105183e-05,
+      "loss": 2.1158,
+      "step": 5668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7560300514037168e-05,
+      "loss": 1.6922,
+      "step": 5669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.755634638196916e-05,
+      "loss": 1.893,
+      "step": 5670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.755239224990115e-05,
+      "loss": 2.0498,
+      "step": 5671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7548438117833138e-05,
+      "loss": 1.7023,
+      "step": 5672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7544483985765123e-05,
+      "loss": 1.9237,
+      "step": 5673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7540529853697116e-05,
+      "loss": 2.0543,
+      "step": 5674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7536575721629104e-05,
+      "loss": 1.7574,
+      "step": 5675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7532621589561093e-05,
+      "loss": 1.888,
+      "step": 5676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.752866745749308e-05,
+      "loss": 2.099,
+      "step": 5677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.752471332542507e-05,
+      "loss": 2.0191,
+      "step": 5678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.752075919335706e-05,
+      "loss": 2.0492,
+      "step": 5679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.751680506128905e-05,
+      "loss": 2.1517,
+      "step": 5680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7512850929221034e-05,
+      "loss": 1.965,
+      "step": 5681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7508896797153026e-05,
+      "loss": 2.1722,
+      "step": 5682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7504942665085015e-05,
+      "loss": 2.0323,
+      "step": 5683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7500988533017004e-05,
+      "loss": 1.985,
+      "step": 5684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7497034400948996e-05,
+      "loss": 1.8694,
+      "step": 5685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.749308026888098e-05,
+      "loss": 2.1056,
+      "step": 5686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.748912613681297e-05,
+      "loss": 1.8989,
+      "step": 5687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.748517200474496e-05,
+      "loss": 2.2157,
+      "step": 5688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.748121787267695e-05,
+      "loss": 1.9699,
+      "step": 5689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7477263740608937e-05,
+      "loss": 2.1977,
+      "step": 5690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7473309608540926e-05,
+      "loss": 1.9373,
+      "step": 5691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7469355476472914e-05,
+      "loss": 2.0045,
+      "step": 5692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7465401344404907e-05,
+      "loss": 1.9395,
+      "step": 5693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7461447212336895e-05,
+      "loss": 1.768,
+      "step": 5694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.745749308026888e-05,
+      "loss": 2.1132,
+      "step": 5695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.745353894820087e-05,
+      "loss": 1.9403,
+      "step": 5696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7449584816132862e-05,
+      "loss": 1.7807,
+      "step": 5697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.744563068406485e-05,
+      "loss": 1.8287,
+      "step": 5698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7441676551996836e-05,
+      "loss": 1.8358,
+      "step": 5699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.743772241992883e-05,
+      "loss": 1.8869,
+      "step": 5700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7433768287860817e-05,
+      "loss": 2.1057,
+      "step": 5701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7429814155792806e-05,
+      "loss": 1.9284,
+      "step": 5702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.742586002372479e-05,
+      "loss": 2.3637,
+      "step": 5703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7421905891656784e-05,
+      "loss": 2.0906,
+      "step": 5704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7417951759588773e-05,
+      "loss": 1.949,
+      "step": 5705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.741399762752076e-05,
+      "loss": 1.9168,
+      "step": 5706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7410043495452747e-05,
+      "loss": 1.8412,
+      "step": 5707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.740608936338474e-05,
+      "loss": 2.1579,
+      "step": 5708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7402135231316728e-05,
+      "loss": 2.1439,
+      "step": 5709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7398181099248717e-05,
+      "loss": 1.8731,
+      "step": 5710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7394226967180702e-05,
+      "loss": 1.9884,
+      "step": 5711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7390272835112694e-05,
+      "loss": 2.2062,
+      "step": 5712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7386318703044683e-05,
+      "loss": 2.07,
+      "step": 5713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7382364570976672e-05,
+      "loss": 1.7718,
+      "step": 5714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7378410438908657e-05,
+      "loss": 2.1206,
+      "step": 5715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.737445630684065e-05,
+      "loss": 2.0495,
+      "step": 5716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.737050217477264e-05,
+      "loss": 2.0551,
+      "step": 5717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7366548042704627e-05,
+      "loss": 1.7986,
+      "step": 5718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.736259391063662e-05,
+      "loss": 1.9739,
+      "step": 5719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7358639778568605e-05,
+      "loss": 1.968,
+      "step": 5720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7354685646500594e-05,
+      "loss": 2.114,
+      "step": 5721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7350731514432583e-05,
+      "loss": 2.0302,
+      "step": 5722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7346777382364575e-05,
+      "loss": 1.9869,
+      "step": 5723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7342823250296564e-05,
+      "loss": 1.9244,
+      "step": 5724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.733886911822855e-05,
+      "loss": 2.0777,
+      "step": 5725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7334914986160538e-05,
+      "loss": 2.1543,
+      "step": 5726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.733096085409253e-05,
+      "loss": 1.8377,
+      "step": 5727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.732700672202452e-05,
+      "loss": 2.0767,
+      "step": 5728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7323052589956504e-05,
+      "loss": 1.9729,
+      "step": 5729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7319098457888493e-05,
+      "loss": 2.2277,
+      "step": 5730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7315144325820485e-05,
+      "loss": 2.0038,
+      "step": 5731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7311190193752474e-05,
+      "loss": 2.251,
+      "step": 5732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.730723606168446e-05,
+      "loss": 2.1851,
+      "step": 5733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7303281929616452e-05,
+      "loss": 1.9564,
+      "step": 5734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.729932779754844e-05,
+      "loss": 1.9799,
+      "step": 5735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.729537366548043e-05,
+      "loss": 1.9743,
+      "step": 5736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7291419533412415e-05,
+      "loss": 2.0806,
+      "step": 5737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7287465401344407e-05,
+      "loss": 1.9348,
+      "step": 5738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7283511269276396e-05,
+      "loss": 1.8574,
+      "step": 5739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7279557137208385e-05,
+      "loss": 2.0008,
+      "step": 5740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.727560300514037e-05,
+      "loss": 1.8027,
+      "step": 5741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7271648873072363e-05,
+      "loss": 2.214,
+      "step": 5742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.726769474100435e-05,
+      "loss": 1.7932,
+      "step": 5743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.726374060893634e-05,
+      "loss": 2.227,
+      "step": 5744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7259786476868326e-05,
+      "loss": 2.3111,
+      "step": 5745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7255832344800318e-05,
+      "loss": 1.9774,
+      "step": 5746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7251878212732307e-05,
+      "loss": 1.8781,
+      "step": 5747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7247924080664295e-05,
+      "loss": 1.8954,
+      "step": 5748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.724396994859628e-05,
+      "loss": 2.1691,
+      "step": 5749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7240015816528277e-05,
+      "loss": 2.0326,
+      "step": 5750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7236061684460262e-05,
+      "loss": 2.192,
+      "step": 5751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.723210755239225e-05,
+      "loss": 1.9926,
+      "step": 5752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7228153420324243e-05,
+      "loss": 2.0942,
+      "step": 5753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7224199288256232e-05,
+      "loss": 2.1315,
+      "step": 5754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7220245156188217e-05,
+      "loss": 1.7504,
+      "step": 5755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7216291024120206e-05,
+      "loss": 1.9858,
+      "step": 5756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7212336892052198e-05,
+      "loss": 2.0224,
+      "step": 5757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7208382759984187e-05,
+      "loss": 2.0613,
+      "step": 5758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7204428627916173e-05,
+      "loss": 1.9328,
+      "step": 5759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.720047449584816e-05,
+      "loss": 1.9366,
+      "step": 5760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7196520363780154e-05,
+      "loss": 1.9385,
+      "step": 5761
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7192566231712142e-05,
+      "loss": 1.8765,
+      "step": 5762
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7188612099644128e-05,
+      "loss": 2.2405,
+      "step": 5763
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7184657967576117e-05,
+      "loss": 1.9362,
+      "step": 5764
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.718070383550811e-05,
+      "loss": 1.9437,
+      "step": 5765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7176749703440098e-05,
+      "loss": 2.098,
+      "step": 5766
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7172795571372083e-05,
+      "loss": 2.1644,
+      "step": 5767
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7168841439304072e-05,
+      "loss": 1.9961,
+      "step": 5768
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7164887307236064e-05,
+      "loss": 2.0457,
+      "step": 5769
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7160933175168053e-05,
+      "loss": 1.9239,
+      "step": 5770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.715697904310004e-05,
+      "loss": 1.9992,
+      "step": 5771
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.715302491103203e-05,
+      "loss": 2.3965,
+      "step": 5772
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.714907077896402e-05,
+      "loss": 1.8373,
+      "step": 5773
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.714511664689601e-05,
+      "loss": 1.8761,
+      "step": 5774
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7141162514827994e-05,
+      "loss": 1.8822,
+      "step": 5775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7137208382759986e-05,
+      "loss": 1.9983,
+      "step": 5776
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7133254250691975e-05,
+      "loss": 1.9566,
+      "step": 5777
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7129300118623964e-05,
+      "loss": 1.6482,
+      "step": 5778
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.712534598655595e-05,
+      "loss": 2.1571,
+      "step": 5779
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7121391854487945e-05,
+      "loss": 2.0086,
+      "step": 5780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.711743772241993e-05,
+      "loss": 2.1151,
+      "step": 5781
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.711348359035192e-05,
+      "loss": 1.8188,
+      "step": 5782
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7109529458283904e-05,
+      "loss": 2.0851,
+      "step": 5783
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.71055753262159e-05,
+      "loss": 1.8748,
+      "step": 5784
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7101621194147885e-05,
+      "loss": 1.8155,
+      "step": 5785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7097667062079874e-05,
+      "loss": 2.0435,
+      "step": 5786
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7093712930011867e-05,
+      "loss": 2.0191,
+      "step": 5787
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7089758797943855e-05,
+      "loss": 1.9656,
+      "step": 5788
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.708580466587584e-05,
+      "loss": 2.1459,
+      "step": 5789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.708185053380783e-05,
+      "loss": 1.9515,
+      "step": 5790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7077896401739822e-05,
+      "loss": 2.1491,
+      "step": 5791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.707394226967181e-05,
+      "loss": 1.8765,
+      "step": 5792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7069988137603796e-05,
+      "loss": 1.8324,
+      "step": 5793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7066034005535785e-05,
+      "loss": 1.7166,
+      "step": 5794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7062079873467777e-05,
+      "loss": 1.7131,
+      "step": 5795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7058125741399766e-05,
+      "loss": 1.9875,
+      "step": 5796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.705417160933175e-05,
+      "loss": 2.0093,
+      "step": 5797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.705021747726374e-05,
+      "loss": 1.9212,
+      "step": 5798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7046263345195732e-05,
+      "loss": 2.0191,
+      "step": 5799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.704230921312772e-05,
+      "loss": 1.9739,
+      "step": 5800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7038355081059707e-05,
+      "loss": 1.8091,
+      "step": 5801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7034400948991696e-05,
+      "loss": 1.9024,
+      "step": 5802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7030446816923688e-05,
+      "loss": 1.9799,
+      "step": 5803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7026492684855677e-05,
+      "loss": 2.0493,
+      "step": 5804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7022538552787662e-05,
+      "loss": 1.6897,
+      "step": 5805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7018584420719654e-05,
+      "loss": 1.6792,
+      "step": 5806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7014630288651643e-05,
+      "loss": 2.2243,
+      "step": 5807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7010676156583632e-05,
+      "loss": 1.8111,
+      "step": 5808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7006722024515617e-05,
+      "loss": 1.9317,
+      "step": 5809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7002767892447613e-05,
+      "loss": 2.046,
+      "step": 5810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.69988137603796e-05,
+      "loss": 1.9172,
+      "step": 5811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6994859628311587e-05,
+      "loss": 2.1798,
+      "step": 5812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6990905496243573e-05,
+      "loss": 2.0864,
+      "step": 5813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6986951364175568e-05,
+      "loss": 1.9095,
+      "step": 5814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6982997232107554e-05,
+      "loss": 2.1647,
+      "step": 5815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6979043100039542e-05,
+      "loss": 1.7692,
+      "step": 5816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6975088967971528e-05,
+      "loss": 1.7717,
+      "step": 5817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6971134835903524e-05,
+      "loss": 2.1885,
+      "step": 5818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.696718070383551e-05,
+      "loss": 1.9019,
+      "step": 5819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6963226571767498e-05,
+      "loss": 2.1109,
+      "step": 5820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.695927243969949e-05,
+      "loss": 2.1528,
+      "step": 5821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.695531830763148e-05,
+      "loss": 1.8405,
+      "step": 5822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6951364175563464e-05,
+      "loss": 2.099,
+      "step": 5823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6947410043495453e-05,
+      "loss": 1.9733,
+      "step": 5824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6943455911427445e-05,
+      "loss": 1.8182,
+      "step": 5825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6939501779359434e-05,
+      "loss": 1.9702,
+      "step": 5826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.693554764729142e-05,
+      "loss": 2.0212,
+      "step": 5827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.693159351522341e-05,
+      "loss": 1.9024,
+      "step": 5828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.69276393831554e-05,
+      "loss": 1.774,
+      "step": 5829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.692368525108739e-05,
+      "loss": 2.0309,
+      "step": 5830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6919731119019375e-05,
+      "loss": 2.2891,
+      "step": 5831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6915776986951364e-05,
+      "loss": 1.8294,
+      "step": 5832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6911822854883356e-05,
+      "loss": 1.925,
+      "step": 5833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6907868722815345e-05,
+      "loss": 1.9714,
+      "step": 5834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.690391459074733e-05,
+      "loss": 1.9409,
+      "step": 5835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.689996045867932e-05,
+      "loss": 1.9489,
+      "step": 5836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.689600632661131e-05,
+      "loss": 2.1165,
+      "step": 5837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.68920521945433e-05,
+      "loss": 1.8062,
+      "step": 5838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6888098062475286e-05,
+      "loss": 1.8156,
+      "step": 5839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.688414393040728e-05,
+      "loss": 2.1749,
+      "step": 5840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6880189798339267e-05,
+      "loss": 1.7246,
+      "step": 5841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6876235666271255e-05,
+      "loss": 2.011,
+      "step": 5842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.687228153420324e-05,
+      "loss": 2.0735,
+      "step": 5843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6868327402135236e-05,
+      "loss": 1.8162,
+      "step": 5844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6864373270067222e-05,
+      "loss": 2.0656,
+      "step": 5845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.686041913799921e-05,
+      "loss": 1.5829,
+      "step": 5846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6856465005931196e-05,
+      "loss": 1.822,
+      "step": 5847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6852510873863192e-05,
+      "loss": 2.0183,
+      "step": 5848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6848556741795177e-05,
+      "loss": 1.6451,
+      "step": 5849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6844602609727166e-05,
+      "loss": 1.8864,
+      "step": 5850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.684064847765915e-05,
+      "loss": 2.0393,
+      "step": 5851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6836694345591147e-05,
+      "loss": 2.0251,
+      "step": 5852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6832740213523132e-05,
+      "loss": 1.7933,
+      "step": 5853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.682878608145512e-05,
+      "loss": 2.0821,
+      "step": 5854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6824831949387107e-05,
+      "loss": 2.0625,
+      "step": 5855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6820877817319102e-05,
+      "loss": 1.8049,
+      "step": 5856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6816923685251088e-05,
+      "loss": 1.882,
+      "step": 5857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6812969553183077e-05,
+      "loss": 1.7815,
+      "step": 5858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.680901542111507e-05,
+      "loss": 1.9715,
+      "step": 5859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6805061289047058e-05,
+      "loss": 1.9925,
+      "step": 5860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6801107156979043e-05,
+      "loss": 1.9474,
+      "step": 5861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6797153024911032e-05,
+      "loss": 1.7642,
+      "step": 5862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6793198892843024e-05,
+      "loss": 1.8714,
+      "step": 5863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6789244760775013e-05,
+      "loss": 1.8442,
+      "step": 5864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6785290628707e-05,
+      "loss": 1.9491,
+      "step": 5865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6781336496638987e-05,
+      "loss": 1.7442,
+      "step": 5866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.677738236457098e-05,
+      "loss": 2.1849,
+      "step": 5867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6773428232502968e-05,
+      "loss": 1.9634,
+      "step": 5868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6769474100434954e-05,
+      "loss": 1.8074,
+      "step": 5869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6765519968366943e-05,
+      "loss": 2.158,
+      "step": 5870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6761565836298935e-05,
+      "loss": 1.9513,
+      "step": 5871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6757611704230924e-05,
+      "loss": 2.0506,
+      "step": 5872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.675365757216291e-05,
+      "loss": 1.9707,
+      "step": 5873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6749703440094905e-05,
+      "loss": 2.0368,
+      "step": 5874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.674574930802689e-05,
+      "loss": 2.0881,
+      "step": 5875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.674179517595888e-05,
+      "loss": 1.8483,
+      "step": 5876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6737841043890864e-05,
+      "loss": 2.0735,
+      "step": 5877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.673388691182286e-05,
+      "loss": 2.0635,
+      "step": 5878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6729932779754845e-05,
+      "loss": 1.8614,
+      "step": 5879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.6725978647686834e-05,
+      "loss": 2.0916,
+      "step": 5880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.672202451561882e-05,
+      "loss": 2.1211,
+      "step": 5881
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6718070383550815e-05,
+      "loss": 1.5829,
+      "step": 5882
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.67141162514828e-05,
+      "loss": 2.3027,
+      "step": 5883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.671016211941479e-05,
+      "loss": 1.9047,
+      "step": 5884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6706207987346775e-05,
+      "loss": 1.7745,
+      "step": 5885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.670225385527877e-05,
+      "loss": 2.261,
+      "step": 5886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6698299723210756e-05,
+      "loss": 1.8355,
+      "step": 5887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6694345591142745e-05,
+      "loss": 1.9898,
+      "step": 5888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.669039145907473e-05,
+      "loss": 2.0314,
+      "step": 5889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6686437327006726e-05,
+      "loss": 2.0133,
+      "step": 5890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.668248319493871e-05,
+      "loss": 1.9567,
+      "step": 5891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.66785290628707e-05,
+      "loss": 2.3204,
+      "step": 5892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6674574930802692e-05,
+      "loss": 1.9366,
+      "step": 5893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.667062079873468e-05,
+      "loss": 1.8597,
+      "step": 5894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 1.8019,
+      "step": 5895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6662712534598655e-05,
+      "loss": 1.9645,
+      "step": 5896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6658758402530648e-05,
+      "loss": 1.9951,
+      "step": 5897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6654804270462636e-05,
+      "loss": 2.0011,
+      "step": 5898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6650850138394622e-05,
+      "loss": 2.1028,
+      "step": 5899
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.664689600632661e-05,
+      "loss": 1.5667,
+      "step": 5900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6642941874258603e-05,
+      "loss": 1.9041,
+      "step": 5901
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6638987742190592e-05,
+      "loss": 2.0494,
+      "step": 5902
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6635033610122577e-05,
+      "loss": 1.9197,
+      "step": 5903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6631079478054566e-05,
+      "loss": 1.6901,
+      "step": 5904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6627125345986558e-05,
+      "loss": 2.1276,
+      "step": 5905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6623171213918547e-05,
+      "loss": 1.7608,
+      "step": 5906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6619217081850533e-05,
+      "loss": 1.7017,
+      "step": 5907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6615262949782528e-05,
+      "loss": 2.0313,
+      "step": 5908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6611308817714514e-05,
+      "loss": 2.2785,
+      "step": 5909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6607354685646502e-05,
+      "loss": 1.9776,
+      "step": 5910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6603400553578488e-05,
+      "loss": 1.7121,
+      "step": 5911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6599446421510483e-05,
+      "loss": 1.8288,
+      "step": 5912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.659549228944247e-05,
+      "loss": 2.0819,
+      "step": 5913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6591538157374458e-05,
+      "loss": 1.9275,
+      "step": 5914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6587584025306443e-05,
+      "loss": 2.0598,
+      "step": 5915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.658362989323844e-05,
+      "loss": 2.2073,
+      "step": 5916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6579675761170424e-05,
+      "loss": 1.9887,
+      "step": 5917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6575721629102413e-05,
+      "loss": 2.0034,
+      "step": 5918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.65717674970344e-05,
+      "loss": 2.0457,
+      "step": 5919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6567813364966394e-05,
+      "loss": 2.0509,
+      "step": 5920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.656385923289838e-05,
+      "loss": 1.9181,
+      "step": 5921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6559905100830368e-05,
+      "loss": 2.0559,
+      "step": 5922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6555950968762354e-05,
+      "loss": 1.8816,
+      "step": 5923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.655199683669435e-05,
+      "loss": 2.0539,
+      "step": 5924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6548042704626335e-05,
+      "loss": 2.1266,
+      "step": 5925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6544088572558324e-05,
+      "loss": 1.8886,
+      "step": 5926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6540134440490316e-05,
+      "loss": 1.9188,
+      "step": 5927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6536180308422305e-05,
+      "loss": 1.7664,
+      "step": 5928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.653222617635429e-05,
+      "loss": 1.9776,
+      "step": 5929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.652827204428628e-05,
+      "loss": 1.9032,
+      "step": 5930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.652431791221827e-05,
+      "loss": 1.6227,
+      "step": 5931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.652036378015026e-05,
+      "loss": 1.9616,
+      "step": 5932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6516409648082245e-05,
+      "loss": 1.8858,
+      "step": 5933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6512455516014234e-05,
+      "loss": 1.8679,
+      "step": 5934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6508501383946226e-05,
+      "loss": 1.9272,
+      "step": 5935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6504547251878215e-05,
+      "loss": 2.0203,
+      "step": 5936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.65005931198102e-05,
+      "loss": 2.0304,
+      "step": 5937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.649663898774219e-05,
+      "loss": 1.9434,
+      "step": 5938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6492684855674182e-05,
+      "loss": 2.0539,
+      "step": 5939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.648873072360617e-05,
+      "loss": 1.7657,
+      "step": 5940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6484776591538156e-05,
+      "loss": 1.9978,
+      "step": 5941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6480822459470145e-05,
+      "loss": 2.0907,
+      "step": 5942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6476868327402137e-05,
+      "loss": 1.8188,
+      "step": 5943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6472914195334126e-05,
+      "loss": 1.7792,
+      "step": 5944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.646896006326611e-05,
+      "loss": 1.8896,
+      "step": 5945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6465005931198107e-05,
+      "loss": 2.0191,
+      "step": 5946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6461051799130092e-05,
+      "loss": 2.1936,
+      "step": 5947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.645709766706208e-05,
+      "loss": 2.0299,
+      "step": 5948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6453143534994067e-05,
+      "loss": 2.244,
+      "step": 5949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6449189402926062e-05,
+      "loss": 2.0443,
+      "step": 5950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6445235270858048e-05,
+      "loss": 1.8928,
+      "step": 5951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6441281138790037e-05,
+      "loss": 2.2184,
+      "step": 5952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6437327006722022e-05,
+      "loss": 2.1852,
+      "step": 5953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6433372874654018e-05,
+      "loss": 1.7738,
+      "step": 5954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6429418742586003e-05,
+      "loss": 2.1556,
+      "step": 5955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6425464610517992e-05,
+      "loss": 2.0745,
+      "step": 5956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.642151047844998e-05,
+      "loss": 2.0234,
+      "step": 5957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6417556346381973e-05,
+      "loss": 1.771,
+      "step": 5958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6413602214313958e-05,
+      "loss": 2.1463,
+      "step": 5959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6409648082245947e-05,
+      "loss": 1.9516,
+      "step": 5960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.640569395017794e-05,
+      "loss": 2.0409,
+      "step": 5961
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6401739818109928e-05,
+      "loss": 1.9546,
+      "step": 5962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6397785686041914e-05,
+      "loss": 1.7302,
+      "step": 5963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6393831553973902e-05,
+      "loss": 1.7913,
+      "step": 5964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6389877421905895e-05,
+      "loss": 2.0533,
+      "step": 5965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6385923289837883e-05,
+      "loss": 1.8711,
+      "step": 5966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.638196915776987e-05,
+      "loss": 1.9712,
+      "step": 5967
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6378015025701858e-05,
+      "loss": 2.2023,
+      "step": 5968
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.637406089363385e-05,
+      "loss": 2.1556,
+      "step": 5969
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.637010676156584e-05,
+      "loss": 1.9377,
+      "step": 5970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6366152629497824e-05,
+      "loss": 1.8984,
+      "step": 5971
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6362198497429813e-05,
+      "loss": 1.7984,
+      "step": 5972
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6358244365361805e-05,
+      "loss": 1.882,
+      "step": 5973
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6354290233293794e-05,
+      "loss": 2.124,
+      "step": 5974
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.635033610122578e-05,
+      "loss": 2.1401,
+      "step": 5975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.634638196915777e-05,
+      "loss": 1.9325,
+      "step": 5976
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.634242783708976e-05,
+      "loss": 1.6663,
+      "step": 5977
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.633847370502175e-05,
+      "loss": 1.8119,
+      "step": 5978
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6334519572953735e-05,
+      "loss": 2.0866,
+      "step": 5979
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.633056544088573e-05,
+      "loss": 2.0548,
+      "step": 5980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6326611308817716e-05,
+      "loss": 1.8244,
+      "step": 5981
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6322657176749705e-05,
+      "loss": 1.9707,
+      "step": 5982
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.631870304468169e-05,
+      "loss": 1.9052,
+      "step": 5983
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6314748912613686e-05,
+      "loss": 1.8995,
+      "step": 5984
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.631079478054567e-05,
+      "loss": 1.8755,
+      "step": 5985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.630684064847766e-05,
+      "loss": 1.9935,
+      "step": 5986
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.630288651640965e-05,
+      "loss": 1.978,
+      "step": 5987
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.629893238434164e-05,
+      "loss": 2.1779,
+      "step": 5988
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6294978252273626e-05,
+      "loss": 2.0766,
+      "step": 5989
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6291024120205615e-05,
+      "loss": 1.8653,
+      "step": 5990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6287069988137604e-05,
+      "loss": 2.3198,
+      "step": 5991
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6283115856069596e-05,
+      "loss": 1.93,
+      "step": 5992
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6279161724001582e-05,
+      "loss": 2.16,
+      "step": 5993
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.627520759193357e-05,
+      "loss": 2.0558,
+      "step": 5994
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6271253459865563e-05,
+      "loss": 1.8538,
+      "step": 5995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.626729932779755e-05,
+      "loss": 1.7733,
+      "step": 5996
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6263345195729537e-05,
+      "loss": 2.0457,
+      "step": 5997
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6259391063661526e-05,
+      "loss": 1.9048,
+      "step": 5998
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6255436931593518e-05,
+      "loss": 1.9124,
+      "step": 5999
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6251482799525507e-05,
+      "loss": 2.0968,
+      "step": 6000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6247528667457492e-05,
+      "loss": 1.8047,
+      "step": 6001
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.624357453538948e-05,
+      "loss": 1.7255,
+      "step": 6002
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6239620403321473e-05,
+      "loss": 2.1348,
+      "step": 6003
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6235666271253462e-05,
+      "loss": 1.9179,
+      "step": 6004
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6231712139185448e-05,
+      "loss": 1.8643,
+      "step": 6005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6227758007117437e-05,
+      "loss": 2.0598,
+      "step": 6006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.622380387504943e-05,
+      "loss": 1.8029,
+      "step": 6007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6219849742981418e-05,
+      "loss": 1.946,
+      "step": 6008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6215895610913403e-05,
+      "loss": 1.7974,
+      "step": 6009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6211941478845392e-05,
+      "loss": 1.6904,
+      "step": 6010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6207987346777384e-05,
+      "loss": 1.8765,
+      "step": 6011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6204033214709373e-05,
+      "loss": 2.0611,
+      "step": 6012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.620007908264136e-05,
+      "loss": 1.9964,
+      "step": 6013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6196124950573354e-05,
+      "loss": 1.7993,
+      "step": 6014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.619217081850534e-05,
+      "loss": 2.0654,
+      "step": 6015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6188216686437328e-05,
+      "loss": 1.9982,
+      "step": 6016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6184262554369317e-05,
+      "loss": 1.7723,
+      "step": 6017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.618030842230131e-05,
+      "loss": 2.3285,
+      "step": 6018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6176354290233295e-05,
+      "loss": 2.0795,
+      "step": 6019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6172400158165284e-05,
+      "loss": 1.8866,
+      "step": 6020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6168446026097272e-05,
+      "loss": 2.0796,
+      "step": 6021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6164491894029265e-05,
+      "loss": 1.7475,
+      "step": 6022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.616053776196125e-05,
+      "loss": 1.9725,
+      "step": 6023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.615658362989324e-05,
+      "loss": 2.1597,
+      "step": 6024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6152629497825228e-05,
+      "loss": 1.9196,
+      "step": 6025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.614867536575722e-05,
+      "loss": 2.3039,
+      "step": 6026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6144721233689205e-05,
+      "loss": 1.9485,
+      "step": 6027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6140767101621194e-05,
+      "loss": 1.909,
+      "step": 6028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6136812969553183e-05,
+      "loss": 1.8548,
+      "step": 6029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6132858837485175e-05,
+      "loss": 1.7875,
+      "step": 6030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.612890470541716e-05,
+      "loss": 1.995,
+      "step": 6031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.612495057334915e-05,
+      "loss": 1.9076,
+      "step": 6032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.612099644128114e-05,
+      "loss": 1.9761,
+      "step": 6033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.611704230921313e-05,
+      "loss": 1.8532,
+      "step": 6034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6113088177145116e-05,
+      "loss": 2.2595,
+      "step": 6035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6109134045077105e-05,
+      "loss": 1.9916,
+      "step": 6036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6105179913009097e-05,
+      "loss": 1.8906,
+      "step": 6037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6101225780941086e-05,
+      "loss": 2.1869,
+      "step": 6038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.609727164887307e-05,
+      "loss": 2.0685,
+      "step": 6039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.609331751680506e-05,
+      "loss": 1.9217,
+      "step": 6040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6089363384737052e-05,
+      "loss": 2.2714,
+      "step": 6041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.608540925266904e-05,
+      "loss": 2.0653,
+      "step": 6042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.608145512060103e-05,
+      "loss": 1.8394,
+      "step": 6043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6077500988533015e-05,
+      "loss": 2.1542,
+      "step": 6044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6073546856465008e-05,
+      "loss": 1.9529,
+      "step": 6045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6069592724396996e-05,
+      "loss": 1.9476,
+      "step": 6046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6065638592328985e-05,
+      "loss": 2.1811,
+      "step": 6047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6061684460260977e-05,
+      "loss": 1.9619,
+      "step": 6048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6057730328192963e-05,
+      "loss": 1.7349,
+      "step": 6049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6053776196124952e-05,
+      "loss": 1.9867,
+      "step": 6050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.604982206405694e-05,
+      "loss": 1.8972,
+      "step": 6051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6045867931988933e-05,
+      "loss": 2.0594,
+      "step": 6052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6041913799920918e-05,
+      "loss": 2.1234,
+      "step": 6053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6037959667852907e-05,
+      "loss": 1.9673,
+      "step": 6054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6034005535784896e-05,
+      "loss": 2.0593,
+      "step": 6055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6030051403716888e-05,
+      "loss": 1.9629,
+      "step": 6056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6026097271648873e-05,
+      "loss": 2.285,
+      "step": 6057
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6022143139580862e-05,
+      "loss": 1.8871,
+      "step": 6058
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.601818900751285e-05,
+      "loss": 1.8092,
+      "step": 6059
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6014234875444843e-05,
+      "loss": 1.9252,
+      "step": 6060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.601028074337683e-05,
+      "loss": 1.9043,
+      "step": 6061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6006326611308818e-05,
+      "loss": 1.9221,
+      "step": 6062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6002372479240806e-05,
+      "loss": 1.9012,
+      "step": 6063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.59984183471728e-05,
+      "loss": 1.829,
+      "step": 6064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5994464215104784e-05,
+      "loss": 1.8677,
+      "step": 6065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5990510083036773e-05,
+      "loss": 1.879,
+      "step": 6066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5986555950968765e-05,
+      "loss": 1.9372,
+      "step": 6067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5982601818900754e-05,
+      "loss": 1.9051,
+      "step": 6068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.597864768683274e-05,
+      "loss": 1.7281,
+      "step": 6069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5974693554764728e-05,
+      "loss": 2.1022,
+      "step": 6070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.597073942269672e-05,
+      "loss": 2.1902,
+      "step": 6071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.596678529062871e-05,
+      "loss": 1.877,
+      "step": 6072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5962831158560698e-05,
+      "loss": 1.8545,
+      "step": 6073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5958877026492684e-05,
+      "loss": 2.0623,
+      "step": 6074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5954922894424676e-05,
+      "loss": 1.7705,
+      "step": 6075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5950968762356665e-05,
+      "loss": 1.9684,
+      "step": 6076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5947014630288653e-05,
+      "loss": 2.0405,
+      "step": 6077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.594306049822064e-05,
+      "loss": 2.0996,
+      "step": 6078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.593910636615263e-05,
+      "loss": 1.9676,
+      "step": 6079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.593515223408462e-05,
+      "loss": 1.9069,
+      "step": 6080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.593119810201661e-05,
+      "loss": 2.1208,
+      "step": 6081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.59272439699486e-05,
+      "loss": 2.064,
+      "step": 6082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5923289837880586e-05,
+      "loss": 1.6332,
+      "step": 6083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5919335705812575e-05,
+      "loss": 1.9933,
+      "step": 6084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5915381573744564e-05,
+      "loss": 1.9458,
+      "step": 6085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5911427441676556e-05,
+      "loss": 2.0605,
+      "step": 6086
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5907473309608542e-05,
+      "loss": 2.1224,
+      "step": 6087
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.590351917754053e-05,
+      "loss": 1.8524,
+      "step": 6088
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.589956504547252e-05,
+      "loss": 1.9909,
+      "step": 6089
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.589561091340451e-05,
+      "loss": 1.9294,
+      "step": 6090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5891656781336497e-05,
+      "loss": 2.0339,
+      "step": 6091
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5887702649268486e-05,
+      "loss": 1.8899,
+      "step": 6092
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5883748517200475e-05,
+      "loss": 1.7087,
+      "step": 6093
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5879794385132467e-05,
+      "loss": 2.0866,
+      "step": 6094
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5875840253064452e-05,
+      "loss": 1.7903,
+      "step": 6095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.587188612099644e-05,
+      "loss": 2.2505,
+      "step": 6096
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.586793198892843e-05,
+      "loss": 1.7034,
+      "step": 6097
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5863977856860422e-05,
+      "loss": 1.7969,
+      "step": 6098
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.586002372479241e-05,
+      "loss": 2.0954,
+      "step": 6099
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5856069592724396e-05,
+      "loss": 2.1127,
+      "step": 6100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.585211546065639e-05,
+      "loss": 1.8874,
+      "step": 6101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5848161328588377e-05,
+      "loss": 1.7818,
+      "step": 6102
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5844207196520366e-05,
+      "loss": 1.7653,
+      "step": 6103
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5840253064452352e-05,
+      "loss": 1.8474,
+      "step": 6104
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5836298932384344e-05,
+      "loss": 1.9828,
+      "step": 6105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5832344800316333e-05,
+      "loss": 2.1301,
+      "step": 6106
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.582839066824832e-05,
+      "loss": 1.8768,
+      "step": 6107
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5824436536180307e-05,
+      "loss": 1.8602,
+      "step": 6108
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.58204824041123e-05,
+      "loss": 1.9246,
+      "step": 6109
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5816528272044288e-05,
+      "loss": 1.8502,
+      "step": 6110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5812574139976277e-05,
+      "loss": 1.7171,
+      "step": 6111
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5808620007908262e-05,
+      "loss": 1.8156,
+      "step": 6112
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5804665875840255e-05,
+      "loss": 1.9373,
+      "step": 6113
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5800711743772243e-05,
+      "loss": 1.9099,
+      "step": 6114
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5796757611704232e-05,
+      "loss": 1.9031,
+      "step": 6115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5792803479636218e-05,
+      "loss": 1.8685,
+      "step": 6116
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.578884934756821e-05,
+      "loss": 2.1058,
+      "step": 6117
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.57848952155002e-05,
+      "loss": 1.9228,
+      "step": 6118
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5780941083432188e-05,
+      "loss": 1.9959,
+      "step": 6119
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.577698695136418e-05,
+      "loss": 2.2028,
+      "step": 6120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5773032819296165e-05,
+      "loss": 1.9871,
+      "step": 6121
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5769078687228154e-05,
+      "loss": 1.9247,
+      "step": 6122
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5765124555160143e-05,
+      "loss": 2.2591,
+      "step": 6123
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5761170423092135e-05,
+      "loss": 1.8902,
+      "step": 6124
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.575721629102412e-05,
+      "loss": 1.8826,
+      "step": 6125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.575326215895611e-05,
+      "loss": 2.259,
+      "step": 6126
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5749308026888098e-05,
+      "loss": 2.1371,
+      "step": 6127
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.574535389482009e-05,
+      "loss": 2.264,
+      "step": 6128
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.574139976275208e-05,
+      "loss": 2.1137,
+      "step": 6129
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5737445630684065e-05,
+      "loss": 1.8364,
+      "step": 6130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5733491498616053e-05,
+      "loss": 1.9679,
+      "step": 6131
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5729537366548046e-05,
+      "loss": 2.0047,
+      "step": 6132
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5725583234480035e-05,
+      "loss": 1.9936,
+      "step": 6133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.572162910241202e-05,
+      "loss": 1.8717,
+      "step": 6134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5717674970344012e-05,
+      "loss": 1.9251,
+      "step": 6135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5713720838276e-05,
+      "loss": 2.2699,
+      "step": 6136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.570976670620799e-05,
+      "loss": 2.0072,
+      "step": 6137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5705812574139975e-05,
+      "loss": 1.8579,
+      "step": 6138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5701858442071967e-05,
+      "loss": 1.8674,
+      "step": 6139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5697904310003956e-05,
+      "loss": 1.9892,
+      "step": 6140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5693950177935945e-05,
+      "loss": 2.2086,
+      "step": 6141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.568999604586793e-05,
+      "loss": 1.8474,
+      "step": 6142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5686041913799923e-05,
+      "loss": 1.8049,
+      "step": 6143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.568208778173191e-05,
+      "loss": 1.6388,
+      "step": 6144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.56781336496639e-05,
+      "loss": 1.7995,
+      "step": 6145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5674179517595886e-05,
+      "loss": 1.9026,
+      "step": 6146
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5670225385527878e-05,
+      "loss": 1.7139,
+      "step": 6147
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5666271253459867e-05,
+      "loss": 1.7894,
+      "step": 6148
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5662317121391856e-05,
+      "loss": 2.4566,
+      "step": 6149
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.565836298932384e-05,
+      "loss": 2.125,
+      "step": 6150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5654408857255833e-05,
+      "loss": 2.0686,
+      "step": 6151
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5650454725187822e-05,
+      "loss": 2.1707,
+      "step": 6152
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.564650059311981e-05,
+      "loss": 1.9827,
+      "step": 6153
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5642546461051803e-05,
+      "loss": 1.8296,
+      "step": 6154
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.563859232898379e-05,
+      "loss": 1.9855,
+      "step": 6155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5634638196915778e-05,
+      "loss": 1.926,
+      "step": 6156
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5630684064847766e-05,
+      "loss": 1.8183,
+      "step": 6157
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.562672993277976e-05,
+      "loss": 2.0487,
+      "step": 6158
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5622775800711747e-05,
+      "loss": 2.241,
+      "step": 6159
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5618821668643733e-05,
+      "loss": 1.8697,
+      "step": 6160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.561486753657572e-05,
+      "loss": 2.0648,
+      "step": 6161
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5610913404507714e-05,
+      "loss": 1.5882,
+      "step": 6162
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5606959272439703e-05,
+      "loss": 2.1271,
+      "step": 6163
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5603005140371688e-05,
+      "loss": 2.001,
+      "step": 6164
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5599051008303677e-05,
+      "loss": 1.9768,
+      "step": 6165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.559509687623567e-05,
+      "loss": 1.9444,
+      "step": 6166
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5591142744167658e-05,
+      "loss": 1.9737,
+      "step": 6167
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5587188612099643e-05,
+      "loss": 1.9633,
+      "step": 6168
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5583234480031636e-05,
+      "loss": 1.9667,
+      "step": 6169
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5579280347963624e-05,
+      "loss": 2.1546,
+      "step": 6170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5575326215895613e-05,
+      "loss": 2.272,
+      "step": 6171
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.55713720838276e-05,
+      "loss": 2.1362,
+      "step": 6172
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.556741795175959e-05,
+      "loss": 1.9743,
+      "step": 6173
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.556346381969158e-05,
+      "loss": 1.9726,
+      "step": 6174
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.555950968762357e-05,
+      "loss": 1.9261,
+      "step": 6175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5555555555555554e-05,
+      "loss": 1.8585,
+      "step": 6176
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5551601423487546e-05,
+      "loss": 1.715,
+      "step": 6177
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5547647291419535e-05,
+      "loss": 1.9041,
+      "step": 6178
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5543693159351524e-05,
+      "loss": 1.9139,
+      "step": 6179
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.553973902728351e-05,
+      "loss": 2.2624,
+      "step": 6180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.55357848952155e-05,
+      "loss": 1.978,
+      "step": 6181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.553183076314749e-05,
+      "loss": 2.1089,
+      "step": 6182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.552787663107948e-05,
+      "loss": 1.9079,
+      "step": 6183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5523922499011465e-05,
+      "loss": 2.0615,
+      "step": 6184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.551996836694346e-05,
+      "loss": 2.1017,
+      "step": 6185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5516014234875446e-05,
+      "loss": 1.8358,
+      "step": 6186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5512060102807435e-05,
+      "loss": 1.7815,
+      "step": 6187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5508105970739427e-05,
+      "loss": 1.9687,
+      "step": 6188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5504151838671416e-05,
+      "loss": 1.7834,
+      "step": 6189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.55001977066034e-05,
+      "loss": 2.107,
+      "step": 6190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.549624357453539e-05,
+      "loss": 1.8197,
+      "step": 6191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5492289442467382e-05,
+      "loss": 1.8676,
+      "step": 6192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.548833531039937e-05,
+      "loss": 1.9403,
+      "step": 6193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5484381178331356e-05,
+      "loss": 2.0828,
+      "step": 6194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5480427046263345e-05,
+      "loss": 2.2173,
+      "step": 6195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5476472914195337e-05,
+      "loss": 1.8652,
+      "step": 6196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5472518782127326e-05,
+      "loss": 1.8704,
+      "step": 6197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.546856465005931e-05,
+      "loss": 2.0131,
+      "step": 6198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.54646105179913e-05,
+      "loss": 1.8241,
+      "step": 6199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5460656385923293e-05,
+      "loss": 2.1099,
+      "step": 6200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.545670225385528e-05,
+      "loss": 1.8102,
+      "step": 6201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5452748121787267e-05,
+      "loss": 2.2422,
+      "step": 6202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5448793989719256e-05,
+      "loss": 2.0453,
+      "step": 6203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5444839857651248e-05,
+      "loss": 2.0057,
+      "step": 6204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5440885725583237e-05,
+      "loss": 1.7489,
+      "step": 6205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5436931593515222e-05,
+      "loss": 2.007,
+      "step": 6206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5432977461447214e-05,
+      "loss": 1.7773,
+      "step": 6207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5429023329379203e-05,
+      "loss": 2.0198,
+      "step": 6208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5425069197311192e-05,
+      "loss": 2.0436,
+      "step": 6209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5421115065243178e-05,
+      "loss": 1.9608,
+      "step": 6210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.541716093317517e-05,
+      "loss": 1.8387,
+      "step": 6211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.541320680110716e-05,
+      "loss": 2.0867,
+      "step": 6212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5409252669039147e-05,
+      "loss": 1.9887,
+      "step": 6213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5405298536971133e-05,
+      "loss": 1.9207,
+      "step": 6214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.540134440490313e-05,
+      "loss": 1.8784,
+      "step": 6215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5397390272835114e-05,
+      "loss": 2.0747,
+      "step": 6216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5393436140767103e-05,
+      "loss": 2.1124,
+      "step": 6217
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5389482008699088e-05,
+      "loss": 1.8659,
+      "step": 6218
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5385527876631084e-05,
+      "loss": 2.101,
+      "step": 6219
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.538157374456307e-05,
+      "loss": 1.8749,
+      "step": 6220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5377619612495058e-05,
+      "loss": 1.6977,
+      "step": 6221
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.537366548042705e-05,
+      "loss": 2.0065,
+      "step": 6222
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.536971134835904e-05,
+      "loss": 1.9186,
+      "step": 6223
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5365757216291025e-05,
+      "loss": 1.9065,
+      "step": 6224
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5361803084223013e-05,
+      "loss": 2.0774,
+      "step": 6225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5357848952155006e-05,
+      "loss": 2.0273,
+      "step": 6226
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5353894820086994e-05,
+      "loss": 1.9581,
+      "step": 6227
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.534994068801898e-05,
+      "loss": 1.9246,
+      "step": 6228
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.534598655595097e-05,
+      "loss": 2.1124,
+      "step": 6229
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.534203242388296e-05,
+      "loss": 1.9739,
+      "step": 6230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.533807829181495e-05,
+      "loss": 2.0657,
+      "step": 6231
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5334124159746935e-05,
+      "loss": 1.9265,
+      "step": 6232
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5330170027678924e-05,
+      "loss": 1.9334,
+      "step": 6233
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5326215895610916e-05,
+      "loss": 1.8773,
+      "step": 6234
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5322261763542905e-05,
+      "loss": 1.9939,
+      "step": 6235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.531830763147489e-05,
+      "loss": 2.0085,
+      "step": 6236
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.531435349940688e-05,
+      "loss": 1.9642,
+      "step": 6237
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.531039936733887e-05,
+      "loss": 2.2859,
+      "step": 6238
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.530644523527086e-05,
+      "loss": 1.7133,
+      "step": 6239
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5302491103202846e-05,
+      "loss": 1.969,
+      "step": 6240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5298536971134838e-05,
+      "loss": 1.8896,
+      "step": 6241
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5294582839066827e-05,
+      "loss": 2.2327,
+      "step": 6242
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5290628706998816e-05,
+      "loss": 1.6346,
+      "step": 6243
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.52866745749308e-05,
+      "loss": 1.8709,
+      "step": 6244
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5282720442862797e-05,
+      "loss": 2.0904,
+      "step": 6245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5278766310794782e-05,
+      "loss": 2.0463,
+      "step": 6246
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.527481217872677e-05,
+      "loss": 2.0622,
+      "step": 6247
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5270858046658756e-05,
+      "loss": 1.9667,
+      "step": 6248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5266903914590752e-05,
+      "loss": 1.923,
+      "step": 6249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5262949782522737e-05,
+      "loss": 2.0662,
+      "step": 6250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5258995650454726e-05,
+      "loss": 2.0028,
+      "step": 6251
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5255041518386712e-05,
+      "loss": 2.0797,
+      "step": 6252
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5251087386318707e-05,
+      "loss": 2.0488,
+      "step": 6253
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5247133254250693e-05,
+      "loss": 1.9974,
+      "step": 6254
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.524317912218268e-05,
+      "loss": 1.6464,
+      "step": 6255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5239224990114674e-05,
+      "loss": 2.0109,
+      "step": 6256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5235270858046663e-05,
+      "loss": 1.8964,
+      "step": 6257
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5231316725978648e-05,
+      "loss": 1.8536,
+      "step": 6258
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5227362593910637e-05,
+      "loss": 2.0538,
+      "step": 6259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.522340846184263e-05,
+      "loss": 2.018,
+      "step": 6260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5219454329774618e-05,
+      "loss": 1.8243,
+      "step": 6261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5215500197706603e-05,
+      "loss": 2.0502,
+      "step": 6262
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5211546065638592e-05,
+      "loss": 2.025,
+      "step": 6263
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5207591933570584e-05,
+      "loss": 2.1321,
+      "step": 6264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5203637801502573e-05,
+      "loss": 1.9675,
+      "step": 6265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.519968366943456e-05,
+      "loss": 1.9935,
+      "step": 6266
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5195729537366547e-05,
+      "loss": 2.1641,
+      "step": 6267
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.519177540529854e-05,
+      "loss": 1.7452,
+      "step": 6268
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.518782127323053e-05,
+      "loss": 1.8207,
+      "step": 6269
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5183867141162514e-05,
+      "loss": 1.981,
+      "step": 6270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5179913009094503e-05,
+      "loss": 1.6896,
+      "step": 6271
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5175958877026495e-05,
+      "loss": 1.9212,
+      "step": 6272
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5172004744958484e-05,
+      "loss": 2.0484,
+      "step": 6273
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.516805061289047e-05,
+      "loss": 1.995,
+      "step": 6274
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5164096480822465e-05,
+      "loss": 1.9742,
+      "step": 6275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.516014234875445e-05,
+      "loss": 2.1288,
+      "step": 6276
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.515618821668644e-05,
+      "loss": 1.851,
+      "step": 6277
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5152234084618425e-05,
+      "loss": 1.9539,
+      "step": 6278
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.514827995255042e-05,
+      "loss": 2.2148,
+      "step": 6279
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5144325820482406e-05,
+      "loss": 1.8197,
+      "step": 6280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5140371688414394e-05,
+      "loss": 1.8067,
+      "step": 6281
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.513641755634638e-05,
+      "loss": 1.7689,
+      "step": 6282
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5132463424278375e-05,
+      "loss": 1.8543,
+      "step": 6283
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.512850929221036e-05,
+      "loss": 1.7568,
+      "step": 6284
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.512455516014235e-05,
+      "loss": 1.6618,
+      "step": 6285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5120601028074335e-05,
+      "loss": 1.7816,
+      "step": 6286
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.511664689600633e-05,
+      "loss": 1.732,
+      "step": 6287
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5112692763938316e-05,
+      "loss": 2.1831,
+      "step": 6288
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5108738631870305e-05,
+      "loss": 1.9698,
+      "step": 6289
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.510478449980229e-05,
+      "loss": 1.9274,
+      "step": 6290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5100830367734286e-05,
+      "loss": 1.7113,
+      "step": 6291
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.509687623566627e-05,
+      "loss": 1.9743,
+      "step": 6292
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.509292210359826e-05,
+      "loss": 1.9911,
+      "step": 6293
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5088967971530253e-05,
+      "loss": 1.6837,
+      "step": 6294
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.508501383946224e-05,
+      "loss": 1.9805,
+      "step": 6295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5081059707394227e-05,
+      "loss": 1.7525,
+      "step": 6296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5077105575326216e-05,
+      "loss": 1.9813,
+      "step": 6297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5073151443258208e-05,
+      "loss": 2.0228,
+      "step": 6298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5069197311190197e-05,
+      "loss": 1.6436,
+      "step": 6299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5065243179122182e-05,
+      "loss": 1.9125,
+      "step": 6300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.506128904705417e-05,
+      "loss": 2.1049,
+      "step": 6301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5057334914986163e-05,
+      "loss": 1.965,
+      "step": 6302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5053380782918152e-05,
+      "loss": 1.7977,
+      "step": 6303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5049426650850137e-05,
+      "loss": 1.9189,
+      "step": 6304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5045472518782126e-05,
+      "loss": 1.9371,
+      "step": 6305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.504151838671412e-05,
+      "loss": 1.9802,
+      "step": 6306
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5037564254646107e-05,
+      "loss": 1.9673,
+      "step": 6307
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5033610122578093e-05,
+      "loss": 2.0156,
+      "step": 6308
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.502965599051009e-05,
+      "loss": 1.8469,
+      "step": 6309
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5025701858442074e-05,
+      "loss": 1.8411,
+      "step": 6310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5021747726374063e-05,
+      "loss": 1.876,
+      "step": 6311
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5017793594306048e-05,
+      "loss": 2.0222,
+      "step": 6312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5013839462238044e-05,
+      "loss": 1.9023,
+      "step": 6313
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.500988533017003e-05,
+      "loss": 2.267,
+      "step": 6314
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5005931198102018e-05,
+      "loss": 1.9534,
+      "step": 6315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5001977066034003e-05,
+      "loss": 2.0791,
+      "step": 6316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4998022933965996e-05,
+      "loss": 1.965,
+      "step": 6317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4994068801897984e-05,
+      "loss": 1.8998,
+      "step": 6318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4990114669829973e-05,
+      "loss": 1.923,
+      "step": 6319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4986160537761962e-05,
+      "loss": 1.9534,
+      "step": 6320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.498220640569395e-05,
+      "loss": 2.2325,
+      "step": 6321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.497825227362594e-05,
+      "loss": 1.9176,
+      "step": 6322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.497429814155793e-05,
+      "loss": 1.7848,
+      "step": 6323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4970344009489917e-05,
+      "loss": 2.125,
+      "step": 6324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4966389877421906e-05,
+      "loss": 1.8523,
+      "step": 6325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4962435745353895e-05,
+      "loss": 2.1535,
+      "step": 6326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4958481613285887e-05,
+      "loss": 1.9467,
+      "step": 6327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4954527481217873e-05,
+      "loss": 2.1857,
+      "step": 6328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4950573349149865e-05,
+      "loss": 2.0309,
+      "step": 6329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.494661921708185e-05,
+      "loss": 2.1124,
+      "step": 6330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4942665085013843e-05,
+      "loss": 1.8405,
+      "step": 6331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4938710952945828e-05,
+      "loss": 1.8787,
+      "step": 6332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.493475682087782e-05,
+      "loss": 2.2158,
+      "step": 6333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4930802688809806e-05,
+      "loss": 2.0121,
+      "step": 6334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4926848556741798e-05,
+      "loss": 2.0146,
+      "step": 6335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4922894424673783e-05,
+      "loss": 2.0055,
+      "step": 6336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4918940292605776e-05,
+      "loss": 1.9422,
+      "step": 6337
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.491498616053776e-05,
+      "loss": 1.9995,
+      "step": 6338
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4911032028469753e-05,
+      "loss": 2.1188,
+      "step": 6339
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.490707789640174e-05,
+      "loss": 2.0254,
+      "step": 6340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.490312376433373e-05,
+      "loss": 1.8208,
+      "step": 6341
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4899169632265716e-05,
+      "loss": 2.1103,
+      "step": 6342
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.489521550019771e-05,
+      "loss": 2.1162,
+      "step": 6343
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4891261368129697e-05,
+      "loss": 2.0055,
+      "step": 6344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4887307236061686e-05,
+      "loss": 2.0043,
+      "step": 6345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4883353103993675e-05,
+      "loss": 1.7584,
+      "step": 6346
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4879398971925664e-05,
+      "loss": 2.1193,
+      "step": 6347
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4875444839857653e-05,
+      "loss": 2.1678,
+      "step": 6348
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.487149070778964e-05,
+      "loss": 1.9693,
+      "step": 6349
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.486753657572163e-05,
+      "loss": 2.3392,
+      "step": 6350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.486358244365362e-05,
+      "loss": 2.1564,
+      "step": 6351
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4859628311585608e-05,
+      "loss": 2.0861,
+      "step": 6352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4855674179517597e-05,
+      "loss": 1.8406,
+      "step": 6353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4851720047449586e-05,
+      "loss": 1.926,
+      "step": 6354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4847765915381574e-05,
+      "loss": 1.7795,
+      "step": 6355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4843811783313563e-05,
+      "loss": 1.7921,
+      "step": 6356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4839857651245552e-05,
+      "loss": 1.967,
+      "step": 6357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.483590351917754e-05,
+      "loss": 2.0393,
+      "step": 6358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.483194938710953e-05,
+      "loss": 1.9113,
+      "step": 6359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.482799525504152e-05,
+      "loss": 2.2289,
+      "step": 6360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.482404112297351e-05,
+      "loss": 1.9608,
+      "step": 6361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4820086990905496e-05,
+      "loss": 1.8686,
+      "step": 6362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.481613285883749e-05,
+      "loss": 1.8723,
+      "step": 6363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4812178726769474e-05,
+      "loss": 1.6785,
+      "step": 6364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4808224594701466e-05,
+      "loss": 1.619,
+      "step": 6365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.480427046263345e-05,
+      "loss": 1.8806,
+      "step": 6366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4800316330565444e-05,
+      "loss": 2.0964,
+      "step": 6367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.479636219849743e-05,
+      "loss": 2.0208,
+      "step": 6368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.479240806642942e-05,
+      "loss": 1.9732,
+      "step": 6369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4788453934361407e-05,
+      "loss": 2.229,
+      "step": 6370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.47844998022934e-05,
+      "loss": 1.9839,
+      "step": 6371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4780545670225384e-05,
+      "loss": 2.1674,
+      "step": 6372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4776591538157377e-05,
+      "loss": 1.865,
+      "step": 6373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4772637406089362e-05,
+      "loss": 2.0669,
+      "step": 6374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4768683274021354e-05,
+      "loss": 1.8714,
+      "step": 6375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.476472914195334e-05,
+      "loss": 1.8173,
+      "step": 6376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4760775009885332e-05,
+      "loss": 1.9481,
+      "step": 6377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.475682087781732e-05,
+      "loss": 1.8441,
+      "step": 6378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.475286674574931e-05,
+      "loss": 1.829,
+      "step": 6379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.47489126136813e-05,
+      "loss": 2.016,
+      "step": 6380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4744958481613287e-05,
+      "loss": 1.7017,
+      "step": 6381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4741004349545276e-05,
+      "loss": 1.7972,
+      "step": 6382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4737050217477265e-05,
+      "loss": 1.9905,
+      "step": 6383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4733096085409254e-05,
+      "loss": 2.1115,
+      "step": 6384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4729141953341243e-05,
+      "loss": 1.8458,
+      "step": 6385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.472518782127323e-05,
+      "loss": 1.7583,
+      "step": 6386
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.472123368920522e-05,
+      "loss": 1.9696,
+      "step": 6387
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.471727955713721e-05,
+      "loss": 1.9906,
+      "step": 6388
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4713325425069198e-05,
+      "loss": 1.7507,
+      "step": 6389
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4709371293001187e-05,
+      "loss": 2.2212,
+      "step": 6390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4705417160933176e-05,
+      "loss": 2.1124,
+      "step": 6391
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4701463028865164e-05,
+      "loss": 1.9267,
+      "step": 6392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4697508896797153e-05,
+      "loss": 1.7825,
+      "step": 6393
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4693554764729142e-05,
+      "loss": 1.8421,
+      "step": 6394
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4689600632661134e-05,
+      "loss": 1.6659,
+      "step": 6395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.468564650059312e-05,
+      "loss": 1.8859,
+      "step": 6396
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4681692368525112e-05,
+      "loss": 1.7841,
+      "step": 6397
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4677738236457097e-05,
+      "loss": 2.2702,
+      "step": 6398
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.467378410438909e-05,
+      "loss": 1.9032,
+      "step": 6399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4669829972321075e-05,
+      "loss": 1.7921,
+      "step": 6400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4665875840253067e-05,
+      "loss": 1.928,
+      "step": 6401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4661921708185053e-05,
+      "loss": 1.7188,
+      "step": 6402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4657967576117045e-05,
+      "loss": 1.888,
+      "step": 6403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.465401344404903e-05,
+      "loss": 1.9419,
+      "step": 6404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4650059311981023e-05,
+      "loss": 2.0137,
+      "step": 6405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4646105179913008e-05,
+      "loss": 1.9082,
+      "step": 6406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4642151047845e-05,
+      "loss": 1.8726,
+      "step": 6407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4638196915776986e-05,
+      "loss": 1.8565,
+      "step": 6408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4634242783708978e-05,
+      "loss": 2.1783,
+      "step": 6409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4630288651640963e-05,
+      "loss": 2.097,
+      "step": 6410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4626334519572956e-05,
+      "loss": 2.0428,
+      "step": 6411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.462238038750494e-05,
+      "loss": 1.8355,
+      "step": 6412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4618426255436933e-05,
+      "loss": 1.8098,
+      "step": 6413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4614472123368922e-05,
+      "loss": 1.8273,
+      "step": 6414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.461051799130091e-05,
+      "loss": 2.2491,
+      "step": 6415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.46065638592329e-05,
+      "loss": 1.9297,
+      "step": 6416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.460260972716489e-05,
+      "loss": 1.9712,
+      "step": 6417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4598655595096877e-05,
+      "loss": 1.8876,
+      "step": 6418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4594701463028866e-05,
+      "loss": 2.2433,
+      "step": 6419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4590747330960855e-05,
+      "loss": 1.9803,
+      "step": 6420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4586793198892844e-05,
+      "loss": 1.8172,
+      "step": 6421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4582839066824833e-05,
+      "loss": 1.8563,
+      "step": 6422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.457888493475682e-05,
+      "loss": 1.9305,
+      "step": 6423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.457493080268881e-05,
+      "loss": 2.1969,
+      "step": 6424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.45709766706208e-05,
+      "loss": 1.9349,
+      "step": 6425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4567022538552788e-05,
+      "loss": 1.9871,
+      "step": 6426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4563068406484777e-05,
+      "loss": 1.7013,
+      "step": 6427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4559114274416766e-05,
+      "loss": 1.833,
+      "step": 6428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4555160142348754e-05,
+      "loss": 1.7786,
+      "step": 6429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4551206010280743e-05,
+      "loss": 1.7486,
+      "step": 6430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4547251878212735e-05,
+      "loss": 1.8711,
+      "step": 6431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.454329774614472e-05,
+      "loss": 1.9001,
+      "step": 6432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4539343614076713e-05,
+      "loss": 1.8683,
+      "step": 6433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.45353894820087e-05,
+      "loss": 1.7326,
+      "step": 6434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.453143534994069e-05,
+      "loss": 1.7981,
+      "step": 6435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4527481217872676e-05,
+      "loss": 2.1088,
+      "step": 6436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.452352708580467e-05,
+      "loss": 1.8132,
+      "step": 6437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4519572953736654e-05,
+      "loss": 2.1923,
+      "step": 6438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4515618821668646e-05,
+      "loss": 1.7958,
+      "step": 6439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.451166468960063e-05,
+      "loss": 2.0476,
+      "step": 6440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4507710557532624e-05,
+      "loss": 1.9951,
+      "step": 6441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.450375642546461e-05,
+      "loss": 1.7345,
+      "step": 6442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.44998022933966e-05,
+      "loss": 2.121,
+      "step": 6443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4495848161328587e-05,
+      "loss": 2.0302,
+      "step": 6444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.449189402926058e-05,
+      "loss": 2.2365,
+      "step": 6445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4487939897192564e-05,
+      "loss": 2.039,
+      "step": 6446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4483985765124557e-05,
+      "loss": 1.6986,
+      "step": 6447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4480031633056545e-05,
+      "loss": 2.0314,
+      "step": 6448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4476077500988534e-05,
+      "loss": 1.9648,
+      "step": 6449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4472123368920523e-05,
+      "loss": 2.2902,
+      "step": 6450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4468169236852512e-05,
+      "loss": 2.0462,
+      "step": 6451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.44642151047845e-05,
+      "loss": 1.7779,
+      "step": 6452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.446026097271649e-05,
+      "loss": 2.0492,
+      "step": 6453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.445630684064848e-05,
+      "loss": 1.9747,
+      "step": 6454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4452352708580467e-05,
+      "loss": 1.7473,
+      "step": 6455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4448398576512456e-05,
+      "loss": 1.9982,
+      "step": 6456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4444444444444445e-05,
+      "loss": 1.9086,
+      "step": 6457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4440490312376434e-05,
+      "loss": 1.9823,
+      "step": 6458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4436536180308423e-05,
+      "loss": 1.8382,
+      "step": 6459
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.443258204824041e-05,
+      "loss": 1.8617,
+      "step": 6460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.44286279161724e-05,
+      "loss": 1.812,
+      "step": 6461
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.442467378410439e-05,
+      "loss": 1.7929,
+      "step": 6462
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4420719652036378e-05,
+      "loss": 1.8584,
+      "step": 6463
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4416765519968367e-05,
+      "loss": 2.0705,
+      "step": 6464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.441281138790036e-05,
+      "loss": 1.9936,
+      "step": 6465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4408857255832344e-05,
+      "loss": 2.345,
+      "step": 6466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4404903123764337e-05,
+      "loss": 2.026,
+      "step": 6467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4400948991696322e-05,
+      "loss": 1.9541,
+      "step": 6468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4396994859628314e-05,
+      "loss": 1.9653,
+      "step": 6469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.43930407275603e-05,
+      "loss": 2.0867,
+      "step": 6470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4389086595492292e-05,
+      "loss": 1.7897,
+      "step": 6471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4385132463424277e-05,
+      "loss": 1.9783,
+      "step": 6472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.438117833135627e-05,
+      "loss": 2.1943,
+      "step": 6473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4377224199288255e-05,
+      "loss": 1.9937,
+      "step": 6474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4373270067220247e-05,
+      "loss": 1.9967,
+      "step": 6475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4369315935152233e-05,
+      "loss": 2.0696,
+      "step": 6476
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4365361803084225e-05,
+      "loss": 1.8395,
+      "step": 6477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4361407671016214e-05,
+      "loss": 2.1878,
+      "step": 6478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4357453538948203e-05,
+      "loss": 1.9988,
+      "step": 6479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.435349940688019e-05,
+      "loss": 1.9768,
+      "step": 6480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.434954527481218e-05,
+      "loss": 1.9627,
+      "step": 6481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.434559114274417e-05,
+      "loss": 1.8606,
+      "step": 6482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4341637010676158e-05,
+      "loss": 1.9636,
+      "step": 6483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4337682878608147e-05,
+      "loss": 2.01,
+      "step": 6484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4333728746540135e-05,
+      "loss": 1.8289,
+      "step": 6485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4329774614472124e-05,
+      "loss": 1.8583,
+      "step": 6486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4325820482404113e-05,
+      "loss": 1.8948,
+      "step": 6487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4321866350336102e-05,
+      "loss": 1.9564,
+      "step": 6488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.431791221826809e-05,
+      "loss": 1.8782,
+      "step": 6489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.431395808620008e-05,
+      "loss": 1.8986,
+      "step": 6490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.431000395413207e-05,
+      "loss": 1.6547,
+      "step": 6491
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4306049822064057e-05,
+      "loss": 1.8296,
+      "step": 6492
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4302095689996046e-05,
+      "loss": 1.9271,
+      "step": 6493
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4298141557928035e-05,
+      "loss": 2.1151,
+      "step": 6494
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4294187425860024e-05,
+      "loss": 1.9083,
+      "step": 6495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4290233293792013e-05,
+      "loss": 1.9992,
+      "step": 6496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4286279161724e-05,
+      "loss": 1.8976,
+      "step": 6497
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.428232502965599e-05,
+      "loss": 2.0708,
+      "step": 6498
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.427837089758798e-05,
+      "loss": 1.9134,
+      "step": 6499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4274416765519968e-05,
+      "loss": 1.9956,
+      "step": 6500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.427046263345196e-05,
+      "loss": 1.9044,
+      "step": 6501
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4266508501383946e-05,
+      "loss": 1.7782,
+      "step": 6502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4262554369315938e-05,
+      "loss": 1.9021,
+      "step": 6503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4258600237247923e-05,
+      "loss": 1.9579,
+      "step": 6504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4254646105179915e-05,
+      "loss": 1.7161,
+      "step": 6505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4250691973111904e-05,
+      "loss": 1.6663,
+      "step": 6506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4246737841043893e-05,
+      "loss": 2.0589,
+      "step": 6507
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4242783708975882e-05,
+      "loss": 1.7545,
+      "step": 6508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.423882957690787e-05,
+      "loss": 1.8908,
+      "step": 6509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.423487544483986e-05,
+      "loss": 1.9451,
+      "step": 6510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.423092131277185e-05,
+      "loss": 1.7254,
+      "step": 6511
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4226967180703837e-05,
+      "loss": 1.8414,
+      "step": 6512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4223013048635826e-05,
+      "loss": 1.8648,
+      "step": 6513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4219058916567815e-05,
+      "loss": 1.9588,
+      "step": 6514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4215104784499804e-05,
+      "loss": 2.1276,
+      "step": 6515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4211150652431793e-05,
+      "loss": 1.8799,
+      "step": 6516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.420719652036378e-05,
+      "loss": 2.0039,
+      "step": 6517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.420324238829577e-05,
+      "loss": 1.9458,
+      "step": 6518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.419928825622776e-05,
+      "loss": 2.0557,
+      "step": 6519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4195334124159748e-05,
+      "loss": 2.216,
+      "step": 6520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4191379992091737e-05,
+      "loss": 1.8209,
+      "step": 6521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4187425860023725e-05,
+      "loss": 2.3614,
+      "step": 6522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4183471727955714e-05,
+      "loss": 1.8633,
+      "step": 6523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4179517595887703e-05,
+      "loss": 2.0775,
+      "step": 6524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4175563463819692e-05,
+      "loss": 1.8143,
+      "step": 6525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.417160933175168e-05,
+      "loss": 2.224,
+      "step": 6526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.416765519968367e-05,
+      "loss": 2.1435,
+      "step": 6527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.416370106761566e-05,
+      "loss": 2.0198,
+      "step": 6528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4159746935547647e-05,
+      "loss": 1.9192,
+      "step": 6529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4155792803479636e-05,
+      "loss": 1.8384,
+      "step": 6530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4151838671411625e-05,
+      "loss": 2.0872,
+      "step": 6531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4147884539343614e-05,
+      "loss": 2.0618,
+      "step": 6532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4143930407275603e-05,
+      "loss": 2.1812,
+      "step": 6533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.413997627520759e-05,
+      "loss": 2.0802,
+      "step": 6534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4136022143139584e-05,
+      "loss": 1.9544,
+      "step": 6535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4132068011071572e-05,
+      "loss": 2.0765,
+      "step": 6536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.412811387900356e-05,
+      "loss": 1.9434,
+      "step": 6537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.412415974693555e-05,
+      "loss": 1.9777,
+      "step": 6538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.412020561486754e-05,
+      "loss": 2.2281,
+      "step": 6539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4116251482799528e-05,
+      "loss": 1.9186,
+      "step": 6540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4112297350731517e-05,
+      "loss": 1.7121,
+      "step": 6541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4108343218663505e-05,
+      "loss": 1.8619,
+      "step": 6542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4104389086595494e-05,
+      "loss": 1.9601,
+      "step": 6543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4100434954527483e-05,
+      "loss": 2.0279,
+      "step": 6544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4096480822459472e-05,
+      "loss": 1.9053,
+      "step": 6545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.409252669039146e-05,
+      "loss": 1.9028,
+      "step": 6546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.408857255832345e-05,
+      "loss": 1.9613,
+      "step": 6547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.408461842625544e-05,
+      "loss": 2.0666,
+      "step": 6548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4080664294187427e-05,
+      "loss": 2.0641,
+      "step": 6549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4076710162119416e-05,
+      "loss": 1.9428,
+      "step": 6550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4072756030051405e-05,
+      "loss": 1.8907,
+      "step": 6551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4068801897983394e-05,
+      "loss": 1.9603,
+      "step": 6552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4064847765915382e-05,
+      "loss": 2.1017,
+      "step": 6553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.406089363384737e-05,
+      "loss": 1.8511,
+      "step": 6554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.405693950177936e-05,
+      "loss": 1.8949,
+      "step": 6555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.405298536971135e-05,
+      "loss": 2.2165,
+      "step": 6556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4049031237643338e-05,
+      "loss": 1.8794,
+      "step": 6557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4045077105575327e-05,
+      "loss": 1.9151,
+      "step": 6558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4041122973507315e-05,
+      "loss": 1.8363,
+      "step": 6559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4037168841439304e-05,
+      "loss": 2.1204,
+      "step": 6560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4033214709371293e-05,
+      "loss": 1.9879,
+      "step": 6561
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4029260577303282e-05,
+      "loss": 1.8006,
+      "step": 6562
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.402530644523527e-05,
+      "loss": 2.2014,
+      "step": 6563
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4021352313167263e-05,
+      "loss": 1.8861,
+      "step": 6564
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.401739818109925e-05,
+      "loss": 1.9483,
+      "step": 6565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.401344404903124e-05,
+      "loss": 1.8158,
+      "step": 6566
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4009489916963226e-05,
+      "loss": 1.8038,
+      "step": 6567
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4005535784895218e-05,
+      "loss": 1.8057,
+      "step": 6568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4001581652827207e-05,
+      "loss": 2.0147,
+      "step": 6569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3997627520759196e-05,
+      "loss": 2.4018,
+      "step": 6570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3993673388691185e-05,
+      "loss": 1.9272,
+      "step": 6571
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3989719256623174e-05,
+      "loss": 2.1,
+      "step": 6572
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3985765124555162e-05,
+      "loss": 1.9224,
+      "step": 6573
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.398181099248715e-05,
+      "loss": 1.6553,
+      "step": 6574
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.397785686041914e-05,
+      "loss": 1.9268,
+      "step": 6575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.397390272835113e-05,
+      "loss": 2.0468,
+      "step": 6576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3969948596283118e-05,
+      "loss": 2.0299,
+      "step": 6577
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3965994464215107e-05,
+      "loss": 2.0146,
+      "step": 6578
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3962040332147095e-05,
+      "loss": 1.7023,
+      "step": 6579
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3958086200079084e-05,
+      "loss": 2.1866,
+      "step": 6580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3954132068011073e-05,
+      "loss": 2.0233,
+      "step": 6581
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3950177935943062e-05,
+      "loss": 2.3069,
+      "step": 6582
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.394622380387505e-05,
+      "loss": 2.0307,
+      "step": 6583
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.394226967180704e-05,
+      "loss": 1.9808,
+      "step": 6584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.393831553973903e-05,
+      "loss": 1.9416,
+      "step": 6585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3934361407671017e-05,
+      "loss": 1.904,
+      "step": 6586
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3930407275603006e-05,
+      "loss": 1.7534,
+      "step": 6587
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3926453143534995e-05,
+      "loss": 2.0845,
+      "step": 6588
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3922499011466984e-05,
+      "loss": 1.8457,
+      "step": 6589
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3918544879398972e-05,
+      "loss": 2.1109,
+      "step": 6590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.391459074733096e-05,
+      "loss": 1.9484,
+      "step": 6591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3910636615262954e-05,
+      "loss": 1.8638,
+      "step": 6592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.390668248319494e-05,
+      "loss": 1.8904,
+      "step": 6593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.390272835112693e-05,
+      "loss": 1.9544,
+      "step": 6594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3898774219058917e-05,
+      "loss": 2.0408,
+      "step": 6595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.389482008699091e-05,
+      "loss": 1.8379,
+      "step": 6596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3890865954922894e-05,
+      "loss": 1.9996,
+      "step": 6597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3886911822854886e-05,
+      "loss": 1.9836,
+      "step": 6598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3882957690786872e-05,
+      "loss": 1.6873,
+      "step": 6599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3879003558718864e-05,
+      "loss": 2.0181,
+      "step": 6600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.387504942665085e-05,
+      "loss": 2.2258,
+      "step": 6601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3871095294582842e-05,
+      "loss": 1.9888,
+      "step": 6602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3867141162514827e-05,
+      "loss": 2.0508,
+      "step": 6603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.386318703044682e-05,
+      "loss": 1.8825,
+      "step": 6604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3859232898378808e-05,
+      "loss": 2.0221,
+      "step": 6605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3855278766310797e-05,
+      "loss": 1.883,
+      "step": 6606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3851324634242786e-05,
+      "loss": 1.9769,
+      "step": 6607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3847370502174775e-05,
+      "loss": 1.8614,
+      "step": 6608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3843416370106764e-05,
+      "loss": 2.0162,
+      "step": 6609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3839462238038752e-05,
+      "loss": 2.067,
+      "step": 6610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.383550810597074e-05,
+      "loss": 1.8668,
+      "step": 6611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.383155397390273e-05,
+      "loss": 2.1308,
+      "step": 6612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.382759984183472e-05,
+      "loss": 2.1754,
+      "step": 6613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3823645709766708e-05,
+      "loss": 2.0666,
+      "step": 6614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3819691577698697e-05,
+      "loss": 1.9485,
+      "step": 6615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3815737445630685e-05,
+      "loss": 2.1049,
+      "step": 6616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3811783313562674e-05,
+      "loss": 2.0015,
+      "step": 6617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3807829181494663e-05,
+      "loss": 1.9781,
+      "step": 6618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3803875049426652e-05,
+      "loss": 1.9122,
+      "step": 6619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.379992091735864e-05,
+      "loss": 1.9527,
+      "step": 6620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.379596678529063e-05,
+      "loss": 2.0163,
+      "step": 6621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3792012653222622e-05,
+      "loss": 1.8694,
+      "step": 6622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3788058521154607e-05,
+      "loss": 1.7812,
+      "step": 6623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.37841043890866e-05,
+      "loss": 2.163,
+      "step": 6624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3780150257018585e-05,
+      "loss": 1.9622,
+      "step": 6625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3776196124950577e-05,
+      "loss": 1.9913,
+      "step": 6626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3772241992882562e-05,
+      "loss": 1.9606,
+      "step": 6627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3768287860814555e-05,
+      "loss": 1.8226,
+      "step": 6628
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.376433372874654e-05,
+      "loss": 1.978,
+      "step": 6629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3760379596678532e-05,
+      "loss": 1.8165,
+      "step": 6630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3756425464610518e-05,
+      "loss": 2.0817,
+      "step": 6631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.375247133254251e-05,
+      "loss": 1.9183,
+      "step": 6632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3748517200474495e-05,
+      "loss": 1.9564,
+      "step": 6633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3744563068406488e-05,
+      "loss": 1.6378,
+      "step": 6634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3740608936338473e-05,
+      "loss": 2.0874,
+      "step": 6635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3736654804270465e-05,
+      "loss": 1.8878,
+      "step": 6636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.373270067220245e-05,
+      "loss": 2.0147,
+      "step": 6637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3728746540134443e-05,
+      "loss": 1.8732,
+      "step": 6638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3724792408066432e-05,
+      "loss": 1.742,
+      "step": 6639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.372083827599842e-05,
+      "loss": 1.8035,
+      "step": 6640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.371688414393041e-05,
+      "loss": 2.2335,
+      "step": 6641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3712930011862398e-05,
+      "loss": 2.0796,
+      "step": 6642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3708975879794387e-05,
+      "loss": 1.8134,
+      "step": 6643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3705021747726376e-05,
+      "loss": 2.1691,
+      "step": 6644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3701067615658365e-05,
+      "loss": 1.8517,
+      "step": 6645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3697113483590354e-05,
+      "loss": 2.1973,
+      "step": 6646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3693159351522342e-05,
+      "loss": 1.7656,
+      "step": 6647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.368920521945433e-05,
+      "loss": 1.9884,
+      "step": 6648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.368525108738632e-05,
+      "loss": 1.9416,
+      "step": 6649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.368129695531831e-05,
+      "loss": 1.9703,
+      "step": 6650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3677342823250298e-05,
+      "loss": 2.1205,
+      "step": 6651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3673388691182287e-05,
+      "loss": 2.0604,
+      "step": 6652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3669434559114275e-05,
+      "loss": 2.1563,
+      "step": 6653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3665480427046264e-05,
+      "loss": 2.0145,
+      "step": 6654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3661526294978253e-05,
+      "loss": 1.8398,
+      "step": 6655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3657572162910245e-05,
+      "loss": 2.1907,
+      "step": 6656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.365361803084223e-05,
+      "loss": 1.9514,
+      "step": 6657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3649663898774223e-05,
+      "loss": 2.0203,
+      "step": 6658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.364570976670621e-05,
+      "loss": 1.8884,
+      "step": 6659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.36417556346382e-05,
+      "loss": 1.8387,
+      "step": 6660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3637801502570186e-05,
+      "loss": 1.8774,
+      "step": 6661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3633847370502178e-05,
+      "loss": 1.8331,
+      "step": 6662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3629893238434164e-05,
+      "loss": 1.8518,
+      "step": 6663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3625939106366156e-05,
+      "loss": 1.9481,
+      "step": 6664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.362198497429814e-05,
+      "loss": 1.9295,
+      "step": 6665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3618030842230133e-05,
+      "loss": 2.0543,
+      "step": 6666
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.361407671016212e-05,
+      "loss": 1.9724,
+      "step": 6667
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.361012257809411e-05,
+      "loss": 1.9608,
+      "step": 6668
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3606168446026097e-05,
+      "loss": 2.1941,
+      "step": 6669
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.360221431395809e-05,
+      "loss": 2.0802,
+      "step": 6670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3598260181890074e-05,
+      "loss": 1.9675,
+      "step": 6671
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3594306049822066e-05,
+      "loss": 2.1431,
+      "step": 6672
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3590351917754052e-05,
+      "loss": 1.9548,
+      "step": 6673
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3586397785686044e-05,
+      "loss": 1.8787,
+      "step": 6674
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3582443653618033e-05,
+      "loss": 1.7425,
+      "step": 6675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3578489521550022e-05,
+      "loss": 1.7535,
+      "step": 6676
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.357453538948201e-05,
+      "loss": 1.7994,
+      "step": 6677
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3570581257414e-05,
+      "loss": 1.8567,
+      "step": 6678
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3566627125345988e-05,
+      "loss": 2.0403,
+      "step": 6679
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3562672993277977e-05,
+      "loss": 2.1156,
+      "step": 6680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3558718861209966e-05,
+      "loss": 1.9307,
+      "step": 6681
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3554764729141955e-05,
+      "loss": 1.9698,
+      "step": 6682
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3550810597073944e-05,
+      "loss": 1.8933,
+      "step": 6683
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3546856465005932e-05,
+      "loss": 1.8696,
+      "step": 6684
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.354290233293792e-05,
+      "loss": 1.7469,
+      "step": 6685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.353894820086991e-05,
+      "loss": 1.9228,
+      "step": 6686
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.35349940688019e-05,
+      "loss": 2.0103,
+      "step": 6687
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3531039936733888e-05,
+      "loss": 1.8591,
+      "step": 6688
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3527085804665877e-05,
+      "loss": 1.8496,
+      "step": 6689
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3523131672597865e-05,
+      "loss": 2.1385,
+      "step": 6690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3519177540529854e-05,
+      "loss": 2.0013,
+      "step": 6691
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3515223408461846e-05,
+      "loss": 1.8946,
+      "step": 6692
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3511269276393832e-05,
+      "loss": 1.9712,
+      "step": 6693
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3507315144325824e-05,
+      "loss": 2.0481,
+      "step": 6694
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.350336101225781e-05,
+      "loss": 1.9216,
+      "step": 6695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.34994068801898e-05,
+      "loss": 1.9231,
+      "step": 6696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3495452748121787e-05,
+      "loss": 1.9074,
+      "step": 6697
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.349149861605378e-05,
+      "loss": 2.2038,
+      "step": 6698
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3487544483985765e-05,
+      "loss": 1.904,
+      "step": 6699
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3483590351917757e-05,
+      "loss": 2.0736,
+      "step": 6700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3479636219849742e-05,
+      "loss": 1.9424,
+      "step": 6701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3475682087781735e-05,
+      "loss": 1.9313,
+      "step": 6702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.347172795571372e-05,
+      "loss": 1.8582,
+      "step": 6703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3467773823645712e-05,
+      "loss": 2.1798,
+      "step": 6704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3463819691577698e-05,
+      "loss": 1.7303,
+      "step": 6705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.345986555950969e-05,
+      "loss": 1.9664,
+      "step": 6706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3455911427441675e-05,
+      "loss": 2.1485,
+      "step": 6707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3451957295373668e-05,
+      "loss": 1.925,
+      "step": 6708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3448003163305656e-05,
+      "loss": 2.1514,
+      "step": 6709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3444049031237645e-05,
+      "loss": 2.1343,
+      "step": 6710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3440094899169634e-05,
+      "loss": 1.8625,
+      "step": 6711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3436140767101623e-05,
+      "loss": 2.0194,
+      "step": 6712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3432186635033612e-05,
+      "loss": 1.8264,
+      "step": 6713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.34282325029656e-05,
+      "loss": 1.9487,
+      "step": 6714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.342427837089759e-05,
+      "loss": 1.9886,
+      "step": 6715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3420324238829578e-05,
+      "loss": 1.8291,
+      "step": 6716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3416370106761567e-05,
+      "loss": 1.9198,
+      "step": 6717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3412415974693556e-05,
+      "loss": 1.9831,
+      "step": 6718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3408461842625545e-05,
+      "loss": 2.1124,
+      "step": 6719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3404507710557534e-05,
+      "loss": 2.0521,
+      "step": 6720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3400553578489522e-05,
+      "loss": 2.0114,
+      "step": 6721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.339659944642151e-05,
+      "loss": 1.725,
+      "step": 6722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.33926453143535e-05,
+      "loss": 1.8638,
+      "step": 6723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.338869118228549e-05,
+      "loss": 1.9386,
+      "step": 6724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3384737050217478e-05,
+      "loss": 1.8733,
+      "step": 6725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.338078291814947e-05,
+      "loss": 1.9919,
+      "step": 6726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3376828786081455e-05,
+      "loss": 1.6981,
+      "step": 6727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3372874654013448e-05,
+      "loss": 1.7439,
+      "step": 6728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3368920521945433e-05,
+      "loss": 2.0733,
+      "step": 6729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3364966389877425e-05,
+      "loss": 1.931,
+      "step": 6730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.336101225780941e-05,
+      "loss": 2.0753,
+      "step": 6731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3357058125741403e-05,
+      "loss": 1.7494,
+      "step": 6732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3353103993673388e-05,
+      "loss": 2.0146,
+      "step": 6733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.334914986160538e-05,
+      "loss": 1.822,
+      "step": 6734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3345195729537366e-05,
+      "loss": 1.9568,
+      "step": 6735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3341241597469358e-05,
+      "loss": 1.9811,
+      "step": 6736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3337287465401344e-05,
+      "loss": 2.0308,
+      "step": 6737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 2.0231,
+      "step": 6738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.332937920126532e-05,
+      "loss": 1.9121,
+      "step": 6739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3325425069197313e-05,
+      "loss": 2.0563,
+      "step": 6740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.33214709371293e-05,
+      "loss": 1.7929,
+      "step": 6741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.331751680506129e-05,
+      "loss": 1.8127,
+      "step": 6742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.331356267299328e-05,
+      "loss": 2.02,
+      "step": 6743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.330960854092527e-05,
+      "loss": 1.9688,
+      "step": 6744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3305654408857258e-05,
+      "loss": 2.0654,
+      "step": 6745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3301700276789246e-05,
+      "loss": 2.0006,
+      "step": 6746
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3297746144721235e-05,
+      "loss": 2.1323,
+      "step": 6747
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3293792012653224e-05,
+      "loss": 1.8752,
+      "step": 6748
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3289837880585213e-05,
+      "loss": 2.0279,
+      "step": 6749
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3285883748517202e-05,
+      "loss": 1.7359,
+      "step": 6750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.328192961644919e-05,
+      "loss": 1.9824,
+      "step": 6751
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.327797548438118e-05,
+      "loss": 2.066,
+      "step": 6752
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3274021352313168e-05,
+      "loss": 1.9578,
+      "step": 6753
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3270067220245157e-05,
+      "loss": 1.9275,
+      "step": 6754
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3266113088177146e-05,
+      "loss": 2.1009,
+      "step": 6755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3262158956109135e-05,
+      "loss": 2.0568,
+      "step": 6756
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3258204824041124e-05,
+      "loss": 1.8747,
+      "step": 6757
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3254250691973112e-05,
+      "loss": 1.9665,
+      "step": 6758
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.32502965599051e-05,
+      "loss": 1.9418,
+      "step": 6759
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.324634242783709e-05,
+      "loss": 1.7841,
+      "step": 6760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.324238829576908e-05,
+      "loss": 2.0706,
+      "step": 6761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.323843416370107e-05,
+      "loss": 1.7538,
+      "step": 6762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3234480031633056e-05,
+      "loss": 1.906,
+      "step": 6763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.323052589956505e-05,
+      "loss": 2.0793,
+      "step": 6764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3226571767497034e-05,
+      "loss": 2.0458,
+      "step": 6765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3222617635429026e-05,
+      "loss": 1.9467,
+      "step": 6766
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3218663503361012e-05,
+      "loss": 2.0761,
+      "step": 6767
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3214709371293004e-05,
+      "loss": 1.9083,
+      "step": 6768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.321075523922499e-05,
+      "loss": 2.0621,
+      "step": 6769
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.320680110715698e-05,
+      "loss": 2.1983,
+      "step": 6770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3202846975088967e-05,
+      "loss": 1.915,
+      "step": 6771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.319889284302096e-05,
+      "loss": 1.962,
+      "step": 6772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3194938710952945e-05,
+      "loss": 1.855,
+      "step": 6773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3190984578884937e-05,
+      "loss": 1.9367,
+      "step": 6774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3187030446816922e-05,
+      "loss": 1.9557,
+      "step": 6775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3183076314748915e-05,
+      "loss": 1.8568,
+      "step": 6776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.31791221826809e-05,
+      "loss": 2.1088,
+      "step": 6777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3175168050612892e-05,
+      "loss": 2.026,
+      "step": 6778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.317121391854488e-05,
+      "loss": 1.7614,
+      "step": 6779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.316725978647687e-05,
+      "loss": 2.1622,
+      "step": 6780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.316330565440886e-05,
+      "loss": 1.8837,
+      "step": 6781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3159351522340848e-05,
+      "loss": 1.9134,
+      "step": 6782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3155397390272836e-05,
+      "loss": 1.9814,
+      "step": 6783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3151443258204825e-05,
+      "loss": 2.1929,
+      "step": 6784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3147489126136814e-05,
+      "loss": 1.9861,
+      "step": 6785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3143534994068803e-05,
+      "loss": 2.0277,
+      "step": 6786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3139580862000792e-05,
+      "loss": 1.9334,
+      "step": 6787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.313562672993278e-05,
+      "loss": 1.9091,
+      "step": 6788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.313167259786477e-05,
+      "loss": 1.9281,
+      "step": 6789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3127718465796758e-05,
+      "loss": 2.0624,
+      "step": 6790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3123764333728747e-05,
+      "loss": 1.9843,
+      "step": 6791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3119810201660736e-05,
+      "loss": 1.7233,
+      "step": 6792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3115856069592725e-05,
+      "loss": 1.8811,
+      "step": 6793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3111901937524714e-05,
+      "loss": 2.0645,
+      "step": 6794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3107947805456702e-05,
+      "loss": 1.8974,
+      "step": 6795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3103993673388695e-05,
+      "loss": 1.8892,
+      "step": 6796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.310003954132068e-05,
+      "loss": 1.9125,
+      "step": 6797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3096085409252672e-05,
+      "loss": 2.0272,
+      "step": 6798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3092131277184658e-05,
+      "loss": 2.0732,
+      "step": 6799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.308817714511665e-05,
+      "loss": 1.9288,
+      "step": 6800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3084223013048635e-05,
+      "loss": 2.0541,
+      "step": 6801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3080268880980628e-05,
+      "loss": 1.8009,
+      "step": 6802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3076314748912613e-05,
+      "loss": 2.0066,
+      "step": 6803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3072360616844605e-05,
+      "loss": 1.911,
+      "step": 6804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.306840648477659e-05,
+      "loss": 1.8641,
+      "step": 6805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3064452352708583e-05,
+      "loss": 2.1033,
+      "step": 6806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3060498220640568e-05,
+      "loss": 1.8711,
+      "step": 6807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.305654408857256e-05,
+      "loss": 1.8922,
+      "step": 6808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3052589956504546e-05,
+      "loss": 1.9194,
+      "step": 6809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3048635824436538e-05,
+      "loss": 2.0675,
+      "step": 6810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3044681692368524e-05,
+      "loss": 1.9028,
+      "step": 6811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3040727560300516e-05,
+      "loss": 1.8359,
+      "step": 6812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3036773428232505e-05,
+      "loss": 1.9768,
+      "step": 6813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3032819296164493e-05,
+      "loss": 1.8753,
+      "step": 6814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3028865164096482e-05,
+      "loss": 2.2111,
+      "step": 6815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.302491103202847e-05,
+      "loss": 2.0076,
+      "step": 6816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.302095689996046e-05,
+      "loss": 1.9126,
+      "step": 6817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.301700276789245e-05,
+      "loss": 1.8054,
+      "step": 6818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3013048635824438e-05,
+      "loss": 1.8136,
+      "step": 6819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3009094503756426e-05,
+      "loss": 2.1196,
+      "step": 6820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3005140371688415e-05,
+      "loss": 2.1178,
+      "step": 6821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.3001186239620404e-05,
+      "loss": 2.0552,
+      "step": 6822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2997232107552393e-05,
+      "loss": 2.1067,
+      "step": 6823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2993277975484382e-05,
+      "loss": 2.0792,
+      "step": 6824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.298932384341637e-05,
+      "loss": 1.9952,
+      "step": 6825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.298536971134836e-05,
+      "loss": 1.7286,
+      "step": 6826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2981415579280348e-05,
+      "loss": 2.1008,
+      "step": 6827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2977461447212337e-05,
+      "loss": 2.161,
+      "step": 6828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2973507315144326e-05,
+      "loss": 1.9284,
+      "step": 6829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2969553183076318e-05,
+      "loss": 2.0279,
+      "step": 6830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2965599051008303e-05,
+      "loss": 1.8263,
+      "step": 6831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2961644918940296e-05,
+      "loss": 1.9639,
+      "step": 6832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.295769078687228e-05,
+      "loss": 1.8909,
+      "step": 6833
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2953736654804273e-05,
+      "loss": 1.9511,
+      "step": 6834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.294978252273626e-05,
+      "loss": 1.9076,
+      "step": 6835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.294582839066825e-05,
+      "loss": 2.0303,
+      "step": 6836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2941874258600236e-05,
+      "loss": 1.8396,
+      "step": 6837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.293792012653223e-05,
+      "loss": 1.9309,
+      "step": 6838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2933965994464214e-05,
+      "loss": 2.2187,
+      "step": 6839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2930011862396206e-05,
+      "loss": 1.992,
+      "step": 6840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2926057730328192e-05,
+      "loss": 1.9284,
+      "step": 6841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2922103598260184e-05,
+      "loss": 1.9226,
+      "step": 6842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.291814946619217e-05,
+      "loss": 1.9326,
+      "step": 6843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.291419533412416e-05,
+      "loss": 2.2198,
+      "step": 6844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2910241202056147e-05,
+      "loss": 2.0388,
+      "step": 6845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.290628706998814e-05,
+      "loss": 1.7645,
+      "step": 6846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2902332937920128e-05,
+      "loss": 1.9218,
+      "step": 6847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2898378805852117e-05,
+      "loss": 1.8197,
+      "step": 6848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2894424673784106e-05,
+      "loss": 2.0087,
+      "step": 6849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2890470541716095e-05,
+      "loss": 1.8812,
+      "step": 6850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2886516409648083e-05,
+      "loss": 1.9261,
+      "step": 6851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2882562277580072e-05,
+      "loss": 1.8551,
+      "step": 6852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.287860814551206e-05,
+      "loss": 2.0695,
+      "step": 6853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.287465401344405e-05,
+      "loss": 1.9682,
+      "step": 6854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.287069988137604e-05,
+      "loss": 2.0843,
+      "step": 6855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2866745749308028e-05,
+      "loss": 1.9119,
+      "step": 6856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2862791617240016e-05,
+      "loss": 2.1766,
+      "step": 6857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2858837485172005e-05,
+      "loss": 1.784,
+      "step": 6858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2854883353103994e-05,
+      "loss": 1.9275,
+      "step": 6859
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2850929221035983e-05,
+      "loss": 1.6777,
+      "step": 6860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.284697508896797e-05,
+      "loss": 1.7544,
+      "step": 6861
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.284302095689996e-05,
+      "loss": 2.0715,
+      "step": 6862
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.283906682483195e-05,
+      "loss": 1.9004,
+      "step": 6863
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2835112692763938e-05,
+      "loss": 1.9057,
+      "step": 6864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2831158560695927e-05,
+      "loss": 1.9834,
+      "step": 6865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.282720442862792e-05,
+      "loss": 2.1886,
+      "step": 6866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2823250296559905e-05,
+      "loss": 1.938,
+      "step": 6867
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2819296164491897e-05,
+      "loss": 1.9721,
+      "step": 6868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2815342032423882e-05,
+      "loss": 1.9051,
+      "step": 6869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2811387900355875e-05,
+      "loss": 1.8593,
+      "step": 6870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.280743376828786e-05,
+      "loss": 1.7273,
+      "step": 6871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2803479636219852e-05,
+      "loss": 2.0473,
+      "step": 6872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2799525504151838e-05,
+      "loss": 1.856,
+      "step": 6873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.279557137208383e-05,
+      "loss": 1.8356,
+      "step": 6874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2791617240015815e-05,
+      "loss": 1.904,
+      "step": 6875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2787663107947807e-05,
+      "loss": 2.0439,
+      "step": 6876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2783708975879793e-05,
+      "loss": 1.992,
+      "step": 6877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2779754843811785e-05,
+      "loss": 1.9698,
+      "step": 6878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.277580071174377e-05,
+      "loss": 1.8933,
+      "step": 6879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2771846579675763e-05,
+      "loss": 1.9166,
+      "step": 6880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2767892447607748e-05,
+      "loss": 1.8534,
+      "step": 6881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.276393831553974e-05,
+      "loss": 2.0065,
+      "step": 6882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.275998418347173e-05,
+      "loss": 1.9713,
+      "step": 6883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2756030051403718e-05,
+      "loss": 1.7689,
+      "step": 6884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2752075919335707e-05,
+      "loss": 1.942,
+      "step": 6885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2748121787267696e-05,
+      "loss": 1.8333,
+      "step": 6886
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2744167655199685e-05,
+      "loss": 1.9382,
+      "step": 6887
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2740213523131673e-05,
+      "loss": 2.0551,
+      "step": 6888
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2736259391063662e-05,
+      "loss": 2.009,
+      "step": 6889
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.273230525899565e-05,
+      "loss": 2.0019,
+      "step": 6890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.272835112692764e-05,
+      "loss": 1.7636,
+      "step": 6891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.272439699485963e-05,
+      "loss": 2.107,
+      "step": 6892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2720442862791618e-05,
+      "loss": 1.6985,
+      "step": 6893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2716488730723606e-05,
+      "loss": 2.0637,
+      "step": 6894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2712534598655595e-05,
+      "loss": 1.7326,
+      "step": 6895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2708580466587584e-05,
+      "loss": 1.9946,
+      "step": 6896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2704626334519573e-05,
+      "loss": 1.858,
+      "step": 6897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.270067220245156e-05,
+      "loss": 1.8736,
+      "step": 6898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.269671807038355e-05,
+      "loss": 1.9532,
+      "step": 6899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2692763938315543e-05,
+      "loss": 1.8954,
+      "step": 6900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2688809806247528e-05,
+      "loss": 1.9222,
+      "step": 6901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.268485567417952e-05,
+      "loss": 1.9638,
+      "step": 6902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2680901542111506e-05,
+      "loss": 1.8504,
+      "step": 6903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2676947410043498e-05,
+      "loss": 2.0151,
+      "step": 6904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2672993277975483e-05,
+      "loss": 1.9341,
+      "step": 6905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2669039145907476e-05,
+      "loss": 1.9847,
+      "step": 6906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.266508501383946e-05,
+      "loss": 2.0106,
+      "step": 6907
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2661130881771453e-05,
+      "loss": 2.0652,
+      "step": 6908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.265717674970344e-05,
+      "loss": 2.0486,
+      "step": 6909
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.265322261763543e-05,
+      "loss": 2.0346,
+      "step": 6910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2649268485567416e-05,
+      "loss": 1.9678,
+      "step": 6911
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.264531435349941e-05,
+      "loss": 1.8472,
+      "step": 6912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2641360221431397e-05,
+      "loss": 1.9145,
+      "step": 6913
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2637406089363386e-05,
+      "loss": 1.9257,
+      "step": 6914
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2633451957295375e-05,
+      "loss": 1.8326,
+      "step": 6915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2629497825227364e-05,
+      "loss": 1.8851,
+      "step": 6916
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2625543693159353e-05,
+      "loss": 1.7238,
+      "step": 6917
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.262158956109134e-05,
+      "loss": 2.0016,
+      "step": 6918
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.261763542902333e-05,
+      "loss": 2.2443,
+      "step": 6919
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.261368129695532e-05,
+      "loss": 1.9301,
+      "step": 6920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2609727164887308e-05,
+      "loss": 2.0134,
+      "step": 6921
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2605773032819297e-05,
+      "loss": 2.0126,
+      "step": 6922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2601818900751286e-05,
+      "loss": 1.7971,
+      "step": 6923
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2597864768683275e-05,
+      "loss": 1.9727,
+      "step": 6924
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2593910636615263e-05,
+      "loss": 1.9488,
+      "step": 6925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2589956504547252e-05,
+      "loss": 2.2234,
+      "step": 6926
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.258600237247924e-05,
+      "loss": 2.0109,
+      "step": 6927
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.258204824041123e-05,
+      "loss": 2.0647,
+      "step": 6928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.257809410834322e-05,
+      "loss": 1.9319,
+      "step": 6929
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2574139976275208e-05,
+      "loss": 1.8572,
+      "step": 6930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2570185844207196e-05,
+      "loss": 1.8379,
+      "step": 6931
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2566231712139185e-05,
+      "loss": 1.7735,
+      "step": 6932
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2562277580071174e-05,
+      "loss": 1.9335,
+      "step": 6933
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2558323448003166e-05,
+      "loss": 1.9747,
+      "step": 6934
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.255436931593515e-05,
+      "loss": 1.8317,
+      "step": 6935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2550415183867144e-05,
+      "loss": 1.9371,
+      "step": 6936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.254646105179913e-05,
+      "loss": 2.0964,
+      "step": 6937
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.254250691973112e-05,
+      "loss": 1.8955,
+      "step": 6938
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2538552787663107e-05,
+      "loss": 1.8397,
+      "step": 6939
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.25345986555951e-05,
+      "loss": 1.9021,
+      "step": 6940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2530644523527088e-05,
+      "loss": 1.8713,
+      "step": 6941
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2526690391459077e-05,
+      "loss": 1.9056,
+      "step": 6942
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2522736259391066e-05,
+      "loss": 1.7617,
+      "step": 6943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2518782127323054e-05,
+      "loss": 1.9359,
+      "step": 6944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2514827995255043e-05,
+      "loss": 2.0395,
+      "step": 6945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2510873863187032e-05,
+      "loss": 1.9323,
+      "step": 6946
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.250691973111902e-05,
+      "loss": 2.0753,
+      "step": 6947
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.250296559905101e-05,
+      "loss": 1.9762,
+      "step": 6948
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2499011466983e-05,
+      "loss": 1.818,
+      "step": 6949
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2495057334914987e-05,
+      "loss": 2.0044,
+      "step": 6950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2491103202846976e-05,
+      "loss": 2.0345,
+      "step": 6951
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2487149070778965e-05,
+      "loss": 1.961,
+      "step": 6952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2483194938710954e-05,
+      "loss": 1.9967,
+      "step": 6953
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2479240806642943e-05,
+      "loss": 1.9766,
+      "step": 6954
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.247528667457493e-05,
+      "loss": 1.7864,
+      "step": 6955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.247133254250692e-05,
+      "loss": 2.0204,
+      "step": 6956
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.246737841043891e-05,
+      "loss": 2.2803,
+      "step": 6957
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2463424278370898e-05,
+      "loss": 1.7565,
+      "step": 6958
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2459470146302887e-05,
+      "loss": 1.9629,
+      "step": 6959
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2455516014234876e-05,
+      "loss": 2.0179,
+      "step": 6960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2451561882166865e-05,
+      "loss": 1.6613,
+      "step": 6961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2447607750098853e-05,
+      "loss": 1.8131,
+      "step": 6962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2443653618030842e-05,
+      "loss": 1.9572,
+      "step": 6963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.243969948596283e-05,
+      "loss": 1.7876,
+      "step": 6964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.243574535389482e-05,
+      "loss": 2.0226,
+      "step": 6965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.243179122182681e-05,
+      "loss": 1.8388,
+      "step": 6966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2427837089758798e-05,
+      "loss": 1.9663,
+      "step": 6967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2423882957690786e-05,
+      "loss": 1.8553,
+      "step": 6968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2419928825622775e-05,
+      "loss": 1.8708,
+      "step": 6969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2415974693554767e-05,
+      "loss": 2.3554,
+      "step": 6970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2412020561486756e-05,
+      "loss": 1.9276,
+      "step": 6971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2408066429418745e-05,
+      "loss": 2.1153,
+      "step": 6972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2404112297350734e-05,
+      "loss": 2.1219,
+      "step": 6973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2400158165282723e-05,
+      "loss": 2.0564,
+      "step": 6974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.239620403321471e-05,
+      "loss": 1.6778,
+      "step": 6975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.23922499011467e-05,
+      "loss": 1.9363,
+      "step": 6976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.238829576907869e-05,
+      "loss": 1.9581,
+      "step": 6977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2384341637010678e-05,
+      "loss": 2.2339,
+      "step": 6978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2380387504942667e-05,
+      "loss": 2.0231,
+      "step": 6979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2376433372874656e-05,
+      "loss": 2.1095,
+      "step": 6980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2372479240806644e-05,
+      "loss": 2.0423,
+      "step": 6981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2368525108738633e-05,
+      "loss": 2.0599,
+      "step": 6982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2364570976670622e-05,
+      "loss": 1.9472,
+      "step": 6983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.236061684460261e-05,
+      "loss": 2.0309,
+      "step": 6984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.23566627125346e-05,
+      "loss": 2.063,
+      "step": 6985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.235270858046659e-05,
+      "loss": 2.03,
+      "step": 6986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2348754448398577e-05,
+      "loss": 1.8463,
+      "step": 6987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2344800316330566e-05,
+      "loss": 1.9015,
+      "step": 6988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2340846184262555e-05,
+      "loss": 1.8324,
+      "step": 6989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2336892052194544e-05,
+      "loss": 1.8306,
+      "step": 6990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2332937920126533e-05,
+      "loss": 1.9894,
+      "step": 6991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.232898378805852e-05,
+      "loss": 2.089,
+      "step": 6992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.232502965599051e-05,
+      "loss": 1.9638,
+      "step": 6993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.23210755239225e-05,
+      "loss": 2.0317,
+      "step": 6994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2317121391854488e-05,
+      "loss": 2.0507,
+      "step": 6995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2313167259786477e-05,
+      "loss": 1.8895,
+      "step": 6996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2309213127718466e-05,
+      "loss": 1.8597,
+      "step": 6997
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2305258995650455e-05,
+      "loss": 1.9714,
+      "step": 6998
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2301304863582447e-05,
+      "loss": 2.1981,
+      "step": 6999
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2297350731514432e-05,
+      "loss": 1.7566,
+      "step": 7000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2293396599446424e-05,
+      "loss": 1.8666,
+      "step": 7001
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.228944246737841e-05,
+      "loss": 1.9121,
+      "step": 7002
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2285488335310402e-05,
+      "loss": 1.8887,
+      "step": 7003
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.228153420324239e-05,
+      "loss": 1.7836,
+      "step": 7004
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.227758007117438e-05,
+      "loss": 2.0866,
+      "step": 7005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.227362593910637e-05,
+      "loss": 2.16,
+      "step": 7006
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2269671807038357e-05,
+      "loss": 1.91,
+      "step": 7007
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2265717674970346e-05,
+      "loss": 2.0311,
+      "step": 7008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2261763542902335e-05,
+      "loss": 1.9228,
+      "step": 7009
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2257809410834324e-05,
+      "loss": 1.8213,
+      "step": 7010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2253855278766313e-05,
+      "loss": 1.8337,
+      "step": 7011
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.22499011466983e-05,
+      "loss": 1.9364,
+      "step": 7012
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.224594701463029e-05,
+      "loss": 1.9198,
+      "step": 7013
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.224199288256228e-05,
+      "loss": 2.0038,
+      "step": 7014
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2238038750494268e-05,
+      "loss": 1.8,
+      "step": 7015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2234084618426257e-05,
+      "loss": 1.7793,
+      "step": 7016
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2230130486358246e-05,
+      "loss": 2.0635,
+      "step": 7017
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.2226176354290234e-05,
+      "loss": 2.0788,
+      "step": 7018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 1.9132,
+      "step": 7019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2218268090154212e-05,
+      "loss": 2.0383,
+      "step": 7020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.22143139580862e-05,
+      "loss": 2.0677,
+      "step": 7021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.221035982601819e-05,
+      "loss": 2.0709,
+      "step": 7022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.220640569395018e-05,
+      "loss": 1.7479,
+      "step": 7023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2202451561882167e-05,
+      "loss": 2.0026,
+      "step": 7024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2198497429814156e-05,
+      "loss": 1.9288,
+      "step": 7025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2194543297746145e-05,
+      "loss": 2.1184,
+      "step": 7026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2190589165678137e-05,
+      "loss": 2.0438,
+      "step": 7027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2186635033610123e-05,
+      "loss": 1.9633,
+      "step": 7028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2182680901542115e-05,
+      "loss": 1.8398,
+      "step": 7029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.21787267694741e-05,
+      "loss": 1.9475,
+      "step": 7030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2174772637406093e-05,
+      "loss": 2.0041,
+      "step": 7031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2170818505338078e-05,
+      "loss": 2.1689,
+      "step": 7032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.216686437327007e-05,
+      "loss": 2.1381,
+      "step": 7033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2162910241202056e-05,
+      "loss": 2.122,
+      "step": 7034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2158956109134048e-05,
+      "loss": 1.9877,
+      "step": 7035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2155001977066033e-05,
+      "loss": 1.8875,
+      "step": 7036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2151047844998026e-05,
+      "loss": 2.2528,
+      "step": 7037
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.214709371293001e-05,
+      "loss": 1.9111,
+      "step": 7038
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2143139580862003e-05,
+      "loss": 2.0472,
+      "step": 7039
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2139185448793992e-05,
+      "loss": 1.5943,
+      "step": 7040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.213523131672598e-05,
+      "loss": 2.1538,
+      "step": 7041
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.213127718465797e-05,
+      "loss": 1.9545,
+      "step": 7042
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.212732305258996e-05,
+      "loss": 1.9323,
+      "step": 7043
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2123368920521947e-05,
+      "loss": 1.9418,
+      "step": 7044
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2119414788453936e-05,
+      "loss": 1.7983,
+      "step": 7045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2115460656385925e-05,
+      "loss": 1.7077,
+      "step": 7046
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2111506524317914e-05,
+      "loss": 1.9129,
+      "step": 7047
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2107552392249903e-05,
+      "loss": 1.7695,
+      "step": 7048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.210359826018189e-05,
+      "loss": 2.005,
+      "step": 7049
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.209964412811388e-05,
+      "loss": 1.8831,
+      "step": 7050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.209568999604587e-05,
+      "loss": 1.9439,
+      "step": 7051
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2091735863977858e-05,
+      "loss": 1.8325,
+      "step": 7052
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2087781731909847e-05,
+      "loss": 2.065,
+      "step": 7053
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2083827599841836e-05,
+      "loss": 1.7768,
+      "step": 7054
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2079873467773824e-05,
+      "loss": 1.8662,
+      "step": 7055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2075919335705813e-05,
+      "loss": 1.9154,
+      "step": 7056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2071965203637805e-05,
+      "loss": 1.9882,
+      "step": 7057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.206801107156979e-05,
+      "loss": 1.8798,
+      "step": 7058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2064056939501783e-05,
+      "loss": 2.0376,
+      "step": 7059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.206010280743377e-05,
+      "loss": 1.9292,
+      "step": 7060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.205614867536576e-05,
+      "loss": 1.8207,
+      "step": 7061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2052194543297746e-05,
+      "loss": 1.8834,
+      "step": 7062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.204824041122974e-05,
+      "loss": 1.8043,
+      "step": 7063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2044286279161724e-05,
+      "loss": 2.0083,
+      "step": 7064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2040332147093716e-05,
+      "loss": 1.9343,
+      "step": 7065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.20363780150257e-05,
+      "loss": 1.9647,
+      "step": 7066
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2032423882957694e-05,
+      "loss": 1.8682,
+      "step": 7067
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.202846975088968e-05,
+      "loss": 1.9145,
+      "step": 7068
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.202451561882167e-05,
+      "loss": 1.901,
+      "step": 7069
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2020561486753657e-05,
+      "loss": 1.912,
+      "step": 7070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.201660735468565e-05,
+      "loss": 1.9401,
+      "step": 7071
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2012653222617635e-05,
+      "loss": 1.9461,
+      "step": 7072
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2008699090549627e-05,
+      "loss": 1.9975,
+      "step": 7073
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2004744958481616e-05,
+      "loss": 2.0722,
+      "step": 7074
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.2000790826413604e-05,
+      "loss": 1.7941,
+      "step": 7075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1996836694345593e-05,
+      "loss": 2.1078,
+      "step": 7076
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1992882562277582e-05,
+      "loss": 2.0652,
+      "step": 7077
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.198892843020957e-05,
+      "loss": 1.8147,
+      "step": 7078
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.198497429814156e-05,
+      "loss": 1.8382,
+      "step": 7079
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.198102016607355e-05,
+      "loss": 1.9735,
+      "step": 7080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1977066034005537e-05,
+      "loss": 1.8409,
+      "step": 7081
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1973111901937526e-05,
+      "loss": 1.8554,
+      "step": 7082
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1969157769869515e-05,
+      "loss": 2.0411,
+      "step": 7083
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1965203637801504e-05,
+      "loss": 1.8116,
+      "step": 7084
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1961249505733493e-05,
+      "loss": 1.9511,
+      "step": 7085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.195729537366548e-05,
+      "loss": 1.9505,
+      "step": 7086
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.195334124159747e-05,
+      "loss": 1.8503,
+      "step": 7087
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.194938710952946e-05,
+      "loss": 1.7535,
+      "step": 7088
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1945432977461448e-05,
+      "loss": 1.9357,
+      "step": 7089
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1941478845393437e-05,
+      "loss": 1.9537,
+      "step": 7090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.193752471332543e-05,
+      "loss": 1.9439,
+      "step": 7091
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1933570581257414e-05,
+      "loss": 2.2076,
+      "step": 7092
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1929616449189407e-05,
+      "loss": 1.8984,
+      "step": 7093
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1925662317121392e-05,
+      "loss": 1.9264,
+      "step": 7094
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1921708185053384e-05,
+      "loss": 1.9991,
+      "step": 7095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.191775405298537e-05,
+      "loss": 2.0336,
+      "step": 7096
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1913799920917362e-05,
+      "loss": 2.0203,
+      "step": 7097
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1909845788849347e-05,
+      "loss": 1.7012,
+      "step": 7098
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.190589165678134e-05,
+      "loss": 1.9644,
+      "step": 7099
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1901937524713325e-05,
+      "loss": 2.0365,
+      "step": 7100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1897983392645317e-05,
+      "loss": 2.0649,
+      "step": 7101
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1894029260577303e-05,
+      "loss": 2.1769,
+      "step": 7102
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1890075128509295e-05,
+      "loss": 1.9682,
+      "step": 7103
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.188612099644128e-05,
+      "loss": 1.9875,
+      "step": 7104
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1882166864373273e-05,
+      "loss": 1.7383,
+      "step": 7105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1878212732305258e-05,
+      "loss": 1.8363,
+      "step": 7106
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.187425860023725e-05,
+      "loss": 1.8973,
+      "step": 7107
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.187030446816924e-05,
+      "loss": 2.0627,
+      "step": 7108
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1866350336101228e-05,
+      "loss": 1.9917,
+      "step": 7109
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1862396204033217e-05,
+      "loss": 2.0212,
+      "step": 7110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1858442071965206e-05,
+      "loss": 1.9549,
+      "step": 7111
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1854487939897194e-05,
+      "loss": 2.2037,
+      "step": 7112
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1850533807829183e-05,
+      "loss": 2.113,
+      "step": 7113
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1846579675761172e-05,
+      "loss": 1.9911,
+      "step": 7114
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.184262554369316e-05,
+      "loss": 1.979,
+      "step": 7115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.183867141162515e-05,
+      "loss": 1.8876,
+      "step": 7116
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.183471727955714e-05,
+      "loss": 1.9218,
+      "step": 7117
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1830763147489127e-05,
+      "loss": 2.0064,
+      "step": 7118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1826809015421116e-05,
+      "loss": 1.7021,
+      "step": 7119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1822854883353105e-05,
+      "loss": 1.8539,
+      "step": 7120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1818900751285094e-05,
+      "loss": 1.7568,
+      "step": 7121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1814946619217083e-05,
+      "loss": 1.8131,
+      "step": 7122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.181099248714907e-05,
+      "loss": 1.9342,
+      "step": 7123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.180703835508106e-05,
+      "loss": 1.8427,
+      "step": 7124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.180308422301305e-05,
+      "loss": 1.8843,
+      "step": 7125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1799130090945038e-05,
+      "loss": 1.7517,
+      "step": 7126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.179517595887703e-05,
+      "loss": 1.9616,
+      "step": 7127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1791221826809016e-05,
+      "loss": 2.0947,
+      "step": 7128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1787267694741008e-05,
+      "loss": 1.8723,
+      "step": 7129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1783313562672993e-05,
+      "loss": 2.0565,
+      "step": 7130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1779359430604985e-05,
+      "loss": 1.8457,
+      "step": 7131
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.177540529853697e-05,
+      "loss": 1.9347,
+      "step": 7132
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1771451166468963e-05,
+      "loss": 1.947,
+      "step": 7133
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.176749703440095e-05,
+      "loss": 2.1924,
+      "step": 7134
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.176354290233294e-05,
+      "loss": 1.7921,
+      "step": 7135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1759588770264926e-05,
+      "loss": 1.8182,
+      "step": 7136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.175563463819692e-05,
+      "loss": 1.7216,
+      "step": 7137
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1751680506128904e-05,
+      "loss": 2.061,
+      "step": 7138
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1747726374060896e-05,
+      "loss": 1.8016,
+      "step": 7139
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.174377224199288e-05,
+      "loss": 1.6591,
+      "step": 7140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1739818109924874e-05,
+      "loss": 1.938,
+      "step": 7141
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.173586397785686e-05,
+      "loss": 2.0308,
+      "step": 7142
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.173190984578885e-05,
+      "loss": 1.6385,
+      "step": 7143
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.172795571372084e-05,
+      "loss": 2.1774,
+      "step": 7144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.172400158165283e-05,
+      "loss": 2.1127,
+      "step": 7145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1720047449584818e-05,
+      "loss": 1.8757,
+      "step": 7146
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1716093317516807e-05,
+      "loss": 2.0566,
+      "step": 7147
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1712139185448796e-05,
+      "loss": 2.1361,
+      "step": 7148
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1708185053380784e-05,
+      "loss": 1.9952,
+      "step": 7149
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1704230921312773e-05,
+      "loss": 1.8113,
+      "step": 7150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1700276789244762e-05,
+      "loss": 1.9127,
+      "step": 7151
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.169632265717675e-05,
+      "loss": 1.8539,
+      "step": 7152
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.169236852510874e-05,
+      "loss": 1.9832,
+      "step": 7153
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.168841439304073e-05,
+      "loss": 1.848,
+      "step": 7154
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1684460260972717e-05,
+      "loss": 1.9424,
+      "step": 7155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1680506128904706e-05,
+      "loss": 1.8699,
+      "step": 7156
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1676551996836695e-05,
+      "loss": 1.9239,
+      "step": 7157
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1672597864768684e-05,
+      "loss": 2.0669,
+      "step": 7158
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1668643732700673e-05,
+      "loss": 1.844,
+      "step": 7159
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.166468960063266e-05,
+      "loss": 2.1049,
+      "step": 7160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1660735468564654e-05,
+      "loss": 1.7342,
+      "step": 7161
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.165678133649664e-05,
+      "loss": 2.1404,
+      "step": 7162
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.165282720442863e-05,
+      "loss": 2.1513,
+      "step": 7163
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1648873072360617e-05,
+      "loss": 1.8587,
+      "step": 7164
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.164491894029261e-05,
+      "loss": 1.6854,
+      "step": 7165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1640964808224594e-05,
+      "loss": 2.056,
+      "step": 7166
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1637010676156587e-05,
+      "loss": 1.8262,
+      "step": 7167
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1633056544088572e-05,
+      "loss": 1.8859,
+      "step": 7168
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1629102412020564e-05,
+      "loss": 1.7756,
+      "step": 7169
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.162514827995255e-05,
+      "loss": 1.9422,
+      "step": 7170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1621194147884542e-05,
+      "loss": 1.8933,
+      "step": 7171
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1617240015816527e-05,
+      "loss": 1.981,
+      "step": 7172
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.161328588374852e-05,
+      "loss": 1.8869,
+      "step": 7173
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1609331751680505e-05,
+      "loss": 1.9091,
+      "step": 7174
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1605377619612497e-05,
+      "loss": 1.7206,
+      "step": 7175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1601423487544483e-05,
+      "loss": 1.9027,
+      "step": 7176
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1597469355476475e-05,
+      "loss": 1.86,
+      "step": 7177
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1593515223408464e-05,
+      "loss": 2.0288,
+      "step": 7178
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1589561091340453e-05,
+      "loss": 1.999,
+      "step": 7179
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.158560695927244e-05,
+      "loss": 1.9455,
+      "step": 7180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.158165282720443e-05,
+      "loss": 1.8839,
+      "step": 7181
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.157769869513642e-05,
+      "loss": 2.0066,
+      "step": 7182
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1573744563068408e-05,
+      "loss": 2.1112,
+      "step": 7183
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1569790431000397e-05,
+      "loss": 1.8977,
+      "step": 7184
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1565836298932386e-05,
+      "loss": 1.8325,
+      "step": 7185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1561882166864374e-05,
+      "loss": 2.0249,
+      "step": 7186
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1557928034796363e-05,
+      "loss": 1.8568,
+      "step": 7187
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1553973902728352e-05,
+      "loss": 2.1164,
+      "step": 7188
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.155001977066034e-05,
+      "loss": 1.9098,
+      "step": 7189
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.154606563859233e-05,
+      "loss": 1.9395,
+      "step": 7190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.154211150652432e-05,
+      "loss": 2.0911,
+      "step": 7191
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1538157374456307e-05,
+      "loss": 1.8648,
+      "step": 7192
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1534203242388296e-05,
+      "loss": 1.8464,
+      "step": 7193
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1530249110320285e-05,
+      "loss": 1.9377,
+      "step": 7194
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1526294978252277e-05,
+      "loss": 1.7414,
+      "step": 7195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1522340846184263e-05,
+      "loss": 1.8263,
+      "step": 7196
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1518386714116255e-05,
+      "loss": 2.1881,
+      "step": 7197
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.151443258204824e-05,
+      "loss": 2.0673,
+      "step": 7198
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1510478449980232e-05,
+      "loss": 2.1027,
+      "step": 7199
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1506524317912218e-05,
+      "loss": 2.2362,
+      "step": 7200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.150257018584421e-05,
+      "loss": 1.961,
+      "step": 7201
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1498616053776196e-05,
+      "loss": 1.898,
+      "step": 7202
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1494661921708188e-05,
+      "loss": 2.0518,
+      "step": 7203
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1490707789640173e-05,
+      "loss": 2.0329,
+      "step": 7204
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1486753657572165e-05,
+      "loss": 1.8704,
+      "step": 7205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.148279952550415e-05,
+      "loss": 1.8661,
+      "step": 7206
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1478845393436143e-05,
+      "loss": 1.8246,
+      "step": 7207
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.147489126136813e-05,
+      "loss": 1.75,
+      "step": 7208
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.147093712930012e-05,
+      "loss": 1.9779,
+      "step": 7209
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1466982997232106e-05,
+      "loss": 1.8926,
+      "step": 7210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.14630288651641e-05,
+      "loss": 2.0189,
+      "step": 7211
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1459074733096084e-05,
+      "loss": 1.9478,
+      "step": 7212
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1455120601028076e-05,
+      "loss": 2.0618,
+      "step": 7213
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1451166468960065e-05,
+      "loss": 1.7881,
+      "step": 7214
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1447212336892054e-05,
+      "loss": 1.686,
+      "step": 7215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1443258204824043e-05,
+      "loss": 2.1138,
+      "step": 7216
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.143930407275603e-05,
+      "loss": 2.2386,
+      "step": 7217
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.143534994068802e-05,
+      "loss": 2.1263,
+      "step": 7218
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.143139580862001e-05,
+      "loss": 1.8644,
+      "step": 7219
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1427441676551998e-05,
+      "loss": 1.7411,
+      "step": 7220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1423487544483987e-05,
+      "loss": 1.8504,
+      "step": 7221
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1419533412415975e-05,
+      "loss": 1.8158,
+      "step": 7222
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1415579280347964e-05,
+      "loss": 2.0952,
+      "step": 7223
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1411625148279953e-05,
+      "loss": 2.0036,
+      "step": 7224
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1407671016211942e-05,
+      "loss": 2.184,
+      "step": 7225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.140371688414393e-05,
+      "loss": 1.9848,
+      "step": 7226
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.139976275207592e-05,
+      "loss": 2.0065,
+      "step": 7227
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.139580862000791e-05,
+      "loss": 1.9995,
+      "step": 7228
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1391854487939897e-05,
+      "loss": 2.0674,
+      "step": 7229
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1387900355871886e-05,
+      "loss": 2.1426,
+      "step": 7230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.138394622380388e-05,
+      "loss": 1.9815,
+      "step": 7231
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1379992091735864e-05,
+      "loss": 1.8862,
+      "step": 7232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1376037959667856e-05,
+      "loss": 2.0536,
+      "step": 7233
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.137208382759984e-05,
+      "loss": 2.0408,
+      "step": 7234
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1368129695531834e-05,
+      "loss": 2.1298,
+      "step": 7235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.136417556346382e-05,
+      "loss": 2.1806,
+      "step": 7236
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.136022143139581e-05,
+      "loss": 1.8056,
+      "step": 7237
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1356267299327797e-05,
+      "loss": 1.9829,
+      "step": 7238
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.135231316725979e-05,
+      "loss": 2.0623,
+      "step": 7239
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1348359035191774e-05,
+      "loss": 2.0759,
+      "step": 7240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1344404903123767e-05,
+      "loss": 2.2214,
+      "step": 7241
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1340450771055752e-05,
+      "loss": 1.9292,
+      "step": 7242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1336496638987744e-05,
+      "loss": 2.0283,
+      "step": 7243
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.133254250691973e-05,
+      "loss": 1.8478,
+      "step": 7244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1328588374851722e-05,
+      "loss": 2.0901,
+      "step": 7245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1324634242783707e-05,
+      "loss": 1.9511,
+      "step": 7246
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.13206801107157e-05,
+      "loss": 1.8282,
+      "step": 7247
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.131672597864769e-05,
+      "loss": 1.8093,
+      "step": 7248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1312771846579677e-05,
+      "loss": 1.9424,
+      "step": 7249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1308817714511666e-05,
+      "loss": 1.9577,
+      "step": 7250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1304863582443655e-05,
+      "loss": 1.8707,
+      "step": 7251
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1300909450375644e-05,
+      "loss": 2.13,
+      "step": 7252
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1296955318307633e-05,
+      "loss": 2.0149,
+      "step": 7253
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.129300118623962e-05,
+      "loss": 2.0882,
+      "step": 7254
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.128904705417161e-05,
+      "loss": 1.9917,
+      "step": 7255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.12850929221036e-05,
+      "loss": 1.9482,
+      "step": 7256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1281138790035588e-05,
+      "loss": 2.1791,
+      "step": 7257
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1277184657967577e-05,
+      "loss": 1.7376,
+      "step": 7258
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1273230525899565e-05,
+      "loss": 1.7404,
+      "step": 7259
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1269276393831554e-05,
+      "loss": 1.8388,
+      "step": 7260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1265322261763543e-05,
+      "loss": 1.7414,
+      "step": 7261
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1261368129695532e-05,
+      "loss": 2.0275,
+      "step": 7262
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.125741399762752e-05,
+      "loss": 2.1771,
+      "step": 7263
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.125345986555951e-05,
+      "loss": 1.9197,
+      "step": 7264
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1249505733491502e-05,
+      "loss": 1.7517,
+      "step": 7265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1245551601423487e-05,
+      "loss": 1.6811,
+      "step": 7266
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.124159746935548e-05,
+      "loss": 1.8892,
+      "step": 7267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1237643337287465e-05,
+      "loss": 1.9577,
+      "step": 7268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1233689205219457e-05,
+      "loss": 2.0309,
+      "step": 7269
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1229735073151443e-05,
+      "loss": 1.997,
+      "step": 7270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1225780941083435e-05,
+      "loss": 1.7871,
+      "step": 7271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.122182680901542e-05,
+      "loss": 1.9954,
+      "step": 7272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1217872676947412e-05,
+      "loss": 1.8532,
+      "step": 7273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1213918544879398e-05,
+      "loss": 2.0752,
+      "step": 7274
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.120996441281139e-05,
+      "loss": 1.9544,
+      "step": 7275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1206010280743376e-05,
+      "loss": 1.9482,
+      "step": 7276
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1202056148675368e-05,
+      "loss": 1.8656,
+      "step": 7277
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1198102016607353e-05,
+      "loss": 2.1163,
+      "step": 7278
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1194147884539345e-05,
+      "loss": 1.9481,
+      "step": 7279
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.119019375247133e-05,
+      "loss": 1.842,
+      "step": 7280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1186239620403323e-05,
+      "loss": 2.009,
+      "step": 7281
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1182285488335312e-05,
+      "loss": 1.9159,
+      "step": 7282
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.11783313562673e-05,
+      "loss": 2.1959,
+      "step": 7283
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.117437722419929e-05,
+      "loss": 1.715,
+      "step": 7284
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.117042309213128e-05,
+      "loss": 2.0115,
+      "step": 7285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1166468960063267e-05,
+      "loss": 1.8716,
+      "step": 7286
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1162514827995256e-05,
+      "loss": 1.8446,
+      "step": 7287
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1158560695927245e-05,
+      "loss": 1.7842,
+      "step": 7288
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1154606563859234e-05,
+      "loss": 2.1907,
+      "step": 7289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1150652431791222e-05,
+      "loss": 2.0439,
+      "step": 7290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.114669829972321e-05,
+      "loss": 2.0906,
+      "step": 7291
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.11427441676552e-05,
+      "loss": 1.9801,
+      "step": 7292
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.113879003558719e-05,
+      "loss": 1.4992,
+      "step": 7293
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1134835903519178e-05,
+      "loss": 1.9337,
+      "step": 7294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1130881771451167e-05,
+      "loss": 2.0133,
+      "step": 7295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1126927639383155e-05,
+      "loss": 2.1489,
+      "step": 7296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1122973507315144e-05,
+      "loss": 1.9527,
+      "step": 7297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1119019375247133e-05,
+      "loss": 2.0949,
+      "step": 7298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1115065243179122e-05,
+      "loss": 2.2135,
+      "step": 7299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.111111111111111e-05,
+      "loss": 2.0725,
+      "step": 7300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1107156979043103e-05,
+      "loss": 1.9008,
+      "step": 7301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.110320284697509e-05,
+      "loss": 2.0774,
+      "step": 7302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.109924871490708e-05,
+      "loss": 2.1011,
+      "step": 7303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1095294582839066e-05,
+      "loss": 2.1133,
+      "step": 7304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1091340450771058e-05,
+      "loss": 2.0133,
+      "step": 7305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1087386318703044e-05,
+      "loss": 2.061,
+      "step": 7306
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1083432186635036e-05,
+      "loss": 1.9698,
+      "step": 7307
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.107947805456702e-05,
+      "loss": 2.0355,
+      "step": 7308
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1075523922499014e-05,
+      "loss": 1.9702,
+      "step": 7309
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1071569790431e-05,
+      "loss": 1.7936,
+      "step": 7310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.106761565836299e-05,
+      "loss": 1.9125,
+      "step": 7311
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1063661526294977e-05,
+      "loss": 1.946,
+      "step": 7312
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.105970739422697e-05,
+      "loss": 1.8904,
+      "step": 7313
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1055753262158954e-05,
+      "loss": 2.048,
+      "step": 7314
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1051799130090947e-05,
+      "loss": 2.0385,
+      "step": 7315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1047844998022932e-05,
+      "loss": 1.8967,
+      "step": 7316
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1043890865954924e-05,
+      "loss": 1.9249,
+      "step": 7317
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1039936733886913e-05,
+      "loss": 2.0232,
+      "step": 7318
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1035982601818902e-05,
+      "loss": 2.0535,
+      "step": 7319
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.103202846975089e-05,
+      "loss": 2.0289,
+      "step": 7320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.102807433768288e-05,
+      "loss": 2.0026,
+      "step": 7321
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.102412020561487e-05,
+      "loss": 1.9139,
+      "step": 7322
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1020166073546857e-05,
+      "loss": 1.9061,
+      "step": 7323
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1016211941478846e-05,
+      "loss": 1.9342,
+      "step": 7324
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1012257809410835e-05,
+      "loss": 1.9593,
+      "step": 7325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1008303677342824e-05,
+      "loss": 1.9468,
+      "step": 7326
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1004349545274812e-05,
+      "loss": 1.7882,
+      "step": 7327
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.10003954132068e-05,
+      "loss": 1.8187,
+      "step": 7328
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.099644128113879e-05,
+      "loss": 2.0455,
+      "step": 7329
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.099248714907078e-05,
+      "loss": 2.1982,
+      "step": 7330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0988533017002768e-05,
+      "loss": 2.0949,
+      "step": 7331
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0984578884934757e-05,
+      "loss": 1.863,
+      "step": 7332
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0980624752866745e-05,
+      "loss": 2.148,
+      "step": 7333
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0976670620798734e-05,
+      "loss": 1.933,
+      "step": 7334
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0972716488730726e-05,
+      "loss": 1.8922,
+      "step": 7335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0968762356662712e-05,
+      "loss": 1.9948,
+      "step": 7336
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0964808224594704e-05,
+      "loss": 1.7939,
+      "step": 7337
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.096085409252669e-05,
+      "loss": 1.8502,
+      "step": 7338
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0956899960458682e-05,
+      "loss": 2.0712,
+      "step": 7339
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0952945828390667e-05,
+      "loss": 2.0644,
+      "step": 7340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.094899169632266e-05,
+      "loss": 1.9551,
+      "step": 7341
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0945037564254645e-05,
+      "loss": 2.0085,
+      "step": 7342
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0941083432186637e-05,
+      "loss": 1.9763,
+      "step": 7343
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0937129300118623e-05,
+      "loss": 2.1525,
+      "step": 7344
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0933175168050615e-05,
+      "loss": 2.1113,
+      "step": 7345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.09292210359826e-05,
+      "loss": 2.0787,
+      "step": 7346
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0925266903914592e-05,
+      "loss": 2.081,
+      "step": 7347
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.092131277184658e-05,
+      "loss": 2.0237,
+      "step": 7348
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.091735863977857e-05,
+      "loss": 1.7446,
+      "step": 7349
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.091340450771056e-05,
+      "loss": 1.803,
+      "step": 7350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0909450375642548e-05,
+      "loss": 1.7387,
+      "step": 7351
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0905496243574537e-05,
+      "loss": 2.0978,
+      "step": 7352
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0901542111506525e-05,
+      "loss": 2.0972,
+      "step": 7353
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0897587979438514e-05,
+      "loss": 2.1361,
+      "step": 7354
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0893633847370503e-05,
+      "loss": 1.9728,
+      "step": 7355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0889679715302492e-05,
+      "loss": 1.9555,
+      "step": 7356
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.088572558323448e-05,
+      "loss": 1.622,
+      "step": 7357
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.088177145116647e-05,
+      "loss": 1.905,
+      "step": 7358
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.087781731909846e-05,
+      "loss": 1.9187,
+      "step": 7359
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0873863187030447e-05,
+      "loss": 2.0361,
+      "step": 7360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0869909054962436e-05,
+      "loss": 1.9286,
+      "step": 7361
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0865954922894425e-05,
+      "loss": 2.0525,
+      "step": 7362
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0862000790826414e-05,
+      "loss": 1.943,
+      "step": 7363
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0858046658758402e-05,
+      "loss": 1.9928,
+      "step": 7364
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.085409252669039e-05,
+      "loss": 1.6971,
+      "step": 7365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.085013839462238e-05,
+      "loss": 1.951,
+      "step": 7366
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.084618426255437e-05,
+      "loss": 1.8905,
+      "step": 7367
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0842230130486358e-05,
+      "loss": 1.8971,
+      "step": 7368
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.083827599841835e-05,
+      "loss": 2.1227,
+      "step": 7369
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0834321866350335e-05,
+      "loss": 1.9198,
+      "step": 7370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0830367734282328e-05,
+      "loss": 2.0309,
+      "step": 7371
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0826413602214313e-05,
+      "loss": 2.2874,
+      "step": 7372
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0822459470146305e-05,
+      "loss": 2.0725,
+      "step": 7373
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.081850533807829e-05,
+      "loss": 1.7871,
+      "step": 7374
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0814551206010283e-05,
+      "loss": 2.1534,
+      "step": 7375
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0810597073942272e-05,
+      "loss": 2.1731,
+      "step": 7376
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.080664294187426e-05,
+      "loss": 1.9068,
+      "step": 7377
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.080268880980625e-05,
+      "loss": 1.9882,
+      "step": 7378
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0798734677738238e-05,
+      "loss": 1.7808,
+      "step": 7379
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0794780545670227e-05,
+      "loss": 2.0275,
+      "step": 7380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0790826413602216e-05,
+      "loss": 1.7097,
+      "step": 7381
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0786872281534205e-05,
+      "loss": 1.7544,
+      "step": 7382
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0782918149466194e-05,
+      "loss": 1.8103,
+      "step": 7383
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0778964017398182e-05,
+      "loss": 2.0264,
+      "step": 7384
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.077500988533017e-05,
+      "loss": 1.8181,
+      "step": 7385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.077105575326216e-05,
+      "loss": 2.0714,
+      "step": 7386
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.076710162119415e-05,
+      "loss": 2.001,
+      "step": 7387
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0763147489126138e-05,
+      "loss": 1.9323,
+      "step": 7388
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0759193357058127e-05,
+      "loss": 1.9521,
+      "step": 7389
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0755239224990115e-05,
+      "loss": 1.9593,
+      "step": 7390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0751285092922104e-05,
+      "loss": 1.8468,
+      "step": 7391
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0747330960854093e-05,
+      "loss": 1.994,
+      "step": 7392
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0743376828786082e-05,
+      "loss": 2.1731,
+      "step": 7393
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.073942269671807e-05,
+      "loss": 1.8036,
+      "step": 7394
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.073546856465006e-05,
+      "loss": 1.8979,
+      "step": 7395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.073151443258205e-05,
+      "loss": 2.1589,
+      "step": 7396
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.0727560300514037e-05,
+      "loss": 2.0508,
+      "step": 7397
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0723606168446026e-05,
+      "loss": 1.9793,
+      "step": 7398
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0719652036378015e-05,
+      "loss": 1.6883,
+      "step": 7399
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0715697904310004e-05,
+      "loss": 1.9696,
+      "step": 7400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0711743772241992e-05,
+      "loss": 1.862,
+      "step": 7401
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.070778964017398e-05,
+      "loss": 1.9147,
+      "step": 7402
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.070383550810597e-05,
+      "loss": 1.8912,
+      "step": 7403
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.069988137603796e-05,
+      "loss": 1.7943,
+      "step": 7404
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.069592724396995e-05,
+      "loss": 1.9505,
+      "step": 7405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.069197311190194e-05,
+      "loss": 1.9476,
+      "step": 7406
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.068801897983393e-05,
+      "loss": 1.9715,
+      "step": 7407
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0684064847765918e-05,
+      "loss": 1.6831,
+      "step": 7408
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0680110715697906e-05,
+      "loss": 2.0417,
+      "step": 7409
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0676156583629895e-05,
+      "loss": 1.8778,
+      "step": 7410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0672202451561884e-05,
+      "loss": 2.0235,
+      "step": 7411
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0668248319493873e-05,
+      "loss": 1.94,
+      "step": 7412
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0664294187425862e-05,
+      "loss": 2.0898,
+      "step": 7413
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.066034005535785e-05,
+      "loss": 2.1273,
+      "step": 7414
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.065638592328984e-05,
+      "loss": 1.945,
+      "step": 7415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0652431791221828e-05,
+      "loss": 1.9482,
+      "step": 7416
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0648477659153817e-05,
+      "loss": 1.6894,
+      "step": 7417
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0644523527085806e-05,
+      "loss": 2.1407,
+      "step": 7418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0640569395017795e-05,
+      "loss": 2.07,
+      "step": 7419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0636615262949784e-05,
+      "loss": 1.9485,
+      "step": 7420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0632661130881772e-05,
+      "loss": 2.0509,
+      "step": 7421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.062870699881376e-05,
+      "loss": 2.0437,
+      "step": 7422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.062475286674575e-05,
+      "loss": 1.8967,
+      "step": 7423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.062079873467774e-05,
+      "loss": 2.003,
+      "step": 7424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0616844602609728e-05,
+      "loss": 1.8976,
+      "step": 7425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0612890470541717e-05,
+      "loss": 1.7984,
+      "step": 7426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0608936338473705e-05,
+      "loss": 1.958,
+      "step": 7427
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0604982206405694e-05,
+      "loss": 2.2027,
+      "step": 7428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0601028074337683e-05,
+      "loss": 1.9644,
+      "step": 7429
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0597073942269672e-05,
+      "loss": 2.1654,
+      "step": 7430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.059311981020166e-05,
+      "loss": 1.8222,
+      "step": 7431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.058916567813365e-05,
+      "loss": 2.1334,
+      "step": 7432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0585211546065638e-05,
+      "loss": 1.9487,
+      "step": 7433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.058125741399763e-05,
+      "loss": 1.9066,
+      "step": 7434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0577303281929616e-05,
+      "loss": 1.9081,
+      "step": 7435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0573349149861608e-05,
+      "loss": 2.0814,
+      "step": 7436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0569395017793594e-05,
+      "loss": 2.096,
+      "step": 7437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0565440885725586e-05,
+      "loss": 1.773,
+      "step": 7438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0561486753657575e-05,
+      "loss": 2.0287,
+      "step": 7439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0557532621589563e-05,
+      "loss": 2.2294,
+      "step": 7440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0553578489521552e-05,
+      "loss": 1.963,
+      "step": 7441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.054962435745354e-05,
+      "loss": 1.8676,
+      "step": 7442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.054567022538553e-05,
+      "loss": 2.1812,
+      "step": 7443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.054171609331752e-05,
+      "loss": 1.9894,
+      "step": 7444
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0537761961249508e-05,
+      "loss": 2.0375,
+      "step": 7445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0533807829181496e-05,
+      "loss": 2.0089,
+      "step": 7446
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0529853697113485e-05,
+      "loss": 2.1362,
+      "step": 7447
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0525899565045474e-05,
+      "loss": 2.0571,
+      "step": 7448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0521945432977463e-05,
+      "loss": 2.3473,
+      "step": 7449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0517991300909452e-05,
+      "loss": 1.9627,
+      "step": 7450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.051403716884144e-05,
+      "loss": 1.7958,
+      "step": 7451
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.051008303677343e-05,
+      "loss": 1.9931,
+      "step": 7452
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0506128904705418e-05,
+      "loss": 2.0704,
+      "step": 7453
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0502174772637407e-05,
+      "loss": 1.8941,
+      "step": 7454
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0498220640569396e-05,
+      "loss": 2.0206,
+      "step": 7455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0494266508501385e-05,
+      "loss": 1.8027,
+      "step": 7456
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0490312376433374e-05,
+      "loss": 1.7102,
+      "step": 7457
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0486358244365362e-05,
+      "loss": 1.5828,
+      "step": 7458
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.048240411229735e-05,
+      "loss": 1.8914,
+      "step": 7459
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.047844998022934e-05,
+      "loss": 2.2974,
+      "step": 7460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.047449584816133e-05,
+      "loss": 1.8082,
+      "step": 7461
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.047054171609332e-05,
+      "loss": 2.0112,
+      "step": 7462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0466587584025307e-05,
+      "loss": 1.9628,
+      "step": 7463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.04626334519573e-05,
+      "loss": 1.7921,
+      "step": 7464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0458679319889284e-05,
+      "loss": 2.0776,
+      "step": 7465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0454725187821276e-05,
+      "loss": 1.8422,
+      "step": 7466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0450771055753262e-05,
+      "loss": 2.0672,
+      "step": 7467
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0446816923685254e-05,
+      "loss": 2.3553,
+      "step": 7468
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.044286279161724e-05,
+      "loss": 2.1564,
+      "step": 7469
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.043890865954923e-05,
+      "loss": 2.0277,
+      "step": 7470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0434954527481217e-05,
+      "loss": 1.7908,
+      "step": 7471
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.043100039541321e-05,
+      "loss": 2.081,
+      "step": 7472
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0427046263345195e-05,
+      "loss": 1.9233,
+      "step": 7473
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0423092131277187e-05,
+      "loss": 1.9832,
+      "step": 7474
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0419137999209176e-05,
+      "loss": 2.0115,
+      "step": 7475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0415183867141165e-05,
+      "loss": 1.8614,
+      "step": 7476
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0411229735073153e-05,
+      "loss": 1.9817,
+      "step": 7477
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0407275603005142e-05,
+      "loss": 1.7546,
+      "step": 7478
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.040332147093713e-05,
+      "loss": 1.6009,
+      "step": 7479
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.039936733886912e-05,
+      "loss": 1.9943,
+      "step": 7480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.039541320680111e-05,
+      "loss": 1.9848,
+      "step": 7481
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0391459074733098e-05,
+      "loss": 2.0357,
+      "step": 7482
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0387504942665086e-05,
+      "loss": 1.8147,
+      "step": 7483
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0383550810597075e-05,
+      "loss": 2.063,
+      "step": 7484
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0379596678529064e-05,
+      "loss": 2.0282,
+      "step": 7485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0375642546461053e-05,
+      "loss": 2.2274,
+      "step": 7486
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0371688414393042e-05,
+      "loss": 1.8997,
+      "step": 7487
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.036773428232503e-05,
+      "loss": 1.7998,
+      "step": 7488
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.036378015025702e-05,
+      "loss": 2.0424,
+      "step": 7489
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0359826018189008e-05,
+      "loss": 2.0637,
+      "step": 7490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0355871886120997e-05,
+      "loss": 2.0508,
+      "step": 7491
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.035191775405299e-05,
+      "loss": 1.8486,
+      "step": 7492
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0347963621984975e-05,
+      "loss": 1.876,
+      "step": 7493
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0344009489916967e-05,
+      "loss": 2.1319,
+      "step": 7494
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0340055357848952e-05,
+      "loss": 2.0706,
+      "step": 7495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0336101225780945e-05,
+      "loss": 1.792,
+      "step": 7496
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.033214709371293e-05,
+      "loss": 2.1423,
+      "step": 7497
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0328192961644922e-05,
+      "loss": 2.011,
+      "step": 7498
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0324238829576908e-05,
+      "loss": 2.1316,
+      "step": 7499
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.03202846975089e-05,
+      "loss": 1.9415,
+      "step": 7500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0316330565440885e-05,
+      "loss": 1.8881,
+      "step": 7501
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0312376433372878e-05,
+      "loss": 1.9918,
+      "step": 7502
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0308422301304863e-05,
+      "loss": 1.8031,
+      "step": 7503
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0304468169236855e-05,
+      "loss": 2.0712,
+      "step": 7504
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.030051403716884e-05,
+      "loss": 1.8876,
+      "step": 7505
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0296559905100833e-05,
+      "loss": 1.945,
+      "step": 7506
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0292605773032818e-05,
+      "loss": 1.7329,
+      "step": 7507
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.028865164096481e-05,
+      "loss": 1.7709,
+      "step": 7508
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.02846975088968e-05,
+      "loss": 2.0509,
+      "step": 7509
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0280743376828788e-05,
+      "loss": 1.888,
+      "step": 7510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0276789244760777e-05,
+      "loss": 1.8835,
+      "step": 7511
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0272835112692766e-05,
+      "loss": 1.7609,
+      "step": 7512
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0268880980624755e-05,
+      "loss": 1.9589,
+      "step": 7513
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0264926848556743e-05,
+      "loss": 2.1302,
+      "step": 7514
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0260972716488732e-05,
+      "loss": 1.9865,
+      "step": 7515
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.025701858442072e-05,
+      "loss": 1.9337,
+      "step": 7516
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.025306445235271e-05,
+      "loss": 1.8442,
+      "step": 7517
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.02491103202847e-05,
+      "loss": 2.0701,
+      "step": 7518
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0245156188216688e-05,
+      "loss": 1.9401,
+      "step": 7519
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0241202056148676e-05,
+      "loss": 1.8916,
+      "step": 7520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0237247924080665e-05,
+      "loss": 2.1552,
+      "step": 7521
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0233293792012654e-05,
+      "loss": 1.6992,
+      "step": 7522
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0229339659944643e-05,
+      "loss": 1.897,
+      "step": 7523
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.0225385527876632e-05,
+      "loss": 2.0083,
+      "step": 7524
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.022143139580862e-05,
+      "loss": 1.8838,
+      "step": 7525
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0217477263740613e-05,
+      "loss": 1.8354,
+      "step": 7526
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0213523131672598e-05,
+      "loss": 2.0489,
+      "step": 7527
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.020956899960459e-05,
+      "loss": 1.8221,
+      "step": 7528
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0205614867536576e-05,
+      "loss": 2.0492,
+      "step": 7529
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0201660735468568e-05,
+      "loss": 1.8993,
+      "step": 7530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0197706603400554e-05,
+      "loss": 2.2549,
+      "step": 7531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0193752471332546e-05,
+      "loss": 1.9608,
+      "step": 7532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.018979833926453e-05,
+      "loss": 1.9692,
+      "step": 7533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0185844207196523e-05,
+      "loss": 1.978,
+      "step": 7534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.018189007512851e-05,
+      "loss": 1.8726,
+      "step": 7535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.01779359430605e-05,
+      "loss": 1.7655,
+      "step": 7536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0173981810992486e-05,
+      "loss": 2.0314,
+      "step": 7537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.017002767892448e-05,
+      "loss": 1.8525,
+      "step": 7538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0166073546856464e-05,
+      "loss": 1.9516,
+      "step": 7539
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0162119414788456e-05,
+      "loss": 1.7609,
+      "step": 7540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0158165282720442e-05,
+      "loss": 1.8581,
+      "step": 7541
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0154211150652434e-05,
+      "loss": 2.0206,
+      "step": 7542
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0150257018584423e-05,
+      "loss": 1.8356,
+      "step": 7543
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.014630288651641e-05,
+      "loss": 1.7638,
+      "step": 7544
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.01423487544484e-05,
+      "loss": 1.9636,
+      "step": 7545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.013839462238039e-05,
+      "loss": 2.0951,
+      "step": 7546
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0134440490312378e-05,
+      "loss": 1.8577,
+      "step": 7547
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0130486358244367e-05,
+      "loss": 1.8868,
+      "step": 7548
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0126532226176356e-05,
+      "loss": 2.0732,
+      "step": 7549
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0122578094108345e-05,
+      "loss": 2.0317,
+      "step": 7550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0118623962040333e-05,
+      "loss": 1.8872,
+      "step": 7551
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0114669829972322e-05,
+      "loss": 1.5931,
+      "step": 7552
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.011071569790431e-05,
+      "loss": 2.054,
+      "step": 7553
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.01067615658363e-05,
+      "loss": 1.7548,
+      "step": 7554
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.010280743376829e-05,
+      "loss": 1.9387,
+      "step": 7555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0098853301700278e-05,
+      "loss": 1.946,
+      "step": 7556
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0094899169632266e-05,
+      "loss": 1.9303,
+      "step": 7557
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0090945037564255e-05,
+      "loss": 1.8491,
+      "step": 7558
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0086990905496244e-05,
+      "loss": 2.1868,
+      "step": 7559
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0083036773428233e-05,
+      "loss": 2.2546,
+      "step": 7560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0079082641360222e-05,
+      "loss": 1.9721,
+      "step": 7561
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0075128509292214e-05,
+      "loss": 2.0519,
+      "step": 7562
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.00711743772242e-05,
+      "loss": 2.0857,
+      "step": 7563
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.006722024515619e-05,
+      "loss": 1.8391,
+      "step": 7564
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0063266113088177e-05,
+      "loss": 1.865,
+      "step": 7565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.005931198102017e-05,
+      "loss": 1.9025,
+      "step": 7566
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0055357848952155e-05,
+      "loss": 1.9629,
+      "step": 7567
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0051403716884147e-05,
+      "loss": 1.8505,
+      "step": 7568
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0047449584816132e-05,
+      "loss": 1.8888,
+      "step": 7569
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0043495452748125e-05,
+      "loss": 2.0332,
+      "step": 7570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.003954132068011e-05,
+      "loss": 1.9234,
+      "step": 7571
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0035587188612102e-05,
+      "loss": 2.3292,
+      "step": 7572
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0031633056544088e-05,
+      "loss": 2.0697,
+      "step": 7573
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.002767892447608e-05,
+      "loss": 2.1435,
+      "step": 7574
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0023724792408065e-05,
+      "loss": 1.6766,
+      "step": 7575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0019770660340058e-05,
+      "loss": 2.1868,
+      "step": 7576
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0015816528272043e-05,
+      "loss": 1.735,
+      "step": 7577
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0011862396204035e-05,
+      "loss": 2.0135,
+      "step": 7578
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0007908264136024e-05,
+      "loss": 1.8217,
+      "step": 7579
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0003954132068013e-05,
+      "loss": 1.8933,
+      "step": 7580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2e-05,
+      "loss": 1.9577,
+      "step": 7581
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999604586793199e-05,
+      "loss": 1.7699,
+      "step": 7582
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999209173586398e-05,
+      "loss": 1.8584,
+      "step": 7583
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9988137603795968e-05,
+      "loss": 2.1259,
+      "step": 7584
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9984183471727957e-05,
+      "loss": 1.8712,
+      "step": 7585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9980229339659946e-05,
+      "loss": 1.9886,
+      "step": 7586
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9976275207591935e-05,
+      "loss": 1.6769,
+      "step": 7587
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9972321075523923e-05,
+      "loss": 1.8224,
+      "step": 7588
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9968366943455912e-05,
+      "loss": 1.9871,
+      "step": 7589
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.99644128113879e-05,
+      "loss": 1.866,
+      "step": 7590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.996045867931989e-05,
+      "loss": 1.8857,
+      "step": 7591
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.995650454725188e-05,
+      "loss": 1.6477,
+      "step": 7592
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9952550415183868e-05,
+      "loss": 1.9628,
+      "step": 7593
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9948596283115856e-05,
+      "loss": 1.8559,
+      "step": 7594
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9944642151047845e-05,
+      "loss": 1.9105,
+      "step": 7595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9940688018979837e-05,
+      "loss": 1.8401,
+      "step": 7596
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9936733886911823e-05,
+      "loss": 1.9583,
+      "step": 7597
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9932779754843815e-05,
+      "loss": 1.9961,
+      "step": 7598
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.99288256227758e-05,
+      "loss": 1.8107,
+      "step": 7599
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9924871490707793e-05,
+      "loss": 2.2654,
+      "step": 7600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9920917358639778e-05,
+      "loss": 1.872,
+      "step": 7601
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.991696322657177e-05,
+      "loss": 1.921,
+      "step": 7602
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9913009094503756e-05,
+      "loss": 1.8837,
+      "step": 7603
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9909054962435748e-05,
+      "loss": 1.9858,
+      "step": 7604
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9905100830367733e-05,
+      "loss": 2.314,
+      "step": 7605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9901146698299726e-05,
+      "loss": 1.8915,
+      "step": 7606
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.989719256623171e-05,
+      "loss": 2.013,
+      "step": 7607
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9893238434163703e-05,
+      "loss": 2.1898,
+      "step": 7608
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.988928430209569e-05,
+      "loss": 2.1042,
+      "step": 7609
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.988533017002768e-05,
+      "loss": 1.8585,
+      "step": 7610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9881376037959666e-05,
+      "loss": 1.9315,
+      "step": 7611
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.987742190589166e-05,
+      "loss": 1.7599,
+      "step": 7612
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9873467773823647e-05,
+      "loss": 1.9246,
+      "step": 7613
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9869513641755636e-05,
+      "loss": 1.9465,
+      "step": 7614
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9865559509687625e-05,
+      "loss": 2.002,
+      "step": 7615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9861605377619614e-05,
+      "loss": 1.7494,
+      "step": 7616
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9857651245551603e-05,
+      "loss": 1.7048,
+      "step": 7617
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.985369711348359e-05,
+      "loss": 1.8109,
+      "step": 7618
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.984974298141558e-05,
+      "loss": 1.6748,
+      "step": 7619
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.984578884934757e-05,
+      "loss": 1.8794,
+      "step": 7620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9841834717279558e-05,
+      "loss": 1.7196,
+      "step": 7621
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9837880585211547e-05,
+      "loss": 2.0847,
+      "step": 7622
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9833926453143536e-05,
+      "loss": 1.8867,
+      "step": 7623
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9829972321075525e-05,
+      "loss": 1.9607,
+      "step": 7624
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9826018189007513e-05,
+      "loss": 2.0535,
+      "step": 7625
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9822064056939502e-05,
+      "loss": 1.8157,
+      "step": 7626
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.981810992487149e-05,
+      "loss": 2.0223,
+      "step": 7627
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.981415579280348e-05,
+      "loss": 2.0618,
+      "step": 7628
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.981020166073547e-05,
+      "loss": 1.9529,
+      "step": 7629
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.980624752866746e-05,
+      "loss": 1.7517,
+      "step": 7630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9802293396599446e-05,
+      "loss": 1.9967,
+      "step": 7631
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.979833926453144e-05,
+      "loss": 1.9152,
+      "step": 7632
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9794385132463424e-05,
+      "loss": 1.9631,
+      "step": 7633
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9790431000395416e-05,
+      "loss": 1.8406,
+      "step": 7634
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.97864768683274e-05,
+      "loss": 2.15,
+      "step": 7635
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9782522736259394e-05,
+      "loss": 1.8237,
+      "step": 7636
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.977856860419138e-05,
+      "loss": 1.928,
+      "step": 7637
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.977461447212337e-05,
+      "loss": 1.9041,
+      "step": 7638
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9770660340055357e-05,
+      "loss": 2.095,
+      "step": 7639
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.976670620798735e-05,
+      "loss": 1.6934,
+      "step": 7640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9762752075919335e-05,
+      "loss": 2.1616,
+      "step": 7641
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9758797943851327e-05,
+      "loss": 1.962,
+      "step": 7642
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9754843811783312e-05,
+      "loss": 1.9969,
+      "step": 7643
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9750889679715305e-05,
+      "loss": 2.058,
+      "step": 7644
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.974693554764729e-05,
+      "loss": 1.9172,
+      "step": 7645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9742981415579282e-05,
+      "loss": 2.1062,
+      "step": 7646
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9739027283511268e-05,
+      "loss": 1.8826,
+      "step": 7647
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.973507315144326e-05,
+      "loss": 2.039,
+      "step": 7648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.973111901937525e-05,
+      "loss": 1.8892,
+      "step": 7649
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9727164887307237e-05,
+      "loss": 1.7759,
+      "step": 7650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9723210755239226e-05,
+      "loss": 2.1858,
+      "step": 7651
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9719256623171215e-05,
+      "loss": 1.9372,
+      "step": 7652
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9715302491103204e-05,
+      "loss": 1.9592,
+      "step": 7653
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9711348359035193e-05,
+      "loss": 1.8052,
+      "step": 7654
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.970739422696718e-05,
+      "loss": 1.8809,
+      "step": 7655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.970344009489917e-05,
+      "loss": 2.0174,
+      "step": 7656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.969948596283116e-05,
+      "loss": 1.9236,
+      "step": 7657
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9695531830763148e-05,
+      "loss": 2.0361,
+      "step": 7658
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9691577698695137e-05,
+      "loss": 1.7684,
+      "step": 7659
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9687623566627126e-05,
+      "loss": 1.7807,
+      "step": 7660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9683669434559115e-05,
+      "loss": 1.8723,
+      "step": 7661
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9679715302491103e-05,
+      "loss": 2.2131,
+      "step": 7662
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9675761170423092e-05,
+      "loss": 1.7912,
+      "step": 7663
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.967180703835508e-05,
+      "loss": 1.9406,
+      "step": 7664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.966785290628707e-05,
+      "loss": 1.677,
+      "step": 7665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9663898774219062e-05,
+      "loss": 1.9765,
+      "step": 7666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9659944642151048e-05,
+      "loss": 2.0369,
+      "step": 7667
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.965599051008304e-05,
+      "loss": 1.8282,
+      "step": 7668
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9652036378015025e-05,
+      "loss": 1.9159,
+      "step": 7669
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9648082245947017e-05,
+      "loss": 2.0872,
+      "step": 7670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9644128113879003e-05,
+      "loss": 1.7018,
+      "step": 7671
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9640173981810995e-05,
+      "loss": 2.0762,
+      "step": 7672
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.963621984974298e-05,
+      "loss": 2.1054,
+      "step": 7673
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9632265717674973e-05,
+      "loss": 1.845,
+      "step": 7674
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9628311585606958e-05,
+      "loss": 1.8946,
+      "step": 7675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.962435745353895e-05,
+      "loss": 1.8909,
+      "step": 7676
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9620403321470936e-05,
+      "loss": 2.0457,
+      "step": 7677
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9616449189402928e-05,
+      "loss": 2.0804,
+      "step": 7678
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9612495057334913e-05,
+      "loss": 1.6931,
+      "step": 7679
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9608540925266906e-05,
+      "loss": 1.9331,
+      "step": 7680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.960458679319889e-05,
+      "loss": 2.0017,
+      "step": 7681
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9600632661130883e-05,
+      "loss": 1.7598,
+      "step": 7682
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9596678529062872e-05,
+      "loss": 1.7309,
+      "step": 7683
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.959272439699486e-05,
+      "loss": 1.8983,
+      "step": 7684
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.958877026492685e-05,
+      "loss": 1.9412,
+      "step": 7685
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.958481613285884e-05,
+      "loss": 2.0806,
+      "step": 7686
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9580862000790827e-05,
+      "loss": 1.9994,
+      "step": 7687
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9576907868722816e-05,
+      "loss": 1.8076,
+      "step": 7688
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9572953736654805e-05,
+      "loss": 1.832,
+      "step": 7689
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9568999604586794e-05,
+      "loss": 1.8518,
+      "step": 7690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9565045472518783e-05,
+      "loss": 1.9901,
+      "step": 7691
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956109134045077e-05,
+      "loss": 1.7612,
+      "step": 7692
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955713720838276e-05,
+      "loss": 1.8056,
+      "step": 7693
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955318307631475e-05,
+      "loss": 1.9565,
+      "step": 7694
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9549228944246738e-05,
+      "loss": 2.171,
+      "step": 7695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9545274812178727e-05,
+      "loss": 2.0214,
+      "step": 7696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9541320680110716e-05,
+      "loss": 1.7516,
+      "step": 7697
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9537366548042705e-05,
+      "loss": 2.0611,
+      "step": 7698
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9533412415974693e-05,
+      "loss": 2.2378,
+      "step": 7699
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9529458283906686e-05,
+      "loss": 1.901,
+      "step": 7700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.952550415183867e-05,
+      "loss": 1.9196,
+      "step": 7701
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9521550019770663e-05,
+      "loss": 1.9796,
+      "step": 7702
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.951759588770265e-05,
+      "loss": 2.1329,
+      "step": 7703
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.951364175563464e-05,
+      "loss": 1.7835,
+      "step": 7704
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9509687623566626e-05,
+      "loss": 1.7047,
+      "step": 7705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.950573349149862e-05,
+      "loss": 2.0789,
+      "step": 7706
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9501779359430604e-05,
+      "loss": 1.9547,
+      "step": 7707
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9497825227362596e-05,
+      "loss": 1.8708,
+      "step": 7708
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.949387109529458e-05,
+      "loss": 1.8851,
+      "step": 7709
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9489916963226574e-05,
+      "loss": 1.8663,
+      "step": 7710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.948596283115856e-05,
+      "loss": 1.9411,
+      "step": 7711
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.948200869909055e-05,
+      "loss": 2.0477,
+      "step": 7712
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9478054567022537e-05,
+      "loss": 1.7688,
+      "step": 7713
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.947410043495453e-05,
+      "loss": 1.7753,
+      "step": 7714
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9470146302886515e-05,
+      "loss": 1.8244,
+      "step": 7715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9466192170818507e-05,
+      "loss": 2.1887,
+      "step": 7716
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9462238038750496e-05,
+      "loss": 2.0354,
+      "step": 7717
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9458283906682484e-05,
+      "loss": 1.888,
+      "step": 7718
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9454329774614473e-05,
+      "loss": 1.8869,
+      "step": 7719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9450375642546462e-05,
+      "loss": 2.0065,
+      "step": 7720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.944642151047845e-05,
+      "loss": 2.0892,
+      "step": 7721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.944246737841044e-05,
+      "loss": 1.736,
+      "step": 7722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.943851324634243e-05,
+      "loss": 1.865,
+      "step": 7723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9434559114274417e-05,
+      "loss": 1.9947,
+      "step": 7724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9430604982206406e-05,
+      "loss": 1.8167,
+      "step": 7725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9426650850138395e-05,
+      "loss": 1.9376,
+      "step": 7726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9422696718070384e-05,
+      "loss": 2.0099,
+      "step": 7727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9418742586002373e-05,
+      "loss": 1.9117,
+      "step": 7728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.941478845393436e-05,
+      "loss": 1.835,
+      "step": 7729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.941083432186635e-05,
+      "loss": 1.9149,
+      "step": 7730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.940688018979834e-05,
+      "loss": 1.9299,
+      "step": 7731
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9402926057730328e-05,
+      "loss": 1.9381,
+      "step": 7732
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9398971925662317e-05,
+      "loss": 1.8292,
+      "step": 7733
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9395017793594306e-05,
+      "loss": 1.8901,
+      "step": 7734
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9391063661526295e-05,
+      "loss": 1.7826,
+      "step": 7735
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9387109529458287e-05,
+      "loss": 2.1197,
+      "step": 7736
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9383155397390272e-05,
+      "loss": 2.0289,
+      "step": 7737
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9379201265322264e-05,
+      "loss": 1.9449,
+      "step": 7738
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.937524713325425e-05,
+      "loss": 2.0883,
+      "step": 7739
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9371293001186242e-05,
+      "loss": 2.0293,
+      "step": 7740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9367338869118228e-05,
+      "loss": 1.9378,
+      "step": 7741
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.936338473705022e-05,
+      "loss": 1.7145,
+      "step": 7742
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9359430604982205e-05,
+      "loss": 2.0646,
+      "step": 7743
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9355476472914197e-05,
+      "loss": 2.0828,
+      "step": 7744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9351522340846183e-05,
+      "loss": 1.9013,
+      "step": 7745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9347568208778175e-05,
+      "loss": 1.9019,
+      "step": 7746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.934361407671016e-05,
+      "loss": 1.9568,
+      "step": 7747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9339659944642153e-05,
+      "loss": 1.9317,
+      "step": 7748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9335705812574138e-05,
+      "loss": 1.895,
+      "step": 7749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.933175168050613e-05,
+      "loss": 2.0102,
+      "step": 7750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9327797548438116e-05,
+      "loss": 2.1761,
+      "step": 7751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9323843416370108e-05,
+      "loss": 2.1685,
+      "step": 7752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9319889284302097e-05,
+      "loss": 1.8232,
+      "step": 7753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9315935152234086e-05,
+      "loss": 1.9728,
+      "step": 7754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9311981020166074e-05,
+      "loss": 1.803,
+      "step": 7755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9308026888098063e-05,
+      "loss": 1.6715,
+      "step": 7756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9304072756030052e-05,
+      "loss": 1.8432,
+      "step": 7757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.930011862396204e-05,
+      "loss": 2.0298,
+      "step": 7758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.929616449189403e-05,
+      "loss": 2.0813,
+      "step": 7759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.929221035982602e-05,
+      "loss": 1.6824,
+      "step": 7760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9288256227758007e-05,
+      "loss": 1.9126,
+      "step": 7761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9284302095689996e-05,
+      "loss": 1.9781,
+      "step": 7762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9280347963621985e-05,
+      "loss": 1.7562,
+      "step": 7763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9276393831553974e-05,
+      "loss": 2.04,
+      "step": 7764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9272439699485963e-05,
+      "loss": 2.159,
+      "step": 7765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.926848556741795e-05,
+      "loss": 2.1253,
+      "step": 7766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.926453143534994e-05,
+      "loss": 2.2575,
+      "step": 7767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.926057730328193e-05,
+      "loss": 1.8281,
+      "step": 7768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9256623171213918e-05,
+      "loss": 2.0908,
+      "step": 7769
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.925266903914591e-05,
+      "loss": 1.7424,
+      "step": 7770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9248714907077896e-05,
+      "loss": 1.7213,
+      "step": 7771
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9244760775009888e-05,
+      "loss": 2.018,
+      "step": 7772
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9240806642941873e-05,
+      "loss": 2.0354,
+      "step": 7773
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9236852510873866e-05,
+      "loss": 1.7317,
+      "step": 7774
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.923289837880585e-05,
+      "loss": 1.8534,
+      "step": 7775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9228944246737843e-05,
+      "loss": 1.8141,
+      "step": 7776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.922499011466983e-05,
+      "loss": 2.0945,
+      "step": 7777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.922103598260182e-05,
+      "loss": 2.33,
+      "step": 7778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9217081850533806e-05,
+      "loss": 1.8781,
+      "step": 7779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.92131277184658e-05,
+      "loss": 2.2118,
+      "step": 7780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9209173586397784e-05,
+      "loss": 1.8029,
+      "step": 7781
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9205219454329776e-05,
+      "loss": 2.0144,
+      "step": 7782
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9201265322261765e-05,
+      "loss": 2.0876,
+      "step": 7783
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9197311190193754e-05,
+      "loss": 1.8704,
+      "step": 7784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9193357058125743e-05,
+      "loss": 1.6182,
+      "step": 7785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.918940292605773e-05,
+      "loss": 1.8681,
+      "step": 7786
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.918544879398972e-05,
+      "loss": 2.1578,
+      "step": 7787
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.918149466192171e-05,
+      "loss": 1.7572,
+      "step": 7788
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9177540529853698e-05,
+      "loss": 1.7743,
+      "step": 7789
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9173586397785687e-05,
+      "loss": 1.8627,
+      "step": 7790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9169632265717676e-05,
+      "loss": 1.7802,
+      "step": 7791
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9165678133649664e-05,
+      "loss": 1.9253,
+      "step": 7792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9161724001581653e-05,
+      "loss": 1.7353,
+      "step": 7793
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9157769869513642e-05,
+      "loss": 1.9468,
+      "step": 7794
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.915381573744563e-05,
+      "loss": 2.0051,
+      "step": 7795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.914986160537762e-05,
+      "loss": 2.0455,
+      "step": 7796
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.914590747330961e-05,
+      "loss": 2.0108,
+      "step": 7797
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9141953341241597e-05,
+      "loss": 2.1061,
+      "step": 7798
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9137999209173586e-05,
+      "loss": 1.7458,
+      "step": 7799
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9134045077105575e-05,
+      "loss": 1.792,
+      "step": 7800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9130090945037564e-05,
+      "loss": 2.0537,
+      "step": 7801
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9126136812969553e-05,
+      "loss": 1.8781,
+      "step": 7802
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.912218268090154e-05,
+      "loss": 1.7936,
+      "step": 7803
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9118228548833534e-05,
+      "loss": 2.1788,
+      "step": 7804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.911427441676552e-05,
+      "loss": 2.0784,
+      "step": 7805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.911032028469751e-05,
+      "loss": 1.9786,
+      "step": 7806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9106366152629497e-05,
+      "loss": 1.8374,
+      "step": 7807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.910241202056149e-05,
+      "loss": 2.2126,
+      "step": 7808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9098457888493475e-05,
+      "loss": 2.1879,
+      "step": 7809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9094503756425467e-05,
+      "loss": 1.808,
+      "step": 7810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9090549624357456e-05,
+      "loss": 2.0286,
+      "step": 7811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9086595492289444e-05,
+      "loss": 1.852,
+      "step": 7812
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9082641360221433e-05,
+      "loss": 1.9434,
+      "step": 7813
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9078687228153422e-05,
+      "loss": 1.9504,
+      "step": 7814
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.907473309608541e-05,
+      "loss": 1.919,
+      "step": 7815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.90707789640174e-05,
+      "loss": 1.832,
+      "step": 7816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.906682483194939e-05,
+      "loss": 1.8765,
+      "step": 7817
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9062870699881377e-05,
+      "loss": 1.8532,
+      "step": 7818
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9058916567813366e-05,
+      "loss": 1.8184,
+      "step": 7819
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9054962435745355e-05,
+      "loss": 1.932,
+      "step": 7820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9051008303677344e-05,
+      "loss": 2.0135,
+      "step": 7821
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9047054171609333e-05,
+      "loss": 2.0428,
+      "step": 7822
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.904310003954132e-05,
+      "loss": 2.0028,
+      "step": 7823
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.903914590747331e-05,
+      "loss": 1.8285,
+      "step": 7824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.90351917754053e-05,
+      "loss": 2.0439,
+      "step": 7825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9031237643337288e-05,
+      "loss": 1.9289,
+      "step": 7826
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9027283511269277e-05,
+      "loss": 1.9787,
+      "step": 7827
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9023329379201266e-05,
+      "loss": 1.9425,
+      "step": 7828
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9019375247133254e-05,
+      "loss": 1.8536,
+      "step": 7829
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9015421115065243e-05,
+      "loss": 1.8575,
+      "step": 7830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9011466982997232e-05,
+      "loss": 1.6446,
+      "step": 7831
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.900751285092922e-05,
+      "loss": 2.2133,
+      "step": 7832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.900355871886121e-05,
+      "loss": 1.9124,
+      "step": 7833
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.89996045867932e-05,
+      "loss": 2.0426,
+      "step": 7834
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8995650454725187e-05,
+      "loss": 1.9739,
+      "step": 7835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8991696322657176e-05,
+      "loss": 2.1214,
+      "step": 7836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8987742190589165e-05,
+      "loss": 2.1415,
+      "step": 7837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8983788058521154e-05,
+      "loss": 1.8233,
+      "step": 7838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8979833926453143e-05,
+      "loss": 2.0486,
+      "step": 7839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8975879794385135e-05,
+      "loss": 1.8055,
+      "step": 7840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8971925662317124e-05,
+      "loss": 2.0761,
+      "step": 7841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8967971530249113e-05,
+      "loss": 1.873,
+      "step": 7842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.89640173981811e-05,
+      "loss": 1.8833,
+      "step": 7843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.896006326611309e-05,
+      "loss": 2.0769,
+      "step": 7844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.895610913404508e-05,
+      "loss": 2.0548,
+      "step": 7845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8952155001977068e-05,
+      "loss": 1.9022,
+      "step": 7846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8948200869909057e-05,
+      "loss": 2.1957,
+      "step": 7847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8944246737841046e-05,
+      "loss": 2.0706,
+      "step": 7848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8940292605773034e-05,
+      "loss": 1.8033,
+      "step": 7849
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8936338473705023e-05,
+      "loss": 1.8375,
+      "step": 7850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8932384341637012e-05,
+      "loss": 1.8544,
+      "step": 7851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8928430209569e-05,
+      "loss": 1.801,
+      "step": 7852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.892447607750099e-05,
+      "loss": 2.0162,
+      "step": 7853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.892052194543298e-05,
+      "loss": 2.0289,
+      "step": 7854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8916567813364967e-05,
+      "loss": 1.9942,
+      "step": 7855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8912613681296956e-05,
+      "loss": 1.8797,
+      "step": 7856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8908659549228945e-05,
+      "loss": 1.8172,
+      "step": 7857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8904705417160934e-05,
+      "loss": 1.999,
+      "step": 7858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8900751285092923e-05,
+      "loss": 1.9923,
+      "step": 7859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.889679715302491e-05,
+      "loss": 2.1355,
+      "step": 7860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.88928430209569e-05,
+      "loss": 2.102,
+      "step": 7861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 1.8203,
+      "step": 7862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8884934756820878e-05,
+      "loss": 1.9085,
+      "step": 7863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8880980624752867e-05,
+      "loss": 1.8555,
+      "step": 7864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8877026492684856e-05,
+      "loss": 1.868,
+      "step": 7865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8873072360616844e-05,
+      "loss": 1.9151,
+      "step": 7866
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8869118228548833e-05,
+      "loss": 1.9408,
+      "step": 7867
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8865164096480822e-05,
+      "loss": 1.8722,
+      "step": 7868
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8861209964412814e-05,
+      "loss": 1.8918,
+      "step": 7869
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.88572558323448e-05,
+      "loss": 1.854,
+      "step": 7870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8853301700276792e-05,
+      "loss": 2.2313,
+      "step": 7871
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8849347568208777e-05,
+      "loss": 2.0687,
+      "step": 7872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.884539343614077e-05,
+      "loss": 1.8481,
+      "step": 7873
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.884143930407276e-05,
+      "loss": 1.9974,
+      "step": 7874
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8837485172004747e-05,
+      "loss": 1.933,
+      "step": 7875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8833531039936736e-05,
+      "loss": 1.8791,
+      "step": 7876
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8829576907868725e-05,
+      "loss": 1.8349,
+      "step": 7877
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8825622775800714e-05,
+      "loss": 1.7907,
+      "step": 7878
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8821668643732703e-05,
+      "loss": 2.1326,
+      "step": 7879
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.881771451166469e-05,
+      "loss": 1.8447,
+      "step": 7880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.881376037959668e-05,
+      "loss": 1.8018,
+      "step": 7881
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.880980624752867e-05,
+      "loss": 1.83,
+      "step": 7882
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8805852115460658e-05,
+      "loss": 2.0071,
+      "step": 7883
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8801897983392647e-05,
+      "loss": 1.7015,
+      "step": 7884
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8797943851324636e-05,
+      "loss": 1.7376,
+      "step": 7885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8793989719256624e-05,
+      "loss": 1.7069,
+      "step": 7886
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8790035587188613e-05,
+      "loss": 2.1077,
+      "step": 7887
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8786081455120602e-05,
+      "loss": 1.8737,
+      "step": 7888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.878212732305259e-05,
+      "loss": 1.8332,
+      "step": 7889
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.877817319098458e-05,
+      "loss": 2.142,
+      "step": 7890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.877421905891657e-05,
+      "loss": 2.057,
+      "step": 7891
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8770264926848557e-05,
+      "loss": 1.9116,
+      "step": 7892
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8766310794780546e-05,
+      "loss": 1.9678,
+      "step": 7893
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8762356662712535e-05,
+      "loss": 1.9807,
+      "step": 7894
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8758402530644524e-05,
+      "loss": 1.7889,
+      "step": 7895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8754448398576513e-05,
+      "loss": 1.8982,
+      "step": 7896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8750494266508505e-05,
+      "loss": 2.0346,
+      "step": 7897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.874654013444049e-05,
+      "loss": 1.9254,
+      "step": 7898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8742586002372482e-05,
+      "loss": 2.1026,
+      "step": 7899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8738631870304468e-05,
+      "loss": 1.9123,
+      "step": 7900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.873467773823646e-05,
+      "loss": 1.7144,
+      "step": 7901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8730723606168446e-05,
+      "loss": 1.876,
+      "step": 7902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8726769474100438e-05,
+      "loss": 1.9157,
+      "step": 7903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8722815342032423e-05,
+      "loss": 1.7706,
+      "step": 7904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8718861209964415e-05,
+      "loss": 1.9666,
+      "step": 7905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.87149070778964e-05,
+      "loss": 1.802,
+      "step": 7906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8710952945828393e-05,
+      "loss": 1.8933,
+      "step": 7907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8706998813760382e-05,
+      "loss": 1.9839,
+      "step": 7908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.870304468169237e-05,
+      "loss": 1.9509,
+      "step": 7909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.869909054962436e-05,
+      "loss": 1.9007,
+      "step": 7910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.869513641755635e-05,
+      "loss": 2.053,
+      "step": 7911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8691182285488337e-05,
+      "loss": 1.8682,
+      "step": 7912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8687228153420326e-05,
+      "loss": 1.6802,
+      "step": 7913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8683274021352315e-05,
+      "loss": 2.0988,
+      "step": 7914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8679319889284304e-05,
+      "loss": 1.7414,
+      "step": 7915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8675365757216293e-05,
+      "loss": 2.0847,
+      "step": 7916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.867141162514828e-05,
+      "loss": 2.0381,
+      "step": 7917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.866745749308027e-05,
+      "loss": 2.0306,
+      "step": 7918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.866350336101226e-05,
+      "loss": 1.9652,
+      "step": 7919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8659549228944248e-05,
+      "loss": 1.8011,
+      "step": 7920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8655595096876237e-05,
+      "loss": 1.8046,
+      "step": 7921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8651640964808226e-05,
+      "loss": 1.9132,
+      "step": 7922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8647686832740214e-05,
+      "loss": 2.0241,
+      "step": 7923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8643732700672203e-05,
+      "loss": 1.94,
+      "step": 7924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8639778568604192e-05,
+      "loss": 1.9794,
+      "step": 7925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.863582443653618e-05,
+      "loss": 2.0737,
+      "step": 7926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8631870304468173e-05,
+      "loss": 1.9911,
+      "step": 7927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.862791617240016e-05,
+      "loss": 2.2566,
+      "step": 7928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.862396204033215e-05,
+      "loss": 1.8266,
+      "step": 7929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8620007908264136e-05,
+      "loss": 1.7207,
+      "step": 7930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.861605377619613e-05,
+      "loss": 1.6699,
+      "step": 7931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8612099644128114e-05,
+      "loss": 1.8865,
+      "step": 7932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8608145512060106e-05,
+      "loss": 1.984,
+      "step": 7933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.860419137999209e-05,
+      "loss": 1.9059,
+      "step": 7934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8600237247924084e-05,
+      "loss": 1.9856,
+      "step": 7935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.859628311585607e-05,
+      "loss": 1.6838,
+      "step": 7936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.859232898378806e-05,
+      "loss": 1.9268,
+      "step": 7937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8588374851720047e-05,
+      "loss": 1.9318,
+      "step": 7938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.858442071965204e-05,
+      "loss": 1.7382,
+      "step": 7939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8580466587584024e-05,
+      "loss": 1.6933,
+      "step": 7940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8576512455516017e-05,
+      "loss": 2.1111,
+      "step": 7941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8572558323448002e-05,
+      "loss": 1.9233,
+      "step": 7942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8568604191379994e-05,
+      "loss": 1.8818,
+      "step": 7943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8564650059311983e-05,
+      "loss": 1.7821,
+      "step": 7944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8560695927243972e-05,
+      "loss": 1.8063,
+      "step": 7945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.855674179517596e-05,
+      "loss": 1.9156,
+      "step": 7946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.855278766310795e-05,
+      "loss": 1.7066,
+      "step": 7947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.854883353103994e-05,
+      "loss": 1.9973,
+      "step": 7948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8544879398971927e-05,
+      "loss": 1.7706,
+      "step": 7949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8540925266903916e-05,
+      "loss": 1.7999,
+      "step": 7950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8536971134835905e-05,
+      "loss": 1.8744,
+      "step": 7951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8533017002767894e-05,
+      "loss": 1.9091,
+      "step": 7952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8529062870699883e-05,
+      "loss": 1.9246,
+      "step": 7953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.852510873863187e-05,
+      "loss": 2.1965,
+      "step": 7954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.852115460656386e-05,
+      "loss": 1.6798,
+      "step": 7955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.851720047449585e-05,
+      "loss": 2.0778,
+      "step": 7956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8513246342427838e-05,
+      "loss": 1.8327,
+      "step": 7957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8509292210359827e-05,
+      "loss": 2.0411,
+      "step": 7958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8505338078291815e-05,
+      "loss": 1.8867,
+      "step": 7959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8501383946223804e-05,
+      "loss": 2.0771,
+      "step": 7960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8497429814155797e-05,
+      "loss": 1.8028,
+      "step": 7961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8493475682087782e-05,
+      "loss": 2.0848,
+      "step": 7962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8489521550019774e-05,
+      "loss": 1.8858,
+      "step": 7963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.848556741795176e-05,
+      "loss": 2.2531,
+      "step": 7964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8481613285883752e-05,
+      "loss": 1.8288,
+      "step": 7965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8477659153815737e-05,
+      "loss": 2.0303,
+      "step": 7966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.847370502174773e-05,
+      "loss": 2.0078,
+      "step": 7967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8469750889679715e-05,
+      "loss": 2.0156,
+      "step": 7968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8465796757611707e-05,
+      "loss": 1.7598,
+      "step": 7969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8461842625543693e-05,
+      "loss": 1.7916,
+      "step": 7970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8457888493475685e-05,
+      "loss": 1.9876,
+      "step": 7971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.845393436140767e-05,
+      "loss": 1.8968,
+      "step": 7972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8449980229339662e-05,
+      "loss": 1.7683,
+      "step": 7973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8446026097271648e-05,
+      "loss": 1.9716,
+      "step": 7974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.844207196520364e-05,
+      "loss": 1.8979,
+      "step": 7975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8438117833135626e-05,
+      "loss": 1.9739,
+      "step": 7976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8434163701067618e-05,
+      "loss": 1.9946,
+      "step": 7977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8430209568999607e-05,
+      "loss": 1.973,
+      "step": 7978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8426255436931595e-05,
+      "loss": 1.9105,
+      "step": 7979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8422301304863584e-05,
+      "loss": 1.8355,
+      "step": 7980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8418347172795573e-05,
+      "loss": 2.055,
+      "step": 7981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8414393040727562e-05,
+      "loss": 1.7402,
+      "step": 7982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.841043890865955e-05,
+      "loss": 2.1202,
+      "step": 7983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.840648477659154e-05,
+      "loss": 2.0924,
+      "step": 7984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.840253064452353e-05,
+      "loss": 1.9487,
+      "step": 7985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8398576512455517e-05,
+      "loss": 1.6638,
+      "step": 7986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8394622380387506e-05,
+      "loss": 1.9231,
+      "step": 7987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8390668248319495e-05,
+      "loss": 1.8196,
+      "step": 7988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8386714116251484e-05,
+      "loss": 2.0397,
+      "step": 7989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8382759984183473e-05,
+      "loss": 2.0351,
+      "step": 7990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837880585211546e-05,
+      "loss": 1.9778,
+      "step": 7991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837485172004745e-05,
+      "loss": 1.8765,
+      "step": 7992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837089758797944e-05,
+      "loss": 2.007,
+      "step": 7993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366943455911428e-05,
+      "loss": 1.8924,
+      "step": 7994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836298932384342e-05,
+      "loss": 2.1429,
+      "step": 7995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8359035191775405e-05,
+      "loss": 1.8836,
+      "step": 7996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8355081059707398e-05,
+      "loss": 1.8922,
+      "step": 7997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8351126927639383e-05,
+      "loss": 1.8531,
+      "step": 7998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347172795571375e-05,
+      "loss": 2.1069,
+      "step": 7999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834321866350336e-05,
+      "loss": 1.7167,
+      "step": 8000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8339264531435353e-05,
+      "loss": 2.3358,
+      "step": 8001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833531039936734e-05,
+      "loss": 1.9408,
+      "step": 8002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833135626729933e-05,
+      "loss": 1.8042,
+      "step": 8003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8327402135231316e-05,
+      "loss": 2.008,
+      "step": 8004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832344800316331e-05,
+      "loss": 1.9276,
+      "step": 8005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8319493871095294e-05,
+      "loss": 1.9957,
+      "step": 8006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8315539739027286e-05,
+      "loss": 2.1004,
+      "step": 8007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.831158560695927e-05,
+      "loss": 1.7396,
+      "step": 8008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8307631474891264e-05,
+      "loss": 1.767,
+      "step": 8009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.830367734282325e-05,
+      "loss": 1.9467,
+      "step": 8010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.829972321075524e-05,
+      "loss": 1.8352,
+      "step": 8011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8295769078687227e-05,
+      "loss": 1.8063,
+      "step": 8012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.829181494661922e-05,
+      "loss": 1.8225,
+      "step": 8013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8287860814551208e-05,
+      "loss": 1.8772,
+      "step": 8014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8283906682483197e-05,
+      "loss": 1.8698,
+      "step": 8015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8279952550415185e-05,
+      "loss": 1.8778,
+      "step": 8016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8275998418347174e-05,
+      "loss": 1.8162,
+      "step": 8017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8272044286279163e-05,
+      "loss": 1.9253,
+      "step": 8018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8268090154211152e-05,
+      "loss": 1.946,
+      "step": 8019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.826413602214314e-05,
+      "loss": 1.7627,
+      "step": 8020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.826018189007513e-05,
+      "loss": 1.9803,
+      "step": 8021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.825622775800712e-05,
+      "loss": 1.7509,
+      "step": 8022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8252273625939107e-05,
+      "loss": 2.0318,
+      "step": 8023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8248319493871096e-05,
+      "loss": 1.9884,
+      "step": 8024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8244365361803085e-05,
+      "loss": 1.9357,
+      "step": 8025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8240411229735074e-05,
+      "loss": 1.7459,
+      "step": 8026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8236457097667063e-05,
+      "loss": 1.9604,
+      "step": 8027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.823250296559905e-05,
+      "loss": 2.1173,
+      "step": 8028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.822854883353104e-05,
+      "loss": 1.7966,
+      "step": 8029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.822459470146303e-05,
+      "loss": 2.1861,
+      "step": 8030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.822064056939502e-05,
+      "loss": 1.973,
+      "step": 8031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8216686437327007e-05,
+      "loss": 1.8724,
+      "step": 8032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8212732305259e-05,
+      "loss": 2.0074,
+      "step": 8033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8208778173190984e-05,
+      "loss": 1.8769,
+      "step": 8034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8204824041122977e-05,
+      "loss": 1.9525,
+      "step": 8035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8200869909054962e-05,
+      "loss": 1.7981,
+      "step": 8036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8196915776986954e-05,
+      "loss": 1.971,
+      "step": 8037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.819296164491894e-05,
+      "loss": 2.0877,
+      "step": 8038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8189007512850932e-05,
+      "loss": 1.8593,
+      "step": 8039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8185053380782917e-05,
+      "loss": 1.8148,
+      "step": 8040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.818109924871491e-05,
+      "loss": 1.9061,
+      "step": 8041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8177145116646895e-05,
+      "loss": 1.8933,
+      "step": 8042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8173190984578887e-05,
+      "loss": 1.9684,
+      "step": 8043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8169236852510873e-05,
+      "loss": 1.8344,
+      "step": 8044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8165282720442865e-05,
+      "loss": 1.8268,
+      "step": 8045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.816132858837485e-05,
+      "loss": 1.7875,
+      "step": 8046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8157374456306842e-05,
+      "loss": 1.7577,
+      "step": 8047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.815342032423883e-05,
+      "loss": 2.2125,
+      "step": 8048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.814946619217082e-05,
+      "loss": 1.8423,
+      "step": 8049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.814551206010281e-05,
+      "loss": 1.854,
+      "step": 8050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8141557928034798e-05,
+      "loss": 1.8248,
+      "step": 8051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8137603795966787e-05,
+      "loss": 2.037,
+      "step": 8052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8133649663898775e-05,
+      "loss": 1.9874,
+      "step": 8053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8129695531830764e-05,
+      "loss": 2.0827,
+      "step": 8054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8125741399762753e-05,
+      "loss": 2.0779,
+      "step": 8055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8121787267694742e-05,
+      "loss": 1.9944,
+      "step": 8056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.811783313562673e-05,
+      "loss": 1.9669,
+      "step": 8057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.811387900355872e-05,
+      "loss": 1.9437,
+      "step": 8058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.810992487149071e-05,
+      "loss": 2.1563,
+      "step": 8059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8105970739422697e-05,
+      "loss": 1.9863,
+      "step": 8060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8102016607354686e-05,
+      "loss": 2.0882,
+      "step": 8061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8098062475286675e-05,
+      "loss": 2.1367,
+      "step": 8062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8094108343218664e-05,
+      "loss": 1.9819,
+      "step": 8063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8090154211150652e-05,
+      "loss": 1.8701,
+      "step": 8064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8086200079082645e-05,
+      "loss": 1.9011,
+      "step": 8065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.808224594701463e-05,
+      "loss": 2.048,
+      "step": 8066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8078291814946622e-05,
+      "loss": 1.9558,
+      "step": 8067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8074337682878608e-05,
+      "loss": 1.9645,
+      "step": 8068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.80703835508106e-05,
+      "loss": 2.0354,
+      "step": 8069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8066429418742585e-05,
+      "loss": 1.9021,
+      "step": 8070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8062475286674578e-05,
+      "loss": 1.801,
+      "step": 8071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8058521154606563e-05,
+      "loss": 1.8532,
+      "step": 8072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8054567022538555e-05,
+      "loss": 1.9283,
+      "step": 8073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.805061289047054e-05,
+      "loss": 2.0165,
+      "step": 8074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8046658758402533e-05,
+      "loss": 2.0117,
+      "step": 8075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.804270462633452e-05,
+      "loss": 1.6398,
+      "step": 8076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.803875049426651e-05,
+      "loss": 1.8508,
+      "step": 8077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8034796362198496e-05,
+      "loss": 2.0713,
+      "step": 8078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8030842230130488e-05,
+      "loss": 2.001,
+      "step": 8079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8026888098062474e-05,
+      "loss": 1.9024,
+      "step": 8080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8022933965994466e-05,
+      "loss": 1.8989,
+      "step": 8081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8018979833926455e-05,
+      "loss": 1.7866,
+      "step": 8082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8015025701858444e-05,
+      "loss": 2.0245,
+      "step": 8083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8011071569790432e-05,
+      "loss": 1.7825,
+      "step": 8084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.800711743772242e-05,
+      "loss": 2.1001,
+      "step": 8085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.800316330565441e-05,
+      "loss": 1.8458,
+      "step": 8086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.79992091735864e-05,
+      "loss": 1.6393,
+      "step": 8087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7995255041518388e-05,
+      "loss": 2.0843,
+      "step": 8088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7991300909450377e-05,
+      "loss": 1.9913,
+      "step": 8089
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7987346777382365e-05,
+      "loss": 2.0278,
+      "step": 8090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7983392645314354e-05,
+      "loss": 1.5737,
+      "step": 8091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7979438513246343e-05,
+      "loss": 1.7476,
+      "step": 8092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7975484381178332e-05,
+      "loss": 2.0031,
+      "step": 8093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.797153024911032e-05,
+      "loss": 2.0849,
+      "step": 8094
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.796757611704231e-05,
+      "loss": 1.9093,
+      "step": 8095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.79636219849743e-05,
+      "loss": 1.8582,
+      "step": 8096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7959667852906287e-05,
+      "loss": 1.9439,
+      "step": 8097
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7955713720838276e-05,
+      "loss": 2.1089,
+      "step": 8098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7951759588770265e-05,
+      "loss": 1.8891,
+      "step": 8099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7947805456702254e-05,
+      "loss": 1.9694,
+      "step": 8100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7943851324634246e-05,
+      "loss": 1.8566,
+      "step": 8101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.793989719256623e-05,
+      "loss": 1.8718,
+      "step": 8102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7935943060498224e-05,
+      "loss": 2.0201,
+      "step": 8103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.793198892843021e-05,
+      "loss": 1.7392,
+      "step": 8104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.79280347963622e-05,
+      "loss": 2.2589,
+      "step": 8105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7924080664294187e-05,
+      "loss": 2.0323,
+      "step": 8106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.792012653222618e-05,
+      "loss": 1.7886,
+      "step": 8107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7916172400158164e-05,
+      "loss": 1.8759,
+      "step": 8108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7912218268090156e-05,
+      "loss": 1.8745,
+      "step": 8109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7908264136022142e-05,
+      "loss": 1.781,
+      "step": 8110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7904310003954134e-05,
+      "loss": 1.8797,
+      "step": 8111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.790035587188612e-05,
+      "loss": 1.9661,
+      "step": 8112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7896401739818112e-05,
+      "loss": 1.9703,
+      "step": 8113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7892447607750097e-05,
+      "loss": 1.9445,
+      "step": 8114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.788849347568209e-05,
+      "loss": 2.0079,
+      "step": 8115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7884539343614075e-05,
+      "loss": 1.8229,
+      "step": 8116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7880585211546067e-05,
+      "loss": 2.0035,
+      "step": 8117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7876631079478056e-05,
+      "loss": 1.9453,
+      "step": 8118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7872676947410045e-05,
+      "loss": 1.6581,
+      "step": 8119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7868722815342034e-05,
+      "loss": 1.9873,
+      "step": 8120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7864768683274022e-05,
+      "loss": 1.9575,
+      "step": 8121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.786081455120601e-05,
+      "loss": 2.1121,
+      "step": 8122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7856860419138e-05,
+      "loss": 1.9168,
+      "step": 8123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.785290628706999e-05,
+      "loss": 2.1102,
+      "step": 8124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7848952155001978e-05,
+      "loss": 1.9918,
+      "step": 8125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7844998022933967e-05,
+      "loss": 1.8483,
+      "step": 8126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7841043890865955e-05,
+      "loss": 1.8491,
+      "step": 8127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7837089758797944e-05,
+      "loss": 1.8587,
+      "step": 8128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7833135626729933e-05,
+      "loss": 1.9384,
+      "step": 8129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7829181494661922e-05,
+      "loss": 1.9875,
+      "step": 8130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.782522736259391e-05,
+      "loss": 1.8398,
+      "step": 8131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.78212732305259e-05,
+      "loss": 2.0334,
+      "step": 8132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.781731909845789e-05,
+      "loss": 2.0255,
+      "step": 8133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7813364966389877e-05,
+      "loss": 2.0356,
+      "step": 8134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.780941083432187e-05,
+      "loss": 1.9484,
+      "step": 8135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7805456702253855e-05,
+      "loss": 2.0725,
+      "step": 8136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7801502570185847e-05,
+      "loss": 1.7801,
+      "step": 8137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7797548438117832e-05,
+      "loss": 2.2159,
+      "step": 8138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7793594306049825e-05,
+      "loss": 2.0227,
+      "step": 8139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.778964017398181e-05,
+      "loss": 2.1691,
+      "step": 8140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7785686041913802e-05,
+      "loss": 2.2124,
+      "step": 8141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7781731909845788e-05,
+      "loss": 1.8006,
+      "step": 8142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.777777777777778e-05,
+      "loss": 1.9171,
+      "step": 8143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7773823645709765e-05,
+      "loss": 2.0031,
+      "step": 8144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7769869513641758e-05,
+      "loss": 1.8709,
+      "step": 8145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7765915381573743e-05,
+      "loss": 2.1261,
+      "step": 8146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7761961249505735e-05,
+      "loss": 2.014,
+      "step": 8147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.775800711743772e-05,
+      "loss": 2.0123,
+      "step": 8148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7754052985369713e-05,
+      "loss": 1.9197,
+      "step": 8149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.77500988533017e-05,
+      "loss": 2.0501,
+      "step": 8150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.774614472123369e-05,
+      "loss": 2.103,
+      "step": 8151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.774219058916568e-05,
+      "loss": 1.8911,
+      "step": 8152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7738236457097668e-05,
+      "loss": 2.0274,
+      "step": 8153
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7734282325029657e-05,
+      "loss": 2.1865,
+      "step": 8154
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7730328192961646e-05,
+      "loss": 1.9992,
+      "step": 8155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7726374060893635e-05,
+      "loss": 1.8571,
+      "step": 8156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7722419928825624e-05,
+      "loss": 1.8019,
+      "step": 8157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7718465796757612e-05,
+      "loss": 1.8965,
+      "step": 8158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.77145116646896e-05,
+      "loss": 1.9824,
+      "step": 8159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.771055753262159e-05,
+      "loss": 1.7514,
+      "step": 8160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.770660340055358e-05,
+      "loss": 1.857,
+      "step": 8161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7702649268485568e-05,
+      "loss": 2.0284,
+      "step": 8162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7698695136417557e-05,
+      "loss": 1.8842,
+      "step": 8163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7694741004349545e-05,
+      "loss": 1.9424,
+      "step": 8164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7690786872281534e-05,
+      "loss": 2.059,
+      "step": 8165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7686832740213523e-05,
+      "loss": 1.8929,
+      "step": 8166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7682878608145512e-05,
+      "loss": 1.6358,
+      "step": 8167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.76789244760775e-05,
+      "loss": 1.7906,
+      "step": 8168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7674970344009493e-05,
+      "loss": 2.0619,
+      "step": 8169
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.767101621194148e-05,
+      "loss": 1.7928,
+      "step": 8170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.766706207987347e-05,
+      "loss": 1.8026,
+      "step": 8171
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7663107947805456e-05,
+      "loss": 1.7293,
+      "step": 8172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7659153815737448e-05,
+      "loss": 1.657,
+      "step": 8173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7655199683669434e-05,
+      "loss": 2.1043,
+      "step": 8174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7651245551601426e-05,
+      "loss": 1.8461,
+      "step": 8175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.764729141953341e-05,
+      "loss": 1.7604,
+      "step": 8176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7643337287465403e-05,
+      "loss": 1.8717,
+      "step": 8177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.763938315539739e-05,
+      "loss": 1.686,
+      "step": 8178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.763542902332938e-05,
+      "loss": 1.8586,
+      "step": 8179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7631474891261367e-05,
+      "loss": 2.0972,
+      "step": 8180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.762752075919336e-05,
+      "loss": 2.1556,
+      "step": 8181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7623566627125344e-05,
+      "loss": 1.9352,
+      "step": 8182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7619612495057336e-05,
+      "loss": 1.9891,
+      "step": 8183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7615658362989322e-05,
+      "loss": 2.1164,
+      "step": 8184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7611704230921314e-05,
+      "loss": 1.6427,
+      "step": 8185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.76077500988533e-05,
+      "loss": 1.9984,
+      "step": 8186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7603795966785292e-05,
+      "loss": 1.7872,
+      "step": 8187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.759984183471728e-05,
+      "loss": 1.9069,
+      "step": 8188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.759588770264927e-05,
+      "loss": 1.9678,
+      "step": 8189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7591933570581258e-05,
+      "loss": 1.883,
+      "step": 8190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7587979438513247e-05,
+      "loss": 1.7509,
+      "step": 8191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7584025306445236e-05,
+      "loss": 1.8648,
+      "step": 8192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7580071174377225e-05,
+      "loss": 2.0213,
+      "step": 8193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7576117042309214e-05,
+      "loss": 1.9015,
+      "step": 8194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7572162910241202e-05,
+      "loss": 2.0578,
+      "step": 8195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.756820877817319e-05,
+      "loss": 1.9592,
+      "step": 8196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.756425464610518e-05,
+      "loss": 1.7589,
+      "step": 8197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.756030051403717e-05,
+      "loss": 2.1018,
+      "step": 8198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7556346381969158e-05,
+      "loss": 1.9539,
+      "step": 8199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7552392249901147e-05,
+      "loss": 1.9813,
+      "step": 8200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7548438117833135e-05,
+      "loss": 1.9508,
+      "step": 8201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7544483985765124e-05,
+      "loss": 1.9261,
+      "step": 8202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7540529853697113e-05,
+      "loss": 1.8757,
+      "step": 8203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7536575721629102e-05,
+      "loss": 2.0353,
+      "step": 8204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7532621589561094e-05,
+      "loss": 1.7477,
+      "step": 8205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.752866745749308e-05,
+      "loss": 2.1177,
+      "step": 8206
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.752471332542507e-05,
+      "loss": 1.9115,
+      "step": 8207
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7520759193357057e-05,
+      "loss": 2.0001,
+      "step": 8208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.751680506128905e-05,
+      "loss": 2.0026,
+      "step": 8209
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7512850929221035e-05,
+      "loss": 1.9485,
+      "step": 8210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7508896797153027e-05,
+      "loss": 1.9705,
+      "step": 8211
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7504942665085012e-05,
+      "loss": 1.7857,
+      "step": 8212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7500988533017005e-05,
+      "loss": 1.8867,
+      "step": 8213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.749703440094899e-05,
+      "loss": 2.0422,
+      "step": 8214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7493080268880982e-05,
+      "loss": 2.0913,
+      "step": 8215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7489126136812968e-05,
+      "loss": 2.1187,
+      "step": 8216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.748517200474496e-05,
+      "loss": 1.8324,
+      "step": 8217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.748121787267695e-05,
+      "loss": 1.874,
+      "step": 8218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7477263740608938e-05,
+      "loss": 1.8679,
+      "step": 8219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7473309608540926e-05,
+      "loss": 1.9855,
+      "step": 8220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7469355476472915e-05,
+      "loss": 1.9876,
+      "step": 8221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7465401344404904e-05,
+      "loss": 1.7704,
+      "step": 8222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7461447212336893e-05,
+      "loss": 1.8456,
+      "step": 8223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7457493080268882e-05,
+      "loss": 1.7288,
+      "step": 8224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.745353894820087e-05,
+      "loss": 1.7972,
+      "step": 8225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.744958481613286e-05,
+      "loss": 1.8399,
+      "step": 8226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7445630684064848e-05,
+      "loss": 2.1898,
+      "step": 8227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7441676551996837e-05,
+      "loss": 1.962,
+      "step": 8228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7437722419928826e-05,
+      "loss": 1.8628,
+      "step": 8229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7433768287860815e-05,
+      "loss": 1.9744,
+      "step": 8230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7429814155792804e-05,
+      "loss": 2.0598,
+      "step": 8231
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7425860023724792e-05,
+      "loss": 1.8645,
+      "step": 8232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.742190589165678e-05,
+      "loss": 1.8275,
+      "step": 8233
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.741795175958877e-05,
+      "loss": 1.9284,
+      "step": 8234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.741399762752076e-05,
+      "loss": 2.1593,
+      "step": 8235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7410043495452748e-05,
+      "loss": 2.224,
+      "step": 8236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7406089363384736e-05,
+      "loss": 1.7842,
+      "step": 8237
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7402135231316725e-05,
+      "loss": 1.9623,
+      "step": 8238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7398181099248718e-05,
+      "loss": 1.9189,
+      "step": 8239
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7394226967180703e-05,
+      "loss": 2.0343,
+      "step": 8240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7390272835112695e-05,
+      "loss": 1.8388,
+      "step": 8241
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.738631870304468e-05,
+      "loss": 1.8059,
+      "step": 8242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7382364570976673e-05,
+      "loss": 2.1609,
+      "step": 8243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7378410438908658e-05,
+      "loss": 1.9877,
+      "step": 8244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.737445630684065e-05,
+      "loss": 2.1458,
+      "step": 8245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.737050217477264e-05,
+      "loss": 2.1749,
+      "step": 8246
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7366548042704628e-05,
+      "loss": 1.9172,
+      "step": 8247
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7362593910636617e-05,
+      "loss": 1.6556,
+      "step": 8248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7358639778568606e-05,
+      "loss": 1.7891,
+      "step": 8249
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7354685646500595e-05,
+      "loss": 1.8611,
+      "step": 8250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7350731514432583e-05,
+      "loss": 1.7956,
+      "step": 8251
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7346777382364572e-05,
+      "loss": 2.2238,
+      "step": 8252
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.734282325029656e-05,
+      "loss": 1.9961,
+      "step": 8253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.733886911822855e-05,
+      "loss": 2.2769,
+      "step": 8254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.733491498616054e-05,
+      "loss": 1.7585,
+      "step": 8255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7330960854092528e-05,
+      "loss": 1.847,
+      "step": 8256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7327006722024516e-05,
+      "loss": 1.9292,
+      "step": 8257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7323052589956505e-05,
+      "loss": 1.787,
+      "step": 8258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7319098457888494e-05,
+      "loss": 1.7738,
+      "step": 8259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7315144325820483e-05,
+      "loss": 1.8212,
+      "step": 8260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7311190193752472e-05,
+      "loss": 1.8615,
+      "step": 8261
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.730723606168446e-05,
+      "loss": 2.112,
+      "step": 8262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.730328192961645e-05,
+      "loss": 2.0125,
+      "step": 8263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7299327797548438e-05,
+      "loss": 2.0493,
+      "step": 8264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7295373665480427e-05,
+      "loss": 1.8711,
+      "step": 8265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7291419533412416e-05,
+      "loss": 1.8345,
+      "step": 8266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7287465401344405e-05,
+      "loss": 1.9243,
+      "step": 8267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7283511269276394e-05,
+      "loss": 2.0282,
+      "step": 8268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7279557137208382e-05,
+      "loss": 1.8535,
+      "step": 8269
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.727560300514037e-05,
+      "loss": 2.1476,
+      "step": 8270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.727164887307236e-05,
+      "loss": 2.192,
+      "step": 8271
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.726769474100435e-05,
+      "loss": 1.8102,
+      "step": 8272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7263740608936338e-05,
+      "loss": 2.0191,
+      "step": 8273
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7259786476868326e-05,
+      "loss": 2.1388,
+      "step": 8274
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.725583234480032e-05,
+      "loss": 1.9622,
+      "step": 8275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7251878212732308e-05,
+      "loss": 1.8871,
+      "step": 8276
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7247924080664296e-05,
+      "loss": 1.7436,
+      "step": 8277
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7243969948596285e-05,
+      "loss": 2.12,
+      "step": 8278
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7240015816528274e-05,
+      "loss": 1.9726,
+      "step": 8279
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7236061684460263e-05,
+      "loss": 1.9895,
+      "step": 8280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.723210755239225e-05,
+      "loss": 2.0498,
+      "step": 8281
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.722815342032424e-05,
+      "loss": 1.8001,
+      "step": 8282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.722419928825623e-05,
+      "loss": 1.9711,
+      "step": 8283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7220245156188218e-05,
+      "loss": 1.897,
+      "step": 8284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7216291024120207e-05,
+      "loss": 1.8872,
+      "step": 8285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7212336892052196e-05,
+      "loss": 2.0011,
+      "step": 8286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7208382759984185e-05,
+      "loss": 2.0757,
+      "step": 8287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7204428627916173e-05,
+      "loss": 2.0567,
+      "step": 8288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7200474495848162e-05,
+      "loss": 1.8038,
+      "step": 8289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.719652036378015e-05,
+      "loss": 1.9959,
+      "step": 8290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.719256623171214e-05,
+      "loss": 1.7772,
+      "step": 8291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.718861209964413e-05,
+      "loss": 2.0271,
+      "step": 8292
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7184657967576118e-05,
+      "loss": 1.9424,
+      "step": 8293
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7180703835508106e-05,
+      "loss": 1.9452,
+      "step": 8294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7176749703440095e-05,
+      "loss": 2.0404,
+      "step": 8295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7172795571372084e-05,
+      "loss": 1.9077,
+      "step": 8296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7168841439304073e-05,
+      "loss": 2.0761,
+      "step": 8297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7164887307236062e-05,
+      "loss": 1.8438,
+      "step": 8298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.716093317516805e-05,
+      "loss": 1.8378,
+      "step": 8299
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.715697904310004e-05,
+      "loss": 1.9348,
+      "step": 8300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7153024911032028e-05,
+      "loss": 1.7874,
+      "step": 8301
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7149070778964017e-05,
+      "loss": 1.9057,
+      "step": 8302
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7149070778964017e-05,
+      "loss": 1.9843,
+      "step": 8303
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7145116646896006e-05,
+      "loss": 1.867,
+      "step": 8304
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7141162514827998e-05,
+      "loss": 2.0024,
+      "step": 8305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7137208382759984e-05,
+      "loss": 2.1241,
+      "step": 8306
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7133254250691976e-05,
+      "loss": 2.0657,
+      "step": 8307
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.712930011862396e-05,
+      "loss": 1.8578,
+      "step": 8308
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7125345986555953e-05,
+      "loss": 1.798,
+      "step": 8309
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7121391854487942e-05,
+      "loss": 2.066,
+      "step": 8310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.711743772241993e-05,
+      "loss": 1.8664,
+      "step": 8311
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.711348359035192e-05,
+      "loss": 1.7769,
+      "step": 8312
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.710952945828391e-05,
+      "loss": 1.8448,
+      "step": 8313
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7105575326215898e-05,
+      "loss": 1.968,
+      "step": 8314
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7101621194147886e-05,
+      "loss": 2.0942,
+      "step": 8315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7097667062079875e-05,
+      "loss": 1.933,
+      "step": 8316
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7093712930011864e-05,
+      "loss": 1.872,
+      "step": 8317
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7089758797943853e-05,
+      "loss": 1.9139,
+      "step": 8318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.708580466587584e-05,
+      "loss": 1.8946,
+      "step": 8319
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.708185053380783e-05,
+      "loss": 1.964,
+      "step": 8320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.707789640173982e-05,
+      "loss": 2.254,
+      "step": 8321
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7073942269671808e-05,
+      "loss": 1.9951,
+      "step": 8322
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7069988137603797e-05,
+      "loss": 1.8412,
+      "step": 8323
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7066034005535786e-05,
+      "loss": 1.9451,
+      "step": 8324
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7062079873467775e-05,
+      "loss": 1.7976,
+      "step": 8325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7058125741399763e-05,
+      "loss": 1.7125,
+      "step": 8326
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7054171609331752e-05,
+      "loss": 2.0456,
+      "step": 8327
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.705021747726374e-05,
+      "loss": 1.7913,
+      "step": 8328
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.704626334519573e-05,
+      "loss": 1.9096,
+      "step": 8329
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.704230921312772e-05,
+      "loss": 1.7731,
+      "step": 8330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7038355081059708e-05,
+      "loss": 1.8357,
+      "step": 8331
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7034400948991696e-05,
+      "loss": 1.9213,
+      "step": 8332
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.703044681692369e-05,
+      "loss": 2.1552,
+      "step": 8333
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7026492684855674e-05,
+      "loss": 1.9732,
+      "step": 8334
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7022538552787666e-05,
+      "loss": 1.7551,
+      "step": 8335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7018584420719652e-05,
+      "loss": 2.1688,
+      "step": 8336
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7014630288651644e-05,
+      "loss": 1.7772,
+      "step": 8337
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.701067615658363e-05,
+      "loss": 2.1929,
+      "step": 8338
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.700672202451562e-05,
+      "loss": 1.8707,
+      "step": 8339
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7002767892447607e-05,
+      "loss": 1.948,
+      "step": 8340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.69988137603796e-05,
+      "loss": 2.0736,
+      "step": 8341
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6994859628311585e-05,
+      "loss": 1.6933,
+      "step": 8342
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6990905496243577e-05,
+      "loss": 1.9258,
+      "step": 8343
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6986951364175566e-05,
+      "loss": 1.8937,
+      "step": 8344
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6982997232107555e-05,
+      "loss": 2.0872,
+      "step": 8345
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6979043100039543e-05,
+      "loss": 2.1159,
+      "step": 8346
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6975088967971532e-05,
+      "loss": 1.9284,
+      "step": 8347
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.697113483590352e-05,
+      "loss": 1.7357,
+      "step": 8348
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.696718070383551e-05,
+      "loss": 1.7584,
+      "step": 8349
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.69632265717675e-05,
+      "loss": 2.2303,
+      "step": 8350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6959272439699487e-05,
+      "loss": 1.9087,
+      "step": 8351
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6955318307631476e-05,
+      "loss": 1.8329,
+      "step": 8352
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6951364175563465e-05,
+      "loss": 1.7895,
+      "step": 8353
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6947410043495454e-05,
+      "loss": 1.9829,
+      "step": 8354
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6943455911427443e-05,
+      "loss": 2.0326,
+      "step": 8355
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.693950177935943e-05,
+      "loss": 2.0176,
+      "step": 8356
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.693554764729142e-05,
+      "loss": 1.8742,
+      "step": 8357
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.693159351522341e-05,
+      "loss": 1.7833,
+      "step": 8358
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6927639383155398e-05,
+      "loss": 2.1237,
+      "step": 8359
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6923685251087387e-05,
+      "loss": 1.9271,
+      "step": 8360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6919731119019376e-05,
+      "loss": 1.8242,
+      "step": 8361
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6915776986951365e-05,
+      "loss": 1.9962,
+      "step": 8362
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6911822854883357e-05,
+      "loss": 1.8497,
+      "step": 8363
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6907868722815342e-05,
+      "loss": 1.6825,
+      "step": 8364
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6903914590747334e-05,
+      "loss": 2.0541,
+      "step": 8365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.689996045867932e-05,
+      "loss": 2.2381,
+      "step": 8366
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6896006326611312e-05,
+      "loss": 1.882,
+      "step": 8367
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6892052194543298e-05,
+      "loss": 1.9497,
+      "step": 8368
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.688809806247529e-05,
+      "loss": 2.1429,
+      "step": 8369
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6884143930407275e-05,
+      "loss": 1.9758,
+      "step": 8370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6880189798339267e-05,
+      "loss": 1.8661,
+      "step": 8371
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6876235666271253e-05,
+      "loss": 1.9436,
+      "step": 8372
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6872281534203245e-05,
+      "loss": 1.9072,
+      "step": 8373
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.686832740213523e-05,
+      "loss": 1.9556,
+      "step": 8374
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6864373270067223e-05,
+      "loss": 1.9314,
+      "step": 8375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6860419137999208e-05,
+      "loss": 1.833,
+      "step": 8376
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.68564650059312e-05,
+      "loss": 2.1204,
+      "step": 8377
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6852510873863186e-05,
+      "loss": 2.0468,
+      "step": 8378
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6848556741795178e-05,
+      "loss": 1.8928,
+      "step": 8379
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6844602609727167e-05,
+      "loss": 1.9567,
+      "step": 8380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6840648477659156e-05,
+      "loss": 2.2437,
+      "step": 8381
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6836694345591145e-05,
+      "loss": 1.7955,
+      "step": 8382
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6832740213523133e-05,
+      "loss": 1.7621,
+      "step": 8383
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6828786081455122e-05,
+      "loss": 1.7218,
+      "step": 8384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.682483194938711e-05,
+      "loss": 2.1172,
+      "step": 8385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.68208778173191e-05,
+      "loss": 2.0205,
+      "step": 8386
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.681692368525109e-05,
+      "loss": 2.1052,
+      "step": 8387
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6812969553183077e-05,
+      "loss": 1.9945,
+      "step": 8388
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6809015421115066e-05,
+      "loss": 1.9316,
+      "step": 8389
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6805061289047055e-05,
+      "loss": 2.2239,
+      "step": 8390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6801107156979044e-05,
+      "loss": 2.0859,
+      "step": 8391
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6797153024911033e-05,
+      "loss": 2.0832,
+      "step": 8392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.679319889284302e-05,
+      "loss": 2.0095,
+      "step": 8393
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.678924476077501e-05,
+      "loss": 1.6757,
+      "step": 8394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6785290628707e-05,
+      "loss": 2.2726,
+      "step": 8395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6781336496638988e-05,
+      "loss": 1.8411,
+      "step": 8396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.677738236457098e-05,
+      "loss": 1.9552,
+      "step": 8397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6773428232502966e-05,
+      "loss": 2.1388,
+      "step": 8398
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6769474100434958e-05,
+      "loss": 1.9567,
+      "step": 8399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6765519968366943e-05,
+      "loss": 1.6902,
+      "step": 8400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6761565836298936e-05,
+      "loss": 1.6842,
+      "step": 8401
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.675761170423092e-05,
+      "loss": 2.0637,
+      "step": 8402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6753657572162913e-05,
+      "loss": 2.0031,
+      "step": 8403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.67497034400949e-05,
+      "loss": 1.9819,
+      "step": 8404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.674574930802689e-05,
+      "loss": 1.93,
+      "step": 8405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6741795175958876e-05,
+      "loss": 2.1657,
+      "step": 8406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.673784104389087e-05,
+      "loss": 1.9003,
+      "step": 8407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6733886911822854e-05,
+      "loss": 1.856,
+      "step": 8408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6729932779754846e-05,
+      "loss": 2.0118,
+      "step": 8409
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.672597864768683e-05,
+      "loss": 2.002,
+      "step": 8410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6722024515618824e-05,
+      "loss": 1.9778,
+      "step": 8411
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.671807038355081e-05,
+      "loss": 2.0882,
+      "step": 8412
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.67141162514828e-05,
+      "loss": 1.9643,
+      "step": 8413
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.671016211941479e-05,
+      "loss": 1.7712,
+      "step": 8414
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.670620798734678e-05,
+      "loss": 1.9454,
+      "step": 8415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6702253855278768e-05,
+      "loss": 2.0286,
+      "step": 8416
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6698299723210757e-05,
+      "loss": 1.992,
+      "step": 8417
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6694345591142746e-05,
+      "loss": 1.7888,
+      "step": 8418
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6690391459074735e-05,
+      "loss": 1.9095,
+      "step": 8419
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6686437327006723e-05,
+      "loss": 1.8886,
+      "step": 8420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6682483194938712e-05,
+      "loss": 1.921,
+      "step": 8421
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.66785290628707e-05,
+      "loss": 1.7958,
+      "step": 8422
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.667457493080269e-05,
+      "loss": 2.195,
+      "step": 8423
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.667062079873468e-05,
+      "loss": 1.9159,
+      "step": 8424
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.7897,
+      "step": 8425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6662712534598656e-05,
+      "loss": 1.7258,
+      "step": 8426
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6658758402530645e-05,
+      "loss": 1.9879,
+      "step": 8427
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6654804270462634e-05,
+      "loss": 2.0682,
+      "step": 8428
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6650850138394623e-05,
+      "loss": 2.1074,
+      "step": 8429
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.664689600632661e-05,
+      "loss": 1.9975,
+      "step": 8430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6642941874258604e-05,
+      "loss": 2.2134,
+      "step": 8431
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.663898774219059e-05,
+      "loss": 2.0829,
+      "step": 8432
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.663503361012258e-05,
+      "loss": 2.0785,
+      "step": 8433
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6631079478054567e-05,
+      "loss": 1.7739,
+      "step": 8434
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.662712534598656e-05,
+      "loss": 1.8785,
+      "step": 8435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6623171213918545e-05,
+      "loss": 1.7117,
+      "step": 8436
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6619217081850537e-05,
+      "loss": 1.9928,
+      "step": 8437
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6615262949782522e-05,
+      "loss": 1.6983,
+      "step": 8438
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6611308817714514e-05,
+      "loss": 1.8543,
+      "step": 8439
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.66073546856465e-05,
+      "loss": 1.8531,
+      "step": 8440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6603400553578492e-05,
+      "loss": 1.834,
+      "step": 8441
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6599446421510478e-05,
+      "loss": 1.9637,
+      "step": 8442
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.659549228944247e-05,
+      "loss": 1.8823,
+      "step": 8443
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6591538157374455e-05,
+      "loss": 1.891,
+      "step": 8444
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6587584025306447e-05,
+      "loss": 1.765,
+      "step": 8445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6583629893238433e-05,
+      "loss": 1.7868,
+      "step": 8446
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6579675761170425e-05,
+      "loss": 1.9422,
+      "step": 8447
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.657572162910241e-05,
+      "loss": 2.0437,
+      "step": 8448
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6571767497034403e-05,
+      "loss": 1.8235,
+      "step": 8449
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.656781336496639e-05,
+      "loss": 1.8579,
+      "step": 8450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.656385923289838e-05,
+      "loss": 1.9835,
+      "step": 8451
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.655990510083037e-05,
+      "loss": 2.0137,
+      "step": 8452
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6555950968762358e-05,
+      "loss": 2.1291,
+      "step": 8453
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6551996836694347e-05,
+      "loss": 2.1301,
+      "step": 8454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6548042704626336e-05,
+      "loss": 1.9597,
+      "step": 8455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6544088572558324e-05,
+      "loss": 1.951,
+      "step": 8456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6540134440490313e-05,
+      "loss": 2.1331,
+      "step": 8457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6536180308422302e-05,
+      "loss": 2.0599,
+      "step": 8458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.653222617635429e-05,
+      "loss": 1.943,
+      "step": 8459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.652827204428628e-05,
+      "loss": 1.9619,
+      "step": 8460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.652431791221827e-05,
+      "loss": 1.6485,
+      "step": 8461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6520363780150257e-05,
+      "loss": 1.8005,
+      "step": 8462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6516409648082246e-05,
+      "loss": 1.9377,
+      "step": 8463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6512455516014235e-05,
+      "loss": 1.984,
+      "step": 8464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6508501383946224e-05,
+      "loss": 1.9191,
+      "step": 8465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6504547251878213e-05,
+      "loss": 1.8076,
+      "step": 8466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6500593119810205e-05,
+      "loss": 2.0104,
+      "step": 8467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.649663898774219e-05,
+      "loss": 2.0105,
+      "step": 8468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6492684855674183e-05,
+      "loss": 1.8214,
+      "step": 8469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6488730723606168e-05,
+      "loss": 1.8816,
+      "step": 8470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.648477659153816e-05,
+      "loss": 1.9517,
+      "step": 8471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6480822459470146e-05,
+      "loss": 1.8923,
+      "step": 8472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6476868327402138e-05,
+      "loss": 1.9139,
+      "step": 8473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6472914195334123e-05,
+      "loss": 1.6481,
+      "step": 8474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6468960063266116e-05,
+      "loss": 1.8765,
+      "step": 8475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.64650059311981e-05,
+      "loss": 2.0699,
+      "step": 8476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6461051799130093e-05,
+      "loss": 1.7952,
+      "step": 8477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.645709766706208e-05,
+      "loss": 1.8232,
+      "step": 8478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.645314353499407e-05,
+      "loss": 1.9824,
+      "step": 8479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6449189402926056e-05,
+      "loss": 1.6381,
+      "step": 8480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.644523527085805e-05,
+      "loss": 2.0525,
+      "step": 8481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6441281138790034e-05,
+      "loss": 1.918,
+      "step": 8482
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6437327006722026e-05,
+      "loss": 1.9293,
+      "step": 8483
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6433372874654015e-05,
+      "loss": 2.1252,
+      "step": 8484
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6429418742586004e-05,
+      "loss": 2.042,
+      "step": 8485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6425464610517993e-05,
+      "loss": 1.8628,
+      "step": 8486
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.642151047844998e-05,
+      "loss": 1.9064,
+      "step": 8487
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.641755634638197e-05,
+      "loss": 2.0306,
+      "step": 8488
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.641360221431396e-05,
+      "loss": 1.9981,
+      "step": 8489
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6409648082245948e-05,
+      "loss": 1.9384,
+      "step": 8490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6405693950177937e-05,
+      "loss": 1.7158,
+      "step": 8491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6401739818109926e-05,
+      "loss": 1.6443,
+      "step": 8492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6397785686041914e-05,
+      "loss": 2.0617,
+      "step": 8493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6393831553973903e-05,
+      "loss": 2.0164,
+      "step": 8494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6389877421905892e-05,
+      "loss": 1.8597,
+      "step": 8495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.638592328983788e-05,
+      "loss": 1.859,
+      "step": 8496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.638196915776987e-05,
+      "loss": 1.8427,
+      "step": 8497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.637801502570186e-05,
+      "loss": 2.0956,
+      "step": 8498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6374060893633847e-05,
+      "loss": 2.0329,
+      "step": 8499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6370106761565836e-05,
+      "loss": 1.9894,
+      "step": 8500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.636615262949783e-05,
+      "loss": 2.0317,
+      "step": 8501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6362198497429814e-05,
+      "loss": 1.6923,
+      "step": 8502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6358244365361806e-05,
+      "loss": 2.08,
+      "step": 8503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.635429023329379e-05,
+      "loss": 1.8041,
+      "step": 8504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6350336101225784e-05,
+      "loss": 2.066,
+      "step": 8505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.634638196915777e-05,
+      "loss": 2.086,
+      "step": 8506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.634242783708976e-05,
+      "loss": 2.1068,
+      "step": 8507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6338473705021747e-05,
+      "loss": 1.9495,
+      "step": 8508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.633451957295374e-05,
+      "loss": 1.7841,
+      "step": 8509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6330565440885725e-05,
+      "loss": 1.7587,
+      "step": 8510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6326611308817717e-05,
+      "loss": 1.9382,
+      "step": 8511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6322657176749702e-05,
+      "loss": 1.8843,
+      "step": 8512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6318703044681694e-05,
+      "loss": 2.0449,
+      "step": 8513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.631474891261368e-05,
+      "loss": 1.79,
+      "step": 8514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6310794780545672e-05,
+      "loss": 1.9748,
+      "step": 8515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6306840648477657e-05,
+      "loss": 2.1131,
+      "step": 8516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.630288651640965e-05,
+      "loss": 1.8957,
+      "step": 8517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.629893238434164e-05,
+      "loss": 1.98,
+      "step": 8518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6294978252273627e-05,
+      "loss": 1.9436,
+      "step": 8519
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6291024120205616e-05,
+      "loss": 1.7927,
+      "step": 8520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6287069988137605e-05,
+      "loss": 1.8933,
+      "step": 8521
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6283115856069594e-05,
+      "loss": 1.9876,
+      "step": 8522
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6279161724001583e-05,
+      "loss": 1.8695,
+      "step": 8523
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.627520759193357e-05,
+      "loss": 1.9274,
+      "step": 8524
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.627125345986556e-05,
+      "loss": 1.9538,
+      "step": 8525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.626729932779755e-05,
+      "loss": 1.8568,
+      "step": 8526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6263345195729538e-05,
+      "loss": 1.7373,
+      "step": 8527
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6259391063661527e-05,
+      "loss": 2.1524,
+      "step": 8528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6255436931593516e-05,
+      "loss": 1.9212,
+      "step": 8529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6251482799525504e-05,
+      "loss": 1.9003,
+      "step": 8530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6247528667457493e-05,
+      "loss": 1.7504,
+      "step": 8531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6243574535389482e-05,
+      "loss": 1.9168,
+      "step": 8532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.623962040332147e-05,
+      "loss": 1.8689,
+      "step": 8533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.623566627125346e-05,
+      "loss": 1.8672,
+      "step": 8534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.623171213918545e-05,
+      "loss": 2.1071,
+      "step": 8535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6227758007117437e-05,
+      "loss": 1.8443,
+      "step": 8536
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.622380387504943e-05,
+      "loss": 2.1703,
+      "step": 8537
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6219849742981415e-05,
+      "loss": 2.0378,
+      "step": 8538
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6215895610913407e-05,
+      "loss": 1.8869,
+      "step": 8539
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6211941478845393e-05,
+      "loss": 1.9693,
+      "step": 8540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6207987346777385e-05,
+      "loss": 1.7377,
+      "step": 8541
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.620403321470937e-05,
+      "loss": 1.8348,
+      "step": 8542
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6200079082641363e-05,
+      "loss": 1.9186,
+      "step": 8543
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6196124950573348e-05,
+      "loss": 2.0659,
+      "step": 8544
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.619217081850534e-05,
+      "loss": 1.9615,
+      "step": 8545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6188216686437326e-05,
+      "loss": 1.8918,
+      "step": 8546
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6184262554369318e-05,
+      "loss": 2.1659,
+      "step": 8547
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6180308422301303e-05,
+      "loss": 1.8594,
+      "step": 8548
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6176354290233296e-05,
+      "loss": 1.9827,
+      "step": 8549
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.617240015816528e-05,
+      "loss": 1.9342,
+      "step": 8550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6168446026097273e-05,
+      "loss": 1.8476,
+      "step": 8551
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.616449189402926e-05,
+      "loss": 2.2466,
+      "step": 8552
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.616053776196125e-05,
+      "loss": 2.0988,
+      "step": 8553
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.615658362989324e-05,
+      "loss": 2.0231,
+      "step": 8554
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.615262949782523e-05,
+      "loss": 1.6948,
+      "step": 8555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6148675365757217e-05,
+      "loss": 2.1345,
+      "step": 8556
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6144721233689206e-05,
+      "loss": 1.6558,
+      "step": 8557
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6140767101621195e-05,
+      "loss": 1.9403,
+      "step": 8558
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6136812969553184e-05,
+      "loss": 1.8171,
+      "step": 8559
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6132858837485173e-05,
+      "loss": 2.08,
+      "step": 8560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.612890470541716e-05,
+      "loss": 1.9342,
+      "step": 8561
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.612495057334915e-05,
+      "loss": 1.7545,
+      "step": 8562
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.612099644128114e-05,
+      "loss": 2.0941,
+      "step": 8563
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6117042309213128e-05,
+      "loss": 1.8357,
+      "step": 8564
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6113088177145117e-05,
+      "loss": 1.8766,
+      "step": 8565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6109134045077106e-05,
+      "loss": 1.9339,
+      "step": 8566
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6105179913009094e-05,
+      "loss": 2.1824,
+      "step": 8567
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6101225780941083e-05,
+      "loss": 1.969,
+      "step": 8568
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6097271648873072e-05,
+      "loss": 1.7081,
+      "step": 8569
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.609331751680506e-05,
+      "loss": 2.1088,
+      "step": 8570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6089363384737053e-05,
+      "loss": 1.8176,
+      "step": 8571
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.608540925266904e-05,
+      "loss": 1.9321,
+      "step": 8572
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.608145512060103e-05,
+      "loss": 1.844,
+      "step": 8573
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6077500988533016e-05,
+      "loss": 1.7984,
+      "step": 8574
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.607354685646501e-05,
+      "loss": 1.7462,
+      "step": 8575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6069592724396994e-05,
+      "loss": 1.8407,
+      "step": 8576
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6065638592328986e-05,
+      "loss": 1.7693,
+      "step": 8577
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.606168446026097e-05,
+      "loss": 1.8325,
+      "step": 8578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6057730328192964e-05,
+      "loss": 1.9254,
+      "step": 8579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.605377619612495e-05,
+      "loss": 1.8472,
+      "step": 8580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.604982206405694e-05,
+      "loss": 2.0909,
+      "step": 8581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6045867931988927e-05,
+      "loss": 2.02,
+      "step": 8582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.604191379992092e-05,
+      "loss": 1.8387,
+      "step": 8583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6037959667852905e-05,
+      "loss": 2.0851,
+      "step": 8584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6034005535784897e-05,
+      "loss": 1.6637,
+      "step": 8585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6030051403716882e-05,
+      "loss": 1.8737,
+      "step": 8586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6026097271648874e-05,
+      "loss": 2.0869,
+      "step": 8587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6022143139580863e-05,
+      "loss": 1.855,
+      "step": 8588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6018189007512852e-05,
+      "loss": 1.6659,
+      "step": 8589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.601423487544484e-05,
+      "loss": 1.8825,
+      "step": 8590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.601028074337683e-05,
+      "loss": 1.8811,
+      "step": 8591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.600632661130882e-05,
+      "loss": 1.9953,
+      "step": 8592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.6002372479240807e-05,
+      "loss": 1.8897,
+      "step": 8593
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5998418347172796e-05,
+      "loss": 2.2458,
+      "step": 8594
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5994464215104785e-05,
+      "loss": 2.0803,
+      "step": 8595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5990510083036774e-05,
+      "loss": 1.9389,
+      "step": 8596
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5986555950968763e-05,
+      "loss": 1.9984,
+      "step": 8597
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.598260181890075e-05,
+      "loss": 1.9418,
+      "step": 8598
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.597864768683274e-05,
+      "loss": 2.0886,
+      "step": 8599
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.597469355476473e-05,
+      "loss": 1.6683,
+      "step": 8600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5970739422696718e-05,
+      "loss": 1.789,
+      "step": 8601
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5966785290628707e-05,
+      "loss": 1.8454,
+      "step": 8602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5962831158560696e-05,
+      "loss": 2.0011,
+      "step": 8603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5958877026492684e-05,
+      "loss": 1.9156,
+      "step": 8604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5954922894424677e-05,
+      "loss": 1.9468,
+      "step": 8605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5950968762356662e-05,
+      "loss": 2.1087,
+      "step": 8606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5947014630288654e-05,
+      "loss": 1.9345,
+      "step": 8607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.594306049822064e-05,
+      "loss": 1.9615,
+      "step": 8608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5939106366152632e-05,
+      "loss": 1.9207,
+      "step": 8609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5935152234084617e-05,
+      "loss": 2.0959,
+      "step": 8610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.593119810201661e-05,
+      "loss": 1.8103,
+      "step": 8611
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5927243969948595e-05,
+      "loss": 1.7878,
+      "step": 8612
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5923289837880587e-05,
+      "loss": 1.6171,
+      "step": 8613
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5919335705812573e-05,
+      "loss": 1.6269,
+      "step": 8614
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5915381573744565e-05,
+      "loss": 2.0201,
+      "step": 8615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.591142744167655e-05,
+      "loss": 2.0218,
+      "step": 8616
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5907473309608543e-05,
+      "loss": 1.8544,
+      "step": 8617
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5903519177540528e-05,
+      "loss": 1.9374,
+      "step": 8618
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.589956504547252e-05,
+      "loss": 2.0798,
+      "step": 8619
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5895610913404506e-05,
+      "loss": 1.9038,
+      "step": 8620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5891656781336498e-05,
+      "loss": 2.0451,
+      "step": 8621
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5887702649268483e-05,
+      "loss": 2.0142,
+      "step": 8622
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5883748517200476e-05,
+      "loss": 1.7602,
+      "step": 8623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5879794385132464e-05,
+      "loss": 2.0764,
+      "step": 8624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5875840253064453e-05,
+      "loss": 1.8935,
+      "step": 8625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5871886120996442e-05,
+      "loss": 1.7,
+      "step": 8626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.586793198892843e-05,
+      "loss": 1.7245,
+      "step": 8627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.586397785686042e-05,
+      "loss": 1.9979,
+      "step": 8628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.586002372479241e-05,
+      "loss": 1.9678,
+      "step": 8629
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5856069592724397e-05,
+      "loss": 1.8729,
+      "step": 8630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5852115460656386e-05,
+      "loss": 1.7547,
+      "step": 8631
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5848161328588375e-05,
+      "loss": 1.9347,
+      "step": 8632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5844207196520364e-05,
+      "loss": 1.7317,
+      "step": 8633
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5840253064452353e-05,
+      "loss": 1.8831,
+      "step": 8634
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.583629893238434e-05,
+      "loss": 2.0202,
+      "step": 8635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.583234480031633e-05,
+      "loss": 1.8958,
+      "step": 8636
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.582839066824832e-05,
+      "loss": 1.8785,
+      "step": 8637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5824436536180308e-05,
+      "loss": 1.7819,
+      "step": 8638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5820482404112297e-05,
+      "loss": 1.9714,
+      "step": 8639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5816528272044286e-05,
+      "loss": 2.0357,
+      "step": 8640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5812574139976278e-05,
+      "loss": 2.0946,
+      "step": 8641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5808620007908263e-05,
+      "loss": 1.9099,
+      "step": 8642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5804665875840255e-05,
+      "loss": 1.8579,
+      "step": 8643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.580071174377224e-05,
+      "loss": 1.8196,
+      "step": 8644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5796757611704233e-05,
+      "loss": 1.8833,
+      "step": 8645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.579280347963622e-05,
+      "loss": 1.6611,
+      "step": 8646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.578884934756821e-05,
+      "loss": 1.8413,
+      "step": 8647
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5784895215500196e-05,
+      "loss": 1.8897,
+      "step": 8648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.578094108343219e-05,
+      "loss": 1.953,
+      "step": 8649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5776986951364174e-05,
+      "loss": 1.8853,
+      "step": 8650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5773032819296166e-05,
+      "loss": 1.7944,
+      "step": 8651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.576907868722815e-05,
+      "loss": 1.9594,
+      "step": 8652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5765124555160144e-05,
+      "loss": 1.9478,
+      "step": 8653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5761170423092133e-05,
+      "loss": 1.7058,
+      "step": 8654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.575721629102412e-05,
+      "loss": 1.907,
+      "step": 8655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.575326215895611e-05,
+      "loss": 1.7646,
+      "step": 8656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.57493080268881e-05,
+      "loss": 1.954,
+      "step": 8657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5745353894820088e-05,
+      "loss": 1.8235,
+      "step": 8658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5741399762752077e-05,
+      "loss": 1.8995,
+      "step": 8659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5737445630684066e-05,
+      "loss": 1.8924,
+      "step": 8660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5733491498616054e-05,
+      "loss": 1.8735,
+      "step": 8661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5729537366548043e-05,
+      "loss": 1.9491,
+      "step": 8662
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5725583234480032e-05,
+      "loss": 2.1183,
+      "step": 8663
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.572162910241202e-05,
+      "loss": 2.1199,
+      "step": 8664
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.571767497034401e-05,
+      "loss": 1.6247,
+      "step": 8665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5713720838276e-05,
+      "loss": 1.9987,
+      "step": 8666
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5709766706207987e-05,
+      "loss": 1.9595,
+      "step": 8667
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5705812574139976e-05,
+      "loss": 1.9655,
+      "step": 8668
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5701858442071965e-05,
+      "loss": 1.8847,
+      "step": 8669
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5697904310003954e-05,
+      "loss": 1.8559,
+      "step": 8670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5693950177935943e-05,
+      "loss": 2.0464,
+      "step": 8671
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.568999604586793e-05,
+      "loss": 1.891,
+      "step": 8672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.568604191379992e-05,
+      "loss": 1.8083,
+      "step": 8673
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.568208778173191e-05,
+      "loss": 2.0492,
+      "step": 8674
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.56781336496639e-05,
+      "loss": 1.9924,
+      "step": 8675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5674179517595887e-05,
+      "loss": 2.1901,
+      "step": 8676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.567022538552788e-05,
+      "loss": 2.0848,
+      "step": 8677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5666271253459864e-05,
+      "loss": 2.0276,
+      "step": 8678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5662317121391857e-05,
+      "loss": 2.3723,
+      "step": 8679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5658362989323842e-05,
+      "loss": 1.9604,
+      "step": 8680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5654408857255834e-05,
+      "loss": 1.9114,
+      "step": 8681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5650454725187823e-05,
+      "loss": 1.9373,
+      "step": 8682
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5646500593119812e-05,
+      "loss": 2.0149,
+      "step": 8683
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.56425464610518e-05,
+      "loss": 2.068,
+      "step": 8684
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.563859232898379e-05,
+      "loss": 1.7929,
+      "step": 8685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.563463819691578e-05,
+      "loss": 2.1314,
+      "step": 8686
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5630684064847767e-05,
+      "loss": 2.029,
+      "step": 8687
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5626729932779756e-05,
+      "loss": 2.1609,
+      "step": 8688
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5622775800711745e-05,
+      "loss": 1.7664,
+      "step": 8689
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5618821668643734e-05,
+      "loss": 1.8599,
+      "step": 8690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5614867536575723e-05,
+      "loss": 1.8498,
+      "step": 8691
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.561091340450771e-05,
+      "loss": 1.7846,
+      "step": 8692
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.56069592724397e-05,
+      "loss": 1.9379,
+      "step": 8693
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.560300514037169e-05,
+      "loss": 1.7843,
+      "step": 8694
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5599051008303678e-05,
+      "loss": 2.009,
+      "step": 8695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5595096876235667e-05,
+      "loss": 2.3269,
+      "step": 8696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5591142744167656e-05,
+      "loss": 2.0449,
+      "step": 8697
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5587188612099644e-05,
+      "loss": 1.8552,
+      "step": 8698
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5583234480031633e-05,
+      "loss": 2.0526,
+      "step": 8699
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5579280347963622e-05,
+      "loss": 1.8933,
+      "step": 8700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.557532621589561e-05,
+      "loss": 2.0875,
+      "step": 8701
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.55713720838276e-05,
+      "loss": 1.8378,
+      "step": 8702
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.556741795175959e-05,
+      "loss": 2.0523,
+      "step": 8703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5563463819691577e-05,
+      "loss": 1.8383,
+      "step": 8704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5559509687623566e-05,
+      "loss": 2.0147,
+      "step": 8705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 1.8093,
+      "step": 8706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5551601423487544e-05,
+      "loss": 1.9225,
+      "step": 8707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5547647291419533e-05,
+      "loss": 1.7564,
+      "step": 8708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.554369315935152e-05,
+      "loss": 1.8746,
+      "step": 8709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.553973902728351e-05,
+      "loss": 1.8354,
+      "step": 8710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5535784895215502e-05,
+      "loss": 2.0284,
+      "step": 8711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.553183076314749e-05,
+      "loss": 1.891,
+      "step": 8712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.552787663107948e-05,
+      "loss": 1.8524,
+      "step": 8713
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.552392249901147e-05,
+      "loss": 2.0305,
+      "step": 8714
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5519968366943458e-05,
+      "loss": 2.0085,
+      "step": 8715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5516014234875447e-05,
+      "loss": 2.0288,
+      "step": 8716
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5512060102807435e-05,
+      "loss": 1.8386,
+      "step": 8717
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5508105970739424e-05,
+      "loss": 1.9441,
+      "step": 8718
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5504151838671413e-05,
+      "loss": 1.8329,
+      "step": 8719
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5500197706603402e-05,
+      "loss": 1.7656,
+      "step": 8720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.549624357453539e-05,
+      "loss": 2.1099,
+      "step": 8721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.549228944246738e-05,
+      "loss": 2.0571,
+      "step": 8722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.548833531039937e-05,
+      "loss": 1.8457,
+      "step": 8723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5484381178331357e-05,
+      "loss": 1.7239,
+      "step": 8724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5480427046263346e-05,
+      "loss": 2.0727,
+      "step": 8725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5476472914195335e-05,
+      "loss": 2.1144,
+      "step": 8726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5472518782127324e-05,
+      "loss": 1.7647,
+      "step": 8727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5468564650059313e-05,
+      "loss": 2.0727,
+      "step": 8728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.54646105179913e-05,
+      "loss": 2.0508,
+      "step": 8729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.546065638592329e-05,
+      "loss": 2.021,
+      "step": 8730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.545670225385528e-05,
+      "loss": 1.9243,
+      "step": 8731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5452748121787268e-05,
+      "loss": 1.766,
+      "step": 8732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5448793989719257e-05,
+      "loss": 1.8984,
+      "step": 8733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5444839857651245e-05,
+      "loss": 1.855,
+      "step": 8734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5440885725583234e-05,
+      "loss": 1.9087,
+      "step": 8735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5436931593515223e-05,
+      "loss": 2.0468,
+      "step": 8736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5432977461447212e-05,
+      "loss": 1.7752,
+      "step": 8737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.54290233293792e-05,
+      "loss": 2.0829,
+      "step": 8738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.542506919731119e-05,
+      "loss": 1.7632,
+      "step": 8739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5421115065243182e-05,
+      "loss": 2.1608,
+      "step": 8740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5417160933175167e-05,
+      "loss": 1.5701,
+      "step": 8741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.541320680110716e-05,
+      "loss": 1.9604,
+      "step": 8742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5409252669039145e-05,
+      "loss": 1.9815,
+      "step": 8743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5405298536971137e-05,
+      "loss": 2.1796,
+      "step": 8744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5401344404903126e-05,
+      "loss": 1.8585,
+      "step": 8745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5397390272835115e-05,
+      "loss": 1.6846,
+      "step": 8746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5393436140767104e-05,
+      "loss": 1.893,
+      "step": 8747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5389482008699092e-05,
+      "loss": 1.9545,
+      "step": 8748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.538552787663108e-05,
+      "loss": 1.9174,
+      "step": 8749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.538157374456307e-05,
+      "loss": 1.7255,
+      "step": 8750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.537761961249506e-05,
+      "loss": 1.7775,
+      "step": 8751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5373665480427048e-05,
+      "loss": 1.9949,
+      "step": 8752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5369711348359037e-05,
+      "loss": 1.8961,
+      "step": 8753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5365757216291025e-05,
+      "loss": 1.8497,
+      "step": 8754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5361803084223014e-05,
+      "loss": 1.9046,
+      "step": 8755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5357848952155003e-05,
+      "loss": 1.8386,
+      "step": 8756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5353894820086992e-05,
+      "loss": 1.8764,
+      "step": 8757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.534994068801898e-05,
+      "loss": 2.1763,
+      "step": 8758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.534598655595097e-05,
+      "loss": 2.0664,
+      "step": 8759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.534203242388296e-05,
+      "loss": 2.1059,
+      "step": 8760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5338078291814947e-05,
+      "loss": 1.8969,
+      "step": 8761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5334124159746936e-05,
+      "loss": 1.7889,
+      "step": 8762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5330170027678925e-05,
+      "loss": 1.8683,
+      "step": 8763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5326215895610914e-05,
+      "loss": 1.8758,
+      "step": 8764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5322261763542903e-05,
+      "loss": 1.7398,
+      "step": 8765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.531830763147489e-05,
+      "loss": 1.7736,
+      "step": 8766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.531435349940688e-05,
+      "loss": 1.9374,
+      "step": 8767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5310399367338872e-05,
+      "loss": 1.7788,
+      "step": 8768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5306445235270858e-05,
+      "loss": 1.8496,
+      "step": 8769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.530249110320285e-05,
+      "loss": 1.9185,
+      "step": 8770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5298536971134835e-05,
+      "loss": 1.7347,
+      "step": 8771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5294582839066828e-05,
+      "loss": 1.9214,
+      "step": 8772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5290628706998813e-05,
+      "loss": 1.8154,
+      "step": 8773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5286674574930805e-05,
+      "loss": 1.9282,
+      "step": 8774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.528272044286279e-05,
+      "loss": 2.0852,
+      "step": 8775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5278766310794783e-05,
+      "loss": 2.127,
+      "step": 8776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.527481217872677e-05,
+      "loss": 1.8429,
+      "step": 8777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.527085804665876e-05,
+      "loss": 1.7273,
+      "step": 8778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.526690391459075e-05,
+      "loss": 1.9062,
+      "step": 8779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5262949782522738e-05,
+      "loss": 2.0518,
+      "step": 8780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5258995650454727e-05,
+      "loss": 1.9945,
+      "step": 8781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5255041518386714e-05,
+      "loss": 2.0296,
+      "step": 8782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5251087386318705e-05,
+      "loss": 2.0281,
+      "step": 8783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5247133254250692e-05,
+      "loss": 2.0094,
+      "step": 8784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5243179122182682e-05,
+      "loss": 2.1334,
+      "step": 8785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.523922499011467e-05,
+      "loss": 2.0684,
+      "step": 8786
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.523527085804666e-05,
+      "loss": 1.8031,
+      "step": 8787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5231316725978647e-05,
+      "loss": 1.7964,
+      "step": 8788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5227362593910638e-05,
+      "loss": 1.6777,
+      "step": 8789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5223408461842625e-05,
+      "loss": 1.7513,
+      "step": 8790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5219454329774615e-05,
+      "loss": 1.9378,
+      "step": 8791
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5215500197706603e-05,
+      "loss": 1.892,
+      "step": 8792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5211546065638593e-05,
+      "loss": 1.8182,
+      "step": 8793
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.520759193357058e-05,
+      "loss": 1.8596,
+      "step": 8794
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.520363780150257e-05,
+      "loss": 1.7638,
+      "step": 8795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5199683669434558e-05,
+      "loss": 1.9493,
+      "step": 8796
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5195729537366548e-05,
+      "loss": 1.8334,
+      "step": 8797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5191775405298539e-05,
+      "loss": 1.9994,
+      "step": 8798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5187821273230526e-05,
+      "loss": 2.1101,
+      "step": 8799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5183867141162517e-05,
+      "loss": 1.8129,
+      "step": 8800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5179913009094504e-05,
+      "loss": 1.6925,
+      "step": 8801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5175958877026494e-05,
+      "loss": 2.0465,
+      "step": 8802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5172004744958481e-05,
+      "loss": 1.8355,
+      "step": 8803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5168050612890472e-05,
+      "loss": 1.9105,
+      "step": 8804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5164096480822459e-05,
+      "loss": 1.9528,
+      "step": 8805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.516014234875445e-05,
+      "loss": 2.0034,
+      "step": 8806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5156188216686437e-05,
+      "loss": 1.768,
+      "step": 8807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5152234084618427e-05,
+      "loss": 2.012,
+      "step": 8808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5148279952550414e-05,
+      "loss": 1.7744,
+      "step": 8809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5144325820482405e-05,
+      "loss": 2.0382,
+      "step": 8810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5140371688414392e-05,
+      "loss": 1.9614,
+      "step": 8811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5136417556346382e-05,
+      "loss": 1.9837,
+      "step": 8812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.513246342427837e-05,
+      "loss": 2.0229,
+      "step": 8813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.512850929221036e-05,
+      "loss": 1.7319,
+      "step": 8814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.512455516014235e-05,
+      "loss": 1.8289,
+      "step": 8815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5120601028074338e-05,
+      "loss": 1.9986,
+      "step": 8816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5116646896006328e-05,
+      "loss": 1.8729,
+      "step": 8817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5112692763938315e-05,
+      "loss": 1.9298,
+      "step": 8818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5108738631870306e-05,
+      "loss": 1.8255,
+      "step": 8819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5104784499802293e-05,
+      "loss": 2.1275,
+      "step": 8820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5100830367734284e-05,
+      "loss": 2.0087,
+      "step": 8821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.509687623566627e-05,
+      "loss": 1.9103,
+      "step": 8822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5092922103598261e-05,
+      "loss": 1.9985,
+      "step": 8823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5088967971530248e-05,
+      "loss": 2.0127,
+      "step": 8824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5085013839462239e-05,
+      "loss": 1.7956,
+      "step": 8825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5081059707394226e-05,
+      "loss": 1.9733,
+      "step": 8826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5077105575326217e-05,
+      "loss": 1.8933,
+      "step": 8827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5073151443258204e-05,
+      "loss": 1.8896,
+      "step": 8828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5069197311190194e-05,
+      "loss": 1.7012,
+      "step": 8829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5065243179122181e-05,
+      "loss": 1.9105,
+      "step": 8830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5061289047054172e-05,
+      "loss": 1.6733,
+      "step": 8831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5057334914986162e-05,
+      "loss": 1.9268,
+      "step": 8832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.505338078291815e-05,
+      "loss": 1.8733,
+      "step": 8833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.504942665085014e-05,
+      "loss": 2.0054,
+      "step": 8834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5045472518782127e-05,
+      "loss": 1.9757,
+      "step": 8835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5041518386714118e-05,
+      "loss": 2.0997,
+      "step": 8836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5037564254646105e-05,
+      "loss": 1.9675,
+      "step": 8837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5033610122578095e-05,
+      "loss": 1.9733,
+      "step": 8838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5029655990510082e-05,
+      "loss": 1.8725,
+      "step": 8839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5025701858442073e-05,
+      "loss": 1.9381,
+      "step": 8840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.502174772637406e-05,
+      "loss": 1.8423,
+      "step": 8841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.501779359430605e-05,
+      "loss": 1.9828,
+      "step": 8842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5013839462238038e-05,
+      "loss": 2.0513,
+      "step": 8843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5009885330170028e-05,
+      "loss": 1.6956,
+      "step": 8844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5005931198102015e-05,
+      "loss": 2.163,
+      "step": 8845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5001977066034006e-05,
+      "loss": 1.9725,
+      "step": 8846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4998022933965993e-05,
+      "loss": 1.853,
+      "step": 8847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4994068801897984e-05,
+      "loss": 1.9764,
+      "step": 8848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4990114669829974e-05,
+      "loss": 1.8068,
+      "step": 8849
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4986160537761961e-05,
+      "loss": 2.0944,
+      "step": 8850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4982206405693952e-05,
+      "loss": 1.9005,
+      "step": 8851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4978252273625939e-05,
+      "loss": 1.7444,
+      "step": 8852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.497429814155793e-05,
+      "loss": 1.9965,
+      "step": 8853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4970344009489917e-05,
+      "loss": 1.9702,
+      "step": 8854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4966389877421907e-05,
+      "loss": 1.904,
+      "step": 8855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4962435745353894e-05,
+      "loss": 2.0071,
+      "step": 8856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4958481613285885e-05,
+      "loss": 2.057,
+      "step": 8857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4954527481217872e-05,
+      "loss": 2.0429,
+      "step": 8858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4950573349149862e-05,
+      "loss": 2.0752,
+      "step": 8859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.494661921708185e-05,
+      "loss": 2.2275,
+      "step": 8860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.494266508501384e-05,
+      "loss": 1.9019,
+      "step": 8861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4938710952945827e-05,
+      "loss": 1.983,
+      "step": 8862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4934756820877818e-05,
+      "loss": 1.7355,
+      "step": 8863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4930802688809807e-05,
+      "loss": 1.7336,
+      "step": 8864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4926848556741795e-05,
+      "loss": 1.9396,
+      "step": 8865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4922894424673786e-05,
+      "loss": 1.7804,
+      "step": 8866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4918940292605773e-05,
+      "loss": 1.9046,
+      "step": 8867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4914986160537764e-05,
+      "loss": 1.9164,
+      "step": 8868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.491103202846975e-05,
+      "loss": 1.9688,
+      "step": 8869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4907077896401741e-05,
+      "loss": 1.8937,
+      "step": 8870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4903123764333728e-05,
+      "loss": 1.8901,
+      "step": 8871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4899169632265719e-05,
+      "loss": 2.1837,
+      "step": 8872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4895215500197706e-05,
+      "loss": 1.8371,
+      "step": 8873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4891261368129697e-05,
+      "loss": 1.8259,
+      "step": 8874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4887307236061684e-05,
+      "loss": 1.941,
+      "step": 8875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4883353103993674e-05,
+      "loss": 1.918,
+      "step": 8876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4879398971925661e-05,
+      "loss": 1.7616,
+      "step": 8877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4875444839857652e-05,
+      "loss": 1.9865,
+      "step": 8878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.487149070778964e-05,
+      "loss": 1.7152,
+      "step": 8879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.486753657572163e-05,
+      "loss": 2.0905,
+      "step": 8880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4863582443653618e-05,
+      "loss": 2.1713,
+      "step": 8881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4859628311585607e-05,
+      "loss": 2.1302,
+      "step": 8882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4855674179517596e-05,
+      "loss": 1.9886,
+      "step": 8883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4851720047449585e-05,
+      "loss": 2.0698,
+      "step": 8884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4847765915381575e-05,
+      "loss": 2.0096,
+      "step": 8885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4843811783313562e-05,
+      "loss": 1.9311,
+      "step": 8886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4839857651245553e-05,
+      "loss": 1.8825,
+      "step": 8887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.483590351917754e-05,
+      "loss": 1.9235,
+      "step": 8888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.483194938710953e-05,
+      "loss": 1.9299,
+      "step": 8889
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4827995255041518e-05,
+      "loss": 1.8353,
+      "step": 8890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4824041122973508e-05,
+      "loss": 2.0601,
+      "step": 8891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4820086990905495e-05,
+      "loss": 1.9347,
+      "step": 8892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4816132858837486e-05,
+      "loss": 1.8541,
+      "step": 8893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4812178726769475e-05,
+      "loss": 1.8219,
+      "step": 8894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4808224594701464e-05,
+      "loss": 1.9928,
+      "step": 8895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4804270462633452e-05,
+      "loss": 2.2481,
+      "step": 8896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4800316330565441e-05,
+      "loss": 1.6565,
+      "step": 8897
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.479636219849743e-05,
+      "loss": 1.8798,
+      "step": 8898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4792408066429419e-05,
+      "loss": 2.2092,
+      "step": 8899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4788453934361408e-05,
+      "loss": 1.9208,
+      "step": 8900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4784499802293397e-05,
+      "loss": 1.8722,
+      "step": 8901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4780545670225387e-05,
+      "loss": 1.83,
+      "step": 8902
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4776591538157374e-05,
+      "loss": 1.6958,
+      "step": 8903
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4772637406089365e-05,
+      "loss": 2.0617,
+      "step": 8904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4768683274021352e-05,
+      "loss": 2.1196,
+      "step": 8905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4764729141953342e-05,
+      "loss": 2.1766,
+      "step": 8906
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4760775009885331e-05,
+      "loss": 1.899,
+      "step": 8907
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.475682087781732e-05,
+      "loss": 1.868,
+      "step": 8908
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4752866745749309e-05,
+      "loss": 1.8994,
+      "step": 8909
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4748912613681298e-05,
+      "loss": 1.8856,
+      "step": 8910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4744958481613286e-05,
+      "loss": 1.7979,
+      "step": 8911
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4741004349545275e-05,
+      "loss": 1.8997,
+      "step": 8912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4737050217477264e-05,
+      "loss": 2.0315,
+      "step": 8913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4733096085409253e-05,
+      "loss": 1.8855,
+      "step": 8914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.4729141953341242e-05,
+      "loss": 1.7983,
+      "step": 8915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.472518782127323e-05,
+      "loss": 1.889,
+      "step": 8916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.472123368920522e-05,
+      "loss": 1.8052,
+      "step": 8917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4717279557137208e-05,
+      "loss": 1.9518,
+      "step": 8918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4713325425069199e-05,
+      "loss": 1.8424,
+      "step": 8919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4709371293001186e-05,
+      "loss": 1.7863,
+      "step": 8920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4705417160933176e-05,
+      "loss": 1.8285,
+      "step": 8921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4701463028865165e-05,
+      "loss": 1.7967,
+      "step": 8922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4697508896797154e-05,
+      "loss": 2.0323,
+      "step": 8923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4693554764729143e-05,
+      "loss": 1.9502,
+      "step": 8924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4689600632661132e-05,
+      "loss": 1.8531,
+      "step": 8925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.468564650059312e-05,
+      "loss": 1.9893,
+      "step": 8926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.468169236852511e-05,
+      "loss": 1.8171,
+      "step": 8927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4677738236457098e-05,
+      "loss": 1.7939,
+      "step": 8928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4673784104389087e-05,
+      "loss": 1.8887,
+      "step": 8929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4669829972321076e-05,
+      "loss": 2.0247,
+      "step": 8930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4665875840253065e-05,
+      "loss": 1.8258,
+      "step": 8931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4661921708185054e-05,
+      "loss": 1.7738,
+      "step": 8932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4657967576117042e-05,
+      "loss": 2.1983,
+      "step": 8933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4654013444049031e-05,
+      "loss": 1.7679,
+      "step": 8934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4650059311981022e-05,
+      "loss": 1.7028,
+      "step": 8935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.464610517991301e-05,
+      "loss": 1.8091,
+      "step": 8936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4642151047845e-05,
+      "loss": 2.0781,
+      "step": 8937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4638196915776988e-05,
+      "loss": 1.9151,
+      "step": 8938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4634242783708977e-05,
+      "loss": 1.8436,
+      "step": 8939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4630288651640966e-05,
+      "loss": 1.7298,
+      "step": 8940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4626334519572955e-05,
+      "loss": 2.0927,
+      "step": 8941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4622380387504944e-05,
+      "loss": 1.9204,
+      "step": 8942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4618426255436932e-05,
+      "loss": 1.9019,
+      "step": 8943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4614472123368921e-05,
+      "loss": 1.9984,
+      "step": 8944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.461051799130091e-05,
+      "loss": 1.8935,
+      "step": 8945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4606563859232899e-05,
+      "loss": 1.8542,
+      "step": 8946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4602609727164888e-05,
+      "loss": 1.9514,
+      "step": 8947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4598655595096876e-05,
+      "loss": 1.9169,
+      "step": 8948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4594701463028865e-05,
+      "loss": 1.8469,
+      "step": 8949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4590747330960856e-05,
+      "loss": 1.9804,
+      "step": 8950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4586793198892843e-05,
+      "loss": 1.9136,
+      "step": 8951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4582839066824833e-05,
+      "loss": 2.0784,
+      "step": 8952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4578884934756822e-05,
+      "loss": 1.9049,
+      "step": 8953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4574930802688811e-05,
+      "loss": 1.9463,
+      "step": 8954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.45709766706208e-05,
+      "loss": 1.6192,
+      "step": 8955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4567022538552789e-05,
+      "loss": 1.7569,
+      "step": 8956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4563068406484778e-05,
+      "loss": 1.9037,
+      "step": 8957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4559114274416766e-05,
+      "loss": 2.2548,
+      "step": 8958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4555160142348755e-05,
+      "loss": 1.9832,
+      "step": 8959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4551206010280744e-05,
+      "loss": 2.0053,
+      "step": 8960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4547251878212733e-05,
+      "loss": 1.8585,
+      "step": 8961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4543297746144722e-05,
+      "loss": 1.8225,
+      "step": 8962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.453934361407671e-05,
+      "loss": 1.9292,
+      "step": 8963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.45353894820087e-05,
+      "loss": 1.9726,
+      "step": 8964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.453143534994069e-05,
+      "loss": 1.8358,
+      "step": 8965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4527481217872677e-05,
+      "loss": 1.7741,
+      "step": 8966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4523527085804668e-05,
+      "loss": 1.9459,
+      "step": 8967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4519572953736655e-05,
+      "loss": 1.8918,
+      "step": 8968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4515618821668645e-05,
+      "loss": 1.8582,
+      "step": 8969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4511664689600634e-05,
+      "loss": 1.8664,
+      "step": 8970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4507710557532623e-05,
+      "loss": 1.9469,
+      "step": 8971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4503756425464612e-05,
+      "loss": 1.7956,
+      "step": 8972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.44998022933966e-05,
+      "loss": 1.9752,
+      "step": 8973
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.449584816132859e-05,
+      "loss": 1.9667,
+      "step": 8974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4491894029260578e-05,
+      "loss": 1.8374,
+      "step": 8975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4487939897192567e-05,
+      "loss": 1.9742,
+      "step": 8976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4483985765124556e-05,
+      "loss": 1.9585,
+      "step": 8977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4480031633056546e-05,
+      "loss": 1.8423,
+      "step": 8978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4476077500988534e-05,
+      "loss": 1.9484,
+      "step": 8979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4472123368920524e-05,
+      "loss": 1.9692,
+      "step": 8980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4468169236852511e-05,
+      "loss": 2.0303,
+      "step": 8981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4464215104784502e-05,
+      "loss": 1.8956,
+      "step": 8982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4460260972716489e-05,
+      "loss": 1.888,
+      "step": 8983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.445630684064848e-05,
+      "loss": 1.9923,
+      "step": 8984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4452352708580466e-05,
+      "loss": 2.0746,
+      "step": 8985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4448398576512457e-05,
+      "loss": 2.0086,
+      "step": 8986
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4444444444444444e-05,
+      "loss": 1.9975,
+      "step": 8987
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4440490312376435e-05,
+      "loss": 2.168,
+      "step": 8988
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4436536180308423e-05,
+      "loss": 1.9508,
+      "step": 8989
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4432582048240412e-05,
+      "loss": 1.8277,
+      "step": 8990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4428627916172401e-05,
+      "loss": 1.9032,
+      "step": 8991
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.442467378410439e-05,
+      "loss": 2.0434,
+      "step": 8992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.442071965203638e-05,
+      "loss": 1.9306,
+      "step": 8993
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4416765519968368e-05,
+      "loss": 1.7875,
+      "step": 8994
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4412811387900358e-05,
+      "loss": 1.8522,
+      "step": 8995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4408857255832345e-05,
+      "loss": 1.8912,
+      "step": 8996
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4404903123764336e-05,
+      "loss": 1.6437,
+      "step": 8997
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4400948991696323e-05,
+      "loss": 2.1181,
+      "step": 8998
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4396994859628313e-05,
+      "loss": 1.7846,
+      "step": 8999
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.43930407275603e-05,
+      "loss": 1.8538,
+      "step": 9000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4389086595492291e-05,
+      "loss": 1.7722,
+      "step": 9001
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4385132463424278e-05,
+      "loss": 1.9442,
+      "step": 9002
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4381178331356269e-05,
+      "loss": 1.7872,
+      "step": 9003
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4377224199288256e-05,
+      "loss": 1.8702,
+      "step": 9004
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4373270067220246e-05,
+      "loss": 2.201,
+      "step": 9005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4369315935152235e-05,
+      "loss": 2.0224,
+      "step": 9006
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4365361803084224e-05,
+      "loss": 1.9659,
+      "step": 9007
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4361407671016215e-05,
+      "loss": 1.8212,
+      "step": 9008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4357453538948202e-05,
+      "loss": 1.985,
+      "step": 9009
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4353499406880192e-05,
+      "loss": 2.0583,
+      "step": 9010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.434954527481218e-05,
+      "loss": 1.8767,
+      "step": 9011
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.434559114274417e-05,
+      "loss": 1.7698,
+      "step": 9012
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4341637010676157e-05,
+      "loss": 2.0397,
+      "step": 9013
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4337682878608148e-05,
+      "loss": 2.0628,
+      "step": 9014
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4333728746540135e-05,
+      "loss": 1.8903,
+      "step": 9015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4329774614472125e-05,
+      "loss": 2.0122,
+      "step": 9016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4325820482404112e-05,
+      "loss": 1.8161,
+      "step": 9017
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4321866350336103e-05,
+      "loss": 1.9278,
+      "step": 9018
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.431791221826809e-05,
+      "loss": 1.9208,
+      "step": 9019
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.431395808620008e-05,
+      "loss": 1.7861,
+      "step": 9020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4310003954132068e-05,
+      "loss": 1.8893,
+      "step": 9021
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4306049822064058e-05,
+      "loss": 1.862,
+      "step": 9022
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4302095689996049e-05,
+      "loss": 1.9428,
+      "step": 9023
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4298141557928036e-05,
+      "loss": 1.8743,
+      "step": 9024
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4294187425860026e-05,
+      "loss": 1.787,
+      "step": 9025
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4290233293792013e-05,
+      "loss": 1.9739,
+      "step": 9026
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4286279161724004e-05,
+      "loss": 1.9576,
+      "step": 9027
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4282325029655991e-05,
+      "loss": 1.9594,
+      "step": 9028
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4278370897587982e-05,
+      "loss": 1.9779,
+      "step": 9029
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4274416765519969e-05,
+      "loss": 2.2584,
+      "step": 9030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.427046263345196e-05,
+      "loss": 2.2427,
+      "step": 9031
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4266508501383946e-05,
+      "loss": 1.8677,
+      "step": 9032
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4262554369315937e-05,
+      "loss": 1.893,
+      "step": 9033
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4258600237247924e-05,
+      "loss": 1.9877,
+      "step": 9034
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4254646105179915e-05,
+      "loss": 2.2404,
+      "step": 9035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4250691973111902e-05,
+      "loss": 1.846,
+      "step": 9036
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4246737841043892e-05,
+      "loss": 2.0308,
+      "step": 9037
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.424278370897588e-05,
+      "loss": 1.9033,
+      "step": 9038
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.423882957690787e-05,
+      "loss": 2.2397,
+      "step": 9039
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.423487544483986e-05,
+      "loss": 1.8464,
+      "step": 9040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4230921312771848e-05,
+      "loss": 1.8008,
+      "step": 9041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4226967180703838e-05,
+      "loss": 1.7722,
+      "step": 9042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4223013048635825e-05,
+      "loss": 1.8476,
+      "step": 9043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4219058916567816e-05,
+      "loss": 1.9434,
+      "step": 9044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4215104784499803e-05,
+      "loss": 1.9139,
+      "step": 9045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4211150652431793e-05,
+      "loss": 2.0052,
+      "step": 9046
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.420719652036378e-05,
+      "loss": 1.7394,
+      "step": 9047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4203242388295771e-05,
+      "loss": 1.8137,
+      "step": 9048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4199288256227758e-05,
+      "loss": 1.8001,
+      "step": 9049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4195334124159749e-05,
+      "loss": 1.9957,
+      "step": 9050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4191379992091736e-05,
+      "loss": 1.7451,
+      "step": 9051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4187425860023726e-05,
+      "loss": 1.5733,
+      "step": 9052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4183471727955713e-05,
+      "loss": 1.7446,
+      "step": 9053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4179517595887704e-05,
+      "loss": 1.8423,
+      "step": 9054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4175563463819691e-05,
+      "loss": 1.9818,
+      "step": 9055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4171609331751682e-05,
+      "loss": 2.0671,
+      "step": 9056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4167655199683672e-05,
+      "loss": 1.9358,
+      "step": 9057
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.416370106761566e-05,
+      "loss": 1.8183,
+      "step": 9058
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.415974693554765e-05,
+      "loss": 1.9308,
+      "step": 9059
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4155792803479637e-05,
+      "loss": 2.2081,
+      "step": 9060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4151838671411627e-05,
+      "loss": 1.8094,
+      "step": 9061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4147884539343615e-05,
+      "loss": 1.8591,
+      "step": 9062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4143930407275605e-05,
+      "loss": 2.1317,
+      "step": 9063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4139976275207592e-05,
+      "loss": 1.9019,
+      "step": 9064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4136022143139583e-05,
+      "loss": 1.8153,
+      "step": 9065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.413206801107157e-05,
+      "loss": 1.783,
+      "step": 9066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.412811387900356e-05,
+      "loss": 1.8996,
+      "step": 9067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4124159746935548e-05,
+      "loss": 1.8815,
+      "step": 9068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4120205614867538e-05,
+      "loss": 1.9411,
+      "step": 9069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4116251482799525e-05,
+      "loss": 1.8404,
+      "step": 9070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4112297350731516e-05,
+      "loss": 1.9619,
+      "step": 9071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4108343218663503e-05,
+      "loss": 1.831,
+      "step": 9072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4104389086595493e-05,
+      "loss": 1.8174,
+      "step": 9073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.410043495452748e-05,
+      "loss": 1.8978,
+      "step": 9074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4096480822459471e-05,
+      "loss": 1.9257,
+      "step": 9075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4092526690391462e-05,
+      "loss": 1.9819,
+      "step": 9076
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4088572558323449e-05,
+      "loss": 1.9854,
+      "step": 9077
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.408461842625544e-05,
+      "loss": 1.7723,
+      "step": 9078
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4080664294187426e-05,
+      "loss": 1.9431,
+      "step": 9079
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4076710162119417e-05,
+      "loss": 1.9263,
+      "step": 9080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4072756030051404e-05,
+      "loss": 2.1651,
+      "step": 9081
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4068801897983395e-05,
+      "loss": 1.8278,
+      "step": 9082
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4064847765915382e-05,
+      "loss": 2.2272,
+      "step": 9083
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4060893633847372e-05,
+      "loss": 1.7757,
+      "step": 9084
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.405693950177936e-05,
+      "loss": 2.1622,
+      "step": 9085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.405298536971135e-05,
+      "loss": 1.5877,
+      "step": 9086
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4049031237643337e-05,
+      "loss": 1.7979,
+      "step": 9087
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4045077105575328e-05,
+      "loss": 2.0256,
+      "step": 9088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4041122973507315e-05,
+      "loss": 1.9515,
+      "step": 9089
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4037168841439305e-05,
+      "loss": 2.2012,
+      "step": 9090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4033214709371292e-05,
+      "loss": 1.953,
+      "step": 9091
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4029260577303283e-05,
+      "loss": 1.8419,
+      "step": 9092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4025306445235273e-05,
+      "loss": 1.9895,
+      "step": 9093
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.402135231316726e-05,
+      "loss": 2.1874,
+      "step": 9094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4017398181099251e-05,
+      "loss": 1.9687,
+      "step": 9095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4013444049031238e-05,
+      "loss": 1.9809,
+      "step": 9096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4009489916963229e-05,
+      "loss": 1.8833,
+      "step": 9097
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4005535784895216e-05,
+      "loss": 1.9446,
+      "step": 9098
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4001581652827206e-05,
+      "loss": 1.8269,
+      "step": 9099
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.3997627520759193e-05,
+      "loss": 1.9621,
+      "step": 9100
+    }
+  ],
+  "max_steps": 12645,
+  "num_train_epochs": 1,
+  "total_flos": 1.0141597324384666e+17,
+  "trial_name": null,
+  "trial_params": null
+}