diff --git "a/checkpoint-9976/trainer_state.json" "b/checkpoint-9976/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-9976/trainer_state.json"
@@ -0,0 +1,59875 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0129975629569457,
+  "eval_steps": 500,
+  "global_step": 9976,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 0.9611,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6909,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-05,
+      "loss": 1.0511,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 0.8962,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 0.9467,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-05,
+      "loss": 0.6699,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2e-05,
+      "loss": 0.899,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8e-05,
+      "loss": 0.8717,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 0.6804,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 1.1548,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999999904545704e-05,
+      "loss": 1.0779,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999999618182824e-05,
+      "loss": 1.2244,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999999914091136e-05,
+      "loss": 0.9183,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999999847273132e-05,
+      "loss": 0.842,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999999761364269e-05,
+      "loss": 1.057,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999996563645495e-05,
+      "loss": 1.0065,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999999532273973e-05,
+      "loss": 0.2884,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999999389092541e-05,
+      "loss": 1.0586,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999999226820254e-05,
+      "loss": 1.1559,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999999045457113e-05,
+      "loss": 0.3994,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999988450031195e-05,
+      "loss": 0.6816,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999998625458275e-05,
+      "loss": 1.3041,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999998386822581e-05,
+      "loss": 1.0317,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999981290960374e-05,
+      "loss": 0.4954,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999978522786466e-05,
+      "loss": 0.9628,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999997556370412e-05,
+      "loss": 0.2448,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999997241371333e-05,
+      "loss": 1.1264,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999969072814134e-05,
+      "loss": 0.5052,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999965541006556e-05,
+      "loss": 0.5601,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999996181829059e-05,
+      "loss": 0.6859,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.99999579046663e-05,
+      "loss": 1.6138,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999995380013368e-05,
+      "loss": 0.5146,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999994950469277e-05,
+      "loss": 0.3252,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999994501834359e-05,
+      "loss": 1.9287,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999994034108618e-05,
+      "loss": 0.9413,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999993547292055e-05,
+      "loss": 0.6465,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999993041384675e-05,
+      "loss": 1.1896,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999992516386481e-05,
+      "loss": 0.9641,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999919722974753e-05,
+      "loss": 0.8682,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999991409117662e-05,
+      "loss": 0.4005,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999990826847045e-05,
+      "loss": 0.9974,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999902254856276e-05,
+      "loss": 0.6602,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999896050334123e-05,
+      "loss": 0.3048,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999988965490406e-05,
+      "loss": 0.8209,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999883068566106e-05,
+      "loss": 0.7759,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999876291320315e-05,
+      "loss": 0.3947,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999869323166715e-05,
+      "loss": 0.4532,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999986216410537e-05,
+      "loss": 0.4846,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.99998548141363e-05,
+      "loss": 0.6761,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999847273259574e-05,
+      "loss": 0.5133,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999839541475225e-05,
+      "loss": 1.4704,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999831618783324e-05,
+      "loss": 1.3765,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99998235051839e-05,
+      "loss": 0.7539,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999981520067701e-05,
+      "loss": 0.9062,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999980670526271e-05,
+      "loss": 0.7255,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999798018941055e-05,
+      "loss": 0.2341,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999978914171209e-05,
+      "loss": 0.6138,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999978007357589e-05,
+      "loss": 1.0273,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999770814532495e-05,
+      "loss": 1.0243,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999761364581975e-05,
+      "loss": 0.7568,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999751723724385e-05,
+      "loss": 0.7743,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999741891959785e-05,
+      "loss": 0.205,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999973186928825e-05,
+      "loss": 0.8313,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999721655709826e-05,
+      "loss": 0.5768,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999711251224576e-05,
+      "loss": 0.5365,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999970065583259e-05,
+      "loss": 1.1081,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999968986953392e-05,
+      "loss": 0.588,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999967889232863e-05,
+      "loss": 0.5623,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999667724216806e-05,
+      "loss": 1.3136,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999965636519849e-05,
+      "loss": 0.4415,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99996448152738e-05,
+      "loss": 0.2088,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999963307444276e-05,
+      "loss": 0.884,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999962114270548e-05,
+      "loss": 1.1938,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999609020062015e-05,
+      "loss": 0.7319,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999959670651246e-05,
+      "loss": 0.4003,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999958420205688e-05,
+      "loss": 0.4763,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999957150669536e-05,
+      "loss": 0.5517,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999955862042798e-05,
+      "loss": 0.9328,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999545543254816e-05,
+      "loss": 0.7181,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999532275175954e-05,
+      "loss": 0.9047,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999951881619149e-05,
+      "loss": 0.7952,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999505166301496e-05,
+      "loss": 1.162,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999491325506063e-05,
+      "loss": 0.2519,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999947729380529e-05,
+      "loss": 0.9357,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999463071199244e-05,
+      "loss": 0.5643,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999944865768802e-05,
+      "loss": 1.2378,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999943405327173e-05,
+      "loss": 1.237,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999941925795046e-05,
+      "loss": 0.6347,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999404271724284e-05,
+      "loss": 0.9233,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999938909459332e-05,
+      "loss": 0.5314,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999373726557655e-05,
+      "loss": 0.861,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999358167617383e-05,
+      "loss": 1.0264,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999934241777261e-05,
+      "loss": 0.2721,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999326477023434e-05,
+      "loss": 1.1545,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999931034536997e-05,
+      "loss": 1.0487,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999294022812295e-05,
+      "loss": 1.117,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999927750935052e-05,
+      "loss": 1.0498,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999260804984765e-05,
+      "loss": 0.6774,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999924390971511e-05,
+      "loss": 0.2515,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999226823541706e-05,
+      "loss": 0.6544,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999920954646462e-05,
+      "loss": 0.9319,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999919207848396e-05,
+      "loss": 0.8882,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999174419599877e-05,
+      "loss": 0.3299,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999915656981246e-05,
+      "loss": 1.0048,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999138529121814e-05,
+      "loss": 0.3582,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999912029752807e-05,
+      "loss": 0.3013,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999101875031325e-05,
+      "loss": 0.4037,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999908326163172e-05,
+      "loss": 0.7843,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999906445732935e-05,
+      "loss": 1.2967,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999904546212436e-05,
+      "loss": 0.6817,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999902627601685e-05,
+      "loss": 0.6538,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999006899006946e-05,
+      "loss": 0.4924,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998987331094775e-05,
+      "loss": 0.2237,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998967572280466e-05,
+      "loss": 0.5863,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999894762256414e-05,
+      "loss": 1.1475,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999892748194591e-05,
+      "loss": 0.7886,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998907150425935e-05,
+      "loss": 0.4083,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998886628004316e-05,
+      "loss": 0.2351,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99988659146812e-05,
+      "loss": 0.4066,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999884501045672e-05,
+      "loss": 0.7706,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999882391533099e-05,
+      "loss": 0.99,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998802629304156e-05,
+      "loss": 1.1271,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998781152376355e-05,
+      "loss": 1.2219,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999875948454773e-05,
+      "loss": 1.3663,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998737625818414e-05,
+      "loss": 0.5026,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999871557618853e-05,
+      "loss": 0.6587,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998693335658245e-05,
+      "loss": 0.5226,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998670904227685e-05,
+      "loss": 0.6075,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999864828189699e-05,
+      "loss": 1.0565,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999862546866632e-05,
+      "loss": 0.5124,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998602464535796e-05,
+      "loss": 0.5832,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999857926950559e-05,
+      "loss": 1.0389,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998555883575834e-05,
+      "loss": 1.1801,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999853230674668e-05,
+      "loss": 0.3315,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999850853901827e-05,
+      "loss": 1.2234,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999848458039078e-05,
+      "loss": 0.4588,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999846043086433e-05,
+      "loss": 0.6374,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999843609043909e-05,
+      "loss": 0.9688,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999841155911523e-05,
+      "loss": 0.8913,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999838683689288e-05,
+      "loss": 0.3771,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998361923772206e-05,
+      "loss": 0.973,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999833681975338e-05,
+      "loss": 0.5952,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998311524836543e-05,
+      "loss": 1.0392,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998286039021866e-05,
+      "loss": 0.233,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99982603623095e-05,
+      "loss": 0.5669,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999823449469963e-05,
+      "loss": 1.0502,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99982084361924e-05,
+      "loss": 0.2361,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999818218678798e-05,
+      "loss": 0.3378,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9998155746486546e-05,
+      "loss": 1.2308,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999812911528826e-05,
+      "loss": 0.7853,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999810229319329e-05,
+      "loss": 0.9563,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9998075280201815e-05,
+      "loss": 0.9404,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9998048076314e-05,
+      "loss": 0.5781,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999802068153001e-05,
+      "loss": 0.3761,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999799309585003e-05,
+      "loss": 1.2978,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997965319274244e-05,
+      "loss": 0.3358,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997937351802814e-05,
+      "loss": 0.7192,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999790919343592e-05,
+      "loss": 0.543,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997880844173746e-05,
+      "loss": 0.456,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999785230401647e-05,
+      "loss": 0.6113,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999782357296426e-05,
+      "loss": 0.6918,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999779465101733e-05,
+      "loss": 0.811,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999776553817584e-05,
+      "loss": 1.1079,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999773623443998e-05,
+      "loss": 0.7662,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997706739809946e-05,
+      "loss": 0.64,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999767705428591e-05,
+      "loss": 0.5964,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997647177868074e-05,
+      "loss": 0.7939,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999761711055663e-05,
+      "loss": 0.3607,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999758685235174e-05,
+      "loss": 1.1062,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999755640325364e-05,
+      "loss": 0.689,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99975257632625e-05,
+      "loss": 0.4168,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99974949323785e-05,
+      "loss": 0.7292,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999746391060188e-05,
+      "loss": 0.4276,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997432697932784e-05,
+      "loss": 1.1441,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999740129437146e-05,
+      "loss": 1.1062,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997369699918074e-05,
+      "loss": 0.5886,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999733791457284e-05,
+      "loss": 0.7991,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997305938335966e-05,
+      "loss": 0.7667,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997273771207644e-05,
+      "loss": 1.0188,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999724141318808e-05,
+      "loss": 0.912,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999720886427749e-05,
+      "loss": 0.7402,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999717612447606e-05,
+      "loss": 0.5352,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997143193784035e-05,
+      "loss": 0.2948,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997110072201586e-05,
+      "loss": 0.7372,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997076759728946e-05,
+      "loss": 0.4266,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997043256366325e-05,
+      "loss": 0.4558,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999700956211392e-05,
+      "loss": 0.307,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999697567697196e-05,
+      "loss": 0.3302,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999694160094067e-05,
+      "loss": 0.7018,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996907334020254e-05,
+      "loss": 0.5692,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999687287621092e-05,
+      "loss": 0.3531,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99968382275129e-05,
+      "loss": 0.8143,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999680338792642e-05,
+      "loss": 0.7237,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999676835745169e-05,
+      "loss": 1.2443,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999673313608894e-05,
+      "loss": 0.5115,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999669772383839e-05,
+      "loss": 0.5782,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999666212070027e-05,
+      "loss": 0.853,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996626326674796e-05,
+      "loss": 0.7049,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999659034176221e-05,
+      "loss": 1.0725,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999655416596273e-05,
+      "loss": 1.0408,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999651779927659e-05,
+      "loss": 0.7292,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999648124170403e-05,
+      "loss": 0.9249,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999644449324526e-05,
+      "loss": 0.6987,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999640755390055e-05,
+      "loss": 0.4626,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996370423670097e-05,
+      "loss": 1.3488,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999633310255415e-05,
+      "loss": 1.1553,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999629559055297e-05,
+      "loss": 0.4537,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999625788766676e-05,
+      "loss": 0.7393,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999621999389578e-05,
+      "loss": 0.9267,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996181909240275e-05,
+      "loss": 0.7131,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996143633700467e-05,
+      "loss": 0.2273,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999610516727662e-05,
+      "loss": 1.0285,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996066509968966e-05,
+      "loss": 0.7143,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996027661777755e-05,
+      "loss": 0.7384,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9995988622703244e-05,
+      "loss": 0.6515,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999594939274567e-05,
+      "loss": 0.4473,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999590997190528e-05,
+      "loss": 0.2267,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9995870360182327e-05,
+      "loss": 1.1585,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999583055757707e-05,
+      "loss": 0.9453,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999579056408976e-05,
+      "loss": 0.9381,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999575037972065e-05,
+      "loss": 0.1904,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9995710004469996e-05,
+      "loss": 0.6304,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9995669438338053e-05,
+      "loss": 0.5582,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9995628681325075e-05,
+      "loss": 0.3723,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999558773343133e-05,
+      "loss": 0.8,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999554659465708e-05,
+      "loss": 1.1826,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999550526500258e-05,
+      "loss": 1.1407,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99954637444681e-05,
+      "loss": 0.3101,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9995422033053894e-05,
+      "loss": 0.3289,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9995380130760234e-05,
+      "loss": 0.5929,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999533803758739e-05,
+      "loss": 0.7245,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999529575353562e-05,
+      "loss": 0.8261,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99952532786052e-05,
+      "loss": 0.4927,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99952106127964e-05,
+      "loss": 0.3573,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9995167756109484e-05,
+      "loss": 0.5373,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999512470854474e-05,
+      "loss": 0.6602,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9995081470102424e-05,
+      "loss": 0.695,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999503804078283e-05,
+      "loss": 0.3144,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9994994420586215e-05,
+      "loss": 0.9949,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9994950609512875e-05,
+      "loss": 0.4096,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999490660756308e-05,
+      "loss": 1.0615,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9994862414737106e-05,
+      "loss": 0.6429,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999481803103523e-05,
+      "loss": 0.4857,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9994773456457764e-05,
+      "loss": 0.333,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999472869100495e-05,
+      "loss": 0.6271,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999468373467711e-05,
+      "loss": 1.3449,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99946385874745e-05,
+      "loss": 0.3742,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9994593249397424e-05,
+      "loss": 0.9754,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999454772044617e-05,
+      "loss": 0.6475,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999450200062102e-05,
+      "loss": 0.4562,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9994456089922274e-05,
+      "loss": 1.127,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999440998835022e-05,
+      "loss": 0.6281,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999436369590515e-05,
+      "loss": 0.3564,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999431721258736e-05,
+      "loss": 1.2867,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999427053839714e-05,
+      "loss": 0.8598,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99942236733348e-05,
+      "loss": 0.8063,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9994176617400625e-05,
+      "loss": 0.7529,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9994129370594934e-05,
+      "loss": 0.9318,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9994081932918e-05,
+      "loss": 0.3904,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999403430437014e-05,
+      "loss": 0.5806,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999398648495166e-05,
+      "loss": 0.9175,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9993938474662856e-05,
+      "loss": 0.9559,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9993890273504046e-05,
+      "loss": 0.3542,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999384188147553e-05,
+      "loss": 1.1604,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99937932985776e-05,
+      "loss": 1.2079,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9993744524810584e-05,
+      "loss": 1.6522,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99936955601748e-05,
+      "loss": 0.8846,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999364640467054e-05,
+      "loss": 0.7613,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9993597058298126e-05,
+      "loss": 0.4491,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999354752105786e-05,
+      "loss": 0.5709,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999349779295008e-05,
+      "loss": 1.2629,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999344787397509e-05,
+      "loss": 0.2976,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99933977641332e-05,
+      "loss": 1.5506,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9993347463424745e-05,
+      "loss": 1.0612,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999329697185003e-05,
+      "loss": 1.0229,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999324628940939e-05,
+      "loss": 0.898,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999319541610313e-05,
+      "loss": 0.9809,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99931443519316e-05,
+      "loss": 0.5301,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9993093096895095e-05,
+      "loss": 0.9113,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999304165099396e-05,
+      "loss": 1.0725,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999299001422852e-05,
+      "loss": 0.7417,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9992938186599105e-05,
+      "loss": 1.0214,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999288616810603e-05,
+      "loss": 0.5221,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9992833958749643e-05,
+      "loss": 0.8052,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999278155853027e-05,
+      "loss": 1.3484,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9992728967448244e-05,
+      "loss": 1.0481,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999267618550391e-05,
+      "loss": 0.7045,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9992623212697585e-05,
+      "loss": 0.6667,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9992570049029625e-05,
+      "loss": 0.1496,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999251669450035e-05,
+      "loss": 1.2328,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9992463149110105e-05,
+      "loss": 0.6702,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9992409412859244e-05,
+      "loss": 1.5594,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9992355485748095e-05,
+      "loss": 1.2045,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9992301367777004e-05,
+      "loss": 1.2895,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999224705894632e-05,
+      "loss": 0.4864,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999219255925638e-05,
+      "loss": 0.5734,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9992137868707544e-05,
+      "loss": 0.8902,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999208298730014e-05,
+      "loss": 0.6732,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999202791503454e-05,
+      "loss": 0.34,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999197265191109e-05,
+      "loss": 0.6457,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999191719793012e-05,
+      "loss": 0.3467,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9991861553092014e-05,
+      "loss": 0.5989,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999180571739709e-05,
+      "loss": 0.9266,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999174969084574e-05,
+      "loss": 0.8037,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9991693473438296e-05,
+      "loss": 0.6816,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999163706517513e-05,
+      "loss": 0.7889,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99915804660566e-05,
+      "loss": 0.8056,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9991523676083056e-05,
+      "loss": 1.0151,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999146669525486e-05,
+      "loss": 0.435,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9991409523572384e-05,
+      "loss": 0.3535,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9991352161035985e-05,
+      "loss": 1.1287,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999129460764603e-05,
+      "loss": 0.3947,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999123686340289e-05,
+      "loss": 0.4871,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999117892830692e-05,
+      "loss": 1.2137,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99911208023585e-05,
+      "loss": 0.2265,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9991062485557996e-05,
+      "loss": 0.2442,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999100397790578e-05,
+      "loss": 0.3961,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9990945279402225e-05,
+      "loss": 1.456,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9990886390047705e-05,
+      "loss": 0.8236,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999082730984259e-05,
+      "loss": 0.5154,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999076803878726e-05,
+      "loss": 1.3931,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999070857688209e-05,
+      "loss": 0.3635,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999064892412746e-05,
+      "loss": 1.1244,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999058908052375e-05,
+      "loss": 0.5527,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999052904607134e-05,
+      "loss": 1.3549,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9990468820770616e-05,
+      "loss": 0.559,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999040840462195e-05,
+      "loss": 0.3449,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999034779762574e-05,
+      "loss": 0.3137,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999028699978237e-05,
+      "loss": 0.8157,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999022601109221e-05,
+      "loss": 1.4109,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9990164831555665e-05,
+      "loss": 1.2802,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999010346117313e-05,
+      "loss": 0.3149,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9990041899944966e-05,
+      "loss": 1.3209,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9989980147871595e-05,
+      "loss": 0.8334,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99899182049534e-05,
+      "loss": 0.9313,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9989856071190774e-05,
+      "loss": 0.5031,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998979374658411e-05,
+      "loss": 0.7843,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9989731231133815e-05,
+      "loss": 0.4697,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998966852484027e-05,
+      "loss": 0.6147,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998960562770389e-05,
+      "loss": 0.3488,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9989542539725064e-05,
+      "loss": 1.2117,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99894792609042e-05,
+      "loss": 0.8119,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998941579124169e-05,
+      "loss": 0.0953,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.998935213073795e-05,
+      "loss": 1.0257,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998928827939339e-05,
+      "loss": 0.6113,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99892242372084e-05,
+      "loss": 0.7502,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99891600041834e-05,
+      "loss": 0.5541,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99890955803188e-05,
+      "loss": 0.4839,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998903096561499e-05,
+      "loss": 1.1221,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99889661600724e-05,
+      "loss": 0.7463,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998890116369145e-05,
+      "loss": 0.6099,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998883597647253e-05,
+      "loss": 0.8686,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998877059841607e-05,
+      "loss": 0.3367,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998870502952247e-05,
+      "loss": 0.3661,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9988639269792174e-05,
+      "loss": 0.4266,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998857331922558e-05,
+      "loss": 0.7553,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998850717782312e-05,
+      "loss": 1.442,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99884408455852e-05,
+      "loss": 1.8646,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998837432251225e-05,
+      "loss": 0.5022,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99883076086047e-05,
+      "loss": 1.053,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9988240703862965e-05,
+      "loss": 0.993,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998817360828748e-05,
+      "loss": 0.386,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998810632187866e-05,
+      "loss": 1.0764,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9988038844636935e-05,
+      "loss": 0.6977,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9987971176562754e-05,
+      "loss": 0.4798,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9987903317656515e-05,
+      "loss": 0.8181,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998783526791868e-05,
+      "loss": 0.5464,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998776702734966e-05,
+      "loss": 0.8875,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9987698595949914e-05,
+      "loss": 1.2418,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9987629973719844e-05,
+      "loss": 0.7488,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998756116065992e-05,
+      "loss": 0.8514,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998749215677055e-05,
+      "loss": 0.3395,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9987422962052196e-05,
+      "loss": 1.1217,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9987353576505294e-05,
+      "loss": 1.093,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998728400013028e-05,
+      "loss": 0.2827,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998721423292759e-05,
+      "loss": 1.1022,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998714427489768e-05,
+      "loss": 0.2735,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998707412604099e-05,
+      "loss": 0.5297,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998700378635797e-05,
+      "loss": 0.2697,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998693325584907e-05,
+      "loss": 0.4194,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9986862534514737e-05,
+      "loss": 0.374,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9986791622355407e-05,
+      "loss": 0.8277,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998672051937154e-05,
+      "loss": 1.4296,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998664922556361e-05,
+      "loss": 0.3553,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998657774093203e-05,
+      "loss": 0.2792,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9986506065477296e-05,
+      "loss": 0.7573,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998643419919984e-05,
+      "loss": 0.8796,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998636214210011e-05,
+      "loss": 0.995,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99862898941786e-05,
+      "loss": 0.3223,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998621745543573e-05,
+      "loss": 0.274,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998614482587199e-05,
+      "loss": 1.292,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9986072005487826e-05,
+      "loss": 1.0212,
+      "step": 392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9985998994283715e-05,
+      "loss": 1.0004,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99859257922601e-05,
+      "loss": 1.0147,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998585239941747e-05,
+      "loss": 0.4574,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998577881575628e-05,
+      "loss": 0.9685,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998570504127701e-05,
+      "loss": 0.79,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998563107598011e-05,
+      "loss": 0.951,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998555691986606e-05,
+      "loss": 0.7738,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998548257293534e-05,
+      "loss": 0.2533,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998540803518842e-05,
+      "loss": 0.5555,
+      "step": 401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998533330662576e-05,
+      "loss": 0.6275,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9985258387247846e-05,
+      "loss": 0.6821,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9985183277055156e-05,
+      "loss": 0.8291,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998510797604816e-05,
+      "loss": 0.7464,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998503248422736e-05,
+      "loss": 1.3062,
+      "step": 406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998495680159321e-05,
+      "loss": 0.8968,
+      "step": 407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9984880928146206e-05,
+      "loss": 0.513,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998480486388682e-05,
+      "loss": 1.023,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9984728608815554e-05,
+      "loss": 1.3292,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998465216293287e-05,
+      "loss": 0.8223,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998457552623927e-05,
+      "loss": 0.431,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9984498698735235e-05,
+      "loss": 0.7604,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998442168042127e-05,
+      "loss": 0.9758,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9984344471297834e-05,
+      "loss": 0.8111,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998426707136545e-05,
+      "loss": 1.031,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998418948062459e-05,
+      "loss": 0.5566,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998411169907576e-05,
+      "loss": 1.3899,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998403372671945e-05,
+      "loss": 0.3331,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998395556355614e-05,
+      "loss": 1.7484,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998387720958637e-05,
+      "loss": 0.7298,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998379866481059e-05,
+      "loss": 0.1869,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9983719929229334e-05,
+      "loss": 0.8926,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998364100284308e-05,
+      "loss": 0.6108,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998356188565235e-05,
+      "loss": 0.9012,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9983482577657635e-05,
+      "loss": 1.1116,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9983403078859446e-05,
+      "loss": 0.2982,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998332338925828e-05,
+      "loss": 0.4509,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998324350885466e-05,
+      "loss": 0.4404,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998316343764908e-05,
+      "loss": 1.7052,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998308317564205e-05,
+      "loss": 0.2172,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998300272283409e-05,
+      "loss": 0.4058,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99829220792257e-05,
+      "loss": 0.3381,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998284124481741e-05,
+      "loss": 0.9059,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998276021960971e-05,
+      "loss": 0.6692,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998267900360315e-05,
+      "loss": 0.5322,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998259759679821e-05,
+      "loss": 0.4405,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998251599919543e-05,
+      "loss": 0.475,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998243421079532e-05,
+      "loss": 0.8556,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998235223159841e-05,
+      "loss": 1.2924,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9982270061605216e-05,
+      "loss": 1.229,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9982187700816264e-05,
+      "loss": 0.2064,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998210514923207e-05,
+      "loss": 0.6372,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998202240685316e-05,
+      "loss": 0.5239,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998193947368008e-05,
+      "loss": 0.5744,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9981856349713336e-05,
+      "loss": 0.9508,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998177303495346e-05,
+      "loss": 0.6277,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9981689529400985e-05,
+      "loss": 0.5385,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998160583305645e-05,
+      "loss": 0.1886,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998152194592038e-05,
+      "loss": 0.8452,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998143786799331e-05,
+      "loss": 0.8216,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998135359927577e-05,
+      "loss": 0.5168,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.99812691397683e-05,
+      "loss": 0.9705,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998118448947144e-05,
+      "loss": 0.7036,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998109964838574e-05,
+      "loss": 0.8518,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9981014616511716e-05,
+      "loss": 1.0004,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9980929393849924e-05,
+      "loss": 1.5896,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.99808439804009e-05,
+      "loss": 1.1848,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998075837616519e-05,
+      "loss": 0.9662,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998067258114334e-05,
+      "loss": 0.3593,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998058659533589e-05,
+      "loss": 0.6831,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9980500418743395e-05,
+      "loss": 1.1698,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9980414051366406e-05,
+      "loss": 0.2089,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9980327493205465e-05,
+      "loss": 0.7581,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998024074426112e-05,
+      "loss": 1.469,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9980153804533927e-05,
+      "loss": 0.57,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998006667402445e-05,
+      "loss": 0.3228,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9979979352733224e-05,
+      "loss": 0.8751,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997989184066082e-05,
+      "loss": 0.6136,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9979804137807784e-05,
+      "loss": 0.7279,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9979716244174684e-05,
+      "loss": 1.1365,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997962815976207e-05,
+      "loss": 0.6784,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997953988457051e-05,
+      "loss": 1.0689,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997945141860056e-05,
+      "loss": 0.6053,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997936276185279e-05,
+      "loss": 0.5844,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9979273914327766e-05,
+      "loss": 0.3803,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9979184876026044e-05,
+      "loss": 0.4299,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997909564694818e-05,
+      "loss": 1.2806,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997900622709478e-05,
+      "loss": 0.9993,
+      "step": 479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997891661646637e-05,
+      "loss": 0.4537,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997882681506355e-05,
+      "loss": 0.5847,
+      "step": 481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997873682288688e-05,
+      "loss": 0.9249,
+      "step": 482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997864663993692e-05,
+      "loss": 0.4477,
+      "step": 483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997855626621427e-05,
+      "loss": 0.665,
+      "step": 484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997846570171949e-05,
+      "loss": 0.8012,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997837494645316e-05,
+      "loss": 0.5784,
+      "step": 486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997828400041586e-05,
+      "loss": 1.209,
+      "step": 487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9978192863608155e-05,
+      "loss": 1.1913,
+      "step": 488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9978101536030635e-05,
+      "loss": 0.418,
+      "step": 489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997801001768389e-05,
+      "loss": 0.4858,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997791830856848e-05,
+      "loss": 1.2531,
+      "step": 491
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997782640868501e-05,
+      "loss": 0.4787,
+      "step": 492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997773431803406e-05,
+      "loss": 0.5832,
+      "step": 493
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997764203661622e-05,
+      "loss": 0.2085,
+      "step": 494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997754956443205e-05,
+      "loss": 0.4369,
+      "step": 495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9977456901482176e-05,
+      "loss": 0.415,
+      "step": 496
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997736404776716e-05,
+      "loss": 1.1403,
+      "step": 497
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9977271003287606e-05,
+      "loss": 0.4629,
+      "step": 498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9977177768044105e-05,
+      "loss": 0.519,
+      "step": 499
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9977084342037245e-05,
+      "loss": 0.3958,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997699072526763e-05,
+      "loss": 0.9738,
+      "step": 501
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9976896917735845e-05,
+      "loss": 1.2351,
+      "step": 502
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997680291944249e-05,
+      "loss": 0.3609,
+      "step": 503
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997670873038817e-05,
+      "loss": 0.5062,
+      "step": 504
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9976614350573475e-05,
+      "loss": 0.199,
+      "step": 505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997651977999901e-05,
+      "loss": 0.471,
+      "step": 506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9976425018665386e-05,
+      "loss": 1.3523,
+      "step": 507
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9976330066573184e-05,
+      "loss": 1.383,
+      "step": 508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997623492372303e-05,
+      "loss": 0.6397,
+      "step": 509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997613959011552e-05,
+      "loss": 0.9004,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9976044065751254e-05,
+      "loss": 0.6136,
+      "step": 511
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997594835063085e-05,
+      "loss": 0.198,
+      "step": 512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9975852444754915e-05,
+      "loss": 0.4103,
+      "step": 513
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9975756348124066e-05,
+      "loss": 0.4988,
+      "step": 514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9975660060738896e-05,
+      "loss": 1.1275,
+      "step": 515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9975563582600036e-05,
+      "loss": 0.8931,
+      "step": 516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997546691370809e-05,
+      "loss": 0.8093,
+      "step": 517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9975370054063666e-05,
+      "loss": 1.2331,
+      "step": 518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.99752730036674e-05,
+      "loss": 0.8776,
+      "step": 519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997517576251991e-05,
+      "loss": 0.5471,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997507833062179e-05,
+      "loss": 1.1267,
+      "step": 521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9974980707973684e-05,
+      "loss": 0.9827,
+      "step": 522
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997488289457619e-05,
+      "loss": 0.4691,
+      "step": 523
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997478489042996e-05,
+      "loss": 0.6836,
+      "step": 524
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997468669553559e-05,
+      "loss": 0.476,
+      "step": 525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9974588309893726e-05,
+      "loss": 0.459,
+      "step": 526
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997448973350498e-05,
+      "loss": 0.4088,
+      "step": 527
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997439096636998e-05,
+      "loss": 1.0805,
+      "step": 528
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997429200848937e-05,
+      "loss": 0.9229,
+      "step": 529
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997419285986376e-05,
+      "loss": 0.4924,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9974093520493784e-05,
+      "loss": 0.8146,
+      "step": 531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9973993990380087e-05,
+      "loss": 1.4679,
+      "step": 532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.99738942695233e-05,
+      "loss": 0.5055,
+      "step": 533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9973794357924047e-05,
+      "loss": 0.4325,
+      "step": 534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9973694255582964e-05,
+      "loss": 0.6653,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.99735939625007e-05,
+      "loss": 0.2578,
+      "step": 536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9973493478677894e-05,
+      "loss": 1.2039,
+      "step": 537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9973392804115166e-05,
+      "loss": 0.9276,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9973291938813174e-05,
+      "loss": 0.9127,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997319088277254e-05,
+      "loss": 0.8766,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997308963599394e-05,
+      "loss": 1.2223,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9972988198478e-05,
+      "loss": 0.3587,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9972886570225356e-05,
+      "loss": 1.2694,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9972784751236664e-05,
+      "loss": 2.0048,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9972682741512585e-05,
+      "loss": 1.2022,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997258054105374e-05,
+      "loss": 0.2993,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9972478149860805e-05,
+      "loss": 0.3411,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997237556793441e-05,
+      "loss": 1.1513,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997227279527523e-05,
+      "loss": 0.6251,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997216983188391e-05,
+      "loss": 1.1043,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997206667776109e-05,
+      "loss": 1.415,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997196333290744e-05,
+      "loss": 0.3739,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997185979732363e-05,
+      "loss": 1.4098,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.99717560710103e-05,
+      "loss": 1.183,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997165215396812e-05,
+      "loss": 0.5319,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9971548046197744e-05,
+      "loss": 1.1395,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997144374769984e-05,
+      "loss": 0.1463,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997133925847506e-05,
+      "loss": 1.1007,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9971234578524085e-05,
+      "loss": 0.9355,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9971129707847584e-05,
+      "loss": 0.7259,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.99710246464462e-05,
+      "loss": 0.4962,
+      "step": 561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997091939432063e-05,
+      "loss": 1.1245,
+      "step": 562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997081395147152e-05,
+      "loss": 1.3628,
+      "step": 563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997070831789955e-05,
+      "loss": 1.073,
+      "step": 564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.99706024936054e-05,
+      "loss": 1.2823,
+      "step": 565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9970496478589736e-05,
+      "loss": 0.7573,
+      "step": 566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9970390272853236e-05,
+      "loss": 0.4683,
+      "step": 567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997028387639657e-05,
+      "loss": 0.3766,
+      "step": 568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9970177289220424e-05,
+      "loss": 0.5968,
+      "step": 569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997007051132546e-05,
+      "loss": 0.9696,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996996354271238e-05,
+      "loss": 1.3313,
+      "step": 571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9969856383381846e-05,
+      "loss": 0.9367,
+      "step": 572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996974903333456e-05,
+      "loss": 0.7735,
+      "step": 573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996964149257118e-05,
+      "loss": 1.2414,
+      "step": 574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.99695337610924e-05,
+      "loss": 0.8832,
+      "step": 575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9969425838898914e-05,
+      "loss": 0.7956,
+      "step": 576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.99693177259914e-05,
+      "loss": 0.4441,
+      "step": 577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996920942237055e-05,
+      "loss": 0.3953,
+      "step": 578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996910092803706e-05,
+      "loss": 1.4373,
+      "step": 579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.99689922429916e-05,
+      "loss": 0.7617,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996888336723488e-05,
+      "loss": 0.3258,
+      "step": 581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9968774300767586e-05,
+      "loss": 0.9018,
+      "step": 582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9968665043590414e-05,
+      "loss": 0.7324,
+      "step": 583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996855559570406e-05,
+      "loss": 0.5455,
+      "step": 584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996844595710922e-05,
+      "loss": 1.1078,
+      "step": 585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996833612780658e-05,
+      "loss": 0.4002,
+      "step": 586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9968226107796866e-05,
+      "loss": 0.9084,
+      "step": 587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996811589708075e-05,
+      "loss": 1.6678,
+      "step": 588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996800549565895e-05,
+      "loss": 0.7559,
+      "step": 589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9967894903532155e-05,
+      "loss": 0.2455,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9967784120701085e-05,
+      "loss": 0.2278,
+      "step": 591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9967673147166436e-05,
+      "loss": 1.1137,
+      "step": 592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996756198292891e-05,
+      "loss": 1.0953,
+      "step": 593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9967450627989226e-05,
+      "loss": 0.6262,
+      "step": 594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996733908234809e-05,
+      "loss": 1.0849,
+      "step": 595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9967227346006194e-05,
+      "loss": 0.5934,
+      "step": 596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996711541896428e-05,
+      "loss": 1.4646,
+      "step": 597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996700330122303e-05,
+      "loss": 0.6864,
+      "step": 598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9966890992783176e-05,
+      "loss": 0.7407,
+      "step": 599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996677849364543e-05,
+      "loss": 0.4342,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.99666658038105e-05,
+      "loss": 0.9907,
+      "step": 601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996655292327912e-05,
+      "loss": 0.5706,
+      "step": 602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9966439852051986e-05,
+      "loss": 0.8936,
+      "step": 603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996632659012983e-05,
+      "loss": 0.5444,
+      "step": 604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996621313751338e-05,
+      "loss": 0.6518,
+      "step": 605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9966099494203344e-05,
+      "loss": 0.5314,
+      "step": 606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9965985660200444e-05,
+      "loss": 0.7321,
+      "step": 607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996587163550542e-05,
+      "loss": 0.4493,
+      "step": 608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996575742011898e-05,
+      "loss": 0.1487,
+      "step": 609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9965643014041866e-05,
+      "loss": 0.8076,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.99655284172748e-05,
+      "loss": 0.9151,
+      "step": 611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9965413629818506e-05,
+      "loss": 0.5829,
+      "step": 612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9965298651673723e-05,
+      "loss": 1.1176,
+      "step": 613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996518348284117e-05,
+      "loss": 0.5693,
+      "step": 614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9965068123321606e-05,
+      "loss": 0.9186,
+      "step": 615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996495257311574e-05,
+      "loss": 0.5334,
+      "step": 616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996483683222431e-05,
+      "loss": 0.8471,
+      "step": 617
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996472090064806e-05,
+      "loss": 1.1946,
+      "step": 618
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996460477838772e-05,
+      "loss": 0.5363,
+      "step": 619
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996448846544405e-05,
+      "loss": 1.471,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996437196181776e-05,
+      "loss": 0.5724,
+      "step": 621
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996425526750961e-05,
+      "loss": 0.6465,
+      "step": 622
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9964138382520345e-05,
+      "loss": 0.7636,
+      "step": 623
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996402130685069e-05,
+      "loss": 0.6175,
+      "step": 624
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9963904040501403e-05,
+      "loss": 0.0177,
+      "step": 625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996378658347324e-05,
+      "loss": 1.4002,
+      "step": 626
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996366893576693e-05,
+      "loss": 1.0862,
+      "step": 627
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996355109738323e-05,
+      "loss": 0.2438,
+      "step": 628
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.99634330683229e-05,
+      "loss": 0.8852,
+      "step": 629
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996331484858667e-05,
+      "loss": 0.606,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9963196438175305e-05,
+      "loss": 1.5637,
+      "step": 631
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996307783708956e-05,
+      "loss": 0.868,
+      "step": 632
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9962959045330184e-05,
+      "loss": 0.755,
+      "step": 633
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996284006289793e-05,
+      "loss": 1.3792,
+      "step": 634
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9962720889793565e-05,
+      "loss": 1.6055,
+      "step": 635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9962601526017845e-05,
+      "loss": 0.815,
+      "step": 636
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9962481971571526e-05,
+      "loss": 0.9183,
+      "step": 637
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9962362226455366e-05,
+      "loss": 0.5483,
+      "step": 638
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996224229067013e-05,
+      "loss": 0.6956,
+      "step": 639
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996212216421659e-05,
+      "loss": 0.2355,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9962001847095494e-05,
+      "loss": 1.359,
+      "step": 641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996188133930763e-05,
+      "loss": 1.1783,
+      "step": 642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996176064085373e-05,
+      "loss": 0.3781,
+      "step": 643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99616397517346e-05,
+      "loss": 1.2409,
+      "step": 644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996151867195099e-05,
+      "loss": 0.4251,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996139740150367e-05,
+      "loss": 1.4537,
+      "step": 646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9961275940393416e-05,
+      "loss": 1.319,
+      "step": 647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9961154288621e-05,
+      "loss": 1.0258,
+      "step": 648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9961032446187193e-05,
+      "loss": 0.4076,
+      "step": 649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996091041309278e-05,
+      "loss": 0.9052,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9960788189338526e-05,
+      "loss": 0.4149,
+      "step": 651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996066577492521e-05,
+      "loss": 0.7334,
+      "step": 652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996054316985362e-05,
+      "loss": 0.4792,
+      "step": 653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996042037412453e-05,
+      "loss": 0.3469,
+      "step": 654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9960297387738725e-05,
+      "loss": 0.9926,
+      "step": 655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9960174210696984e-05,
+      "loss": 0.8215,
+      "step": 656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9960050843000086e-05,
+      "loss": 0.9677,
+      "step": 657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9959927284648826e-05,
+      "loss": 1.1923,
+      "step": 658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995980353564399e-05,
+      "loss": 0.09,
+      "step": 659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995967959598635e-05,
+      "loss": 0.6642,
+      "step": 660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995955546567672e-05,
+      "loss": 0.4872,
+      "step": 661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9959431144715876e-05,
+      "loss": 1.0875,
+      "step": 662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99593066331046e-05,
+      "loss": 0.7099,
+      "step": 663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9959181930843694e-05,
+      "loss": 1.2392,
+      "step": 664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995905703793395e-05,
+      "loss": 1.2947,
+      "step": 665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995893195437617e-05,
+      "loss": 1.2215,
+      "step": 666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9958806680171145e-05,
+      "loss": 1.0522,
+      "step": 667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9958681215319665e-05,
+      "loss": 0.6942,
+      "step": 668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9958555559822536e-05,
+      "loss": 0.6779,
+      "step": 669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9958429713680564e-05,
+      "loss": 0.6918,
+      "step": 670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9958303676894535e-05,
+      "loss": 0.5521,
+      "step": 671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995817744946526e-05,
+      "loss": 0.448,
+      "step": 672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9958051031393545e-05,
+      "loss": 0.3696,
+      "step": 673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995792442268018e-05,
+      "loss": 0.3717,
+      "step": 674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9957797623325985e-05,
+      "loss": 0.4186,
+      "step": 675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995767063333176e-05,
+      "loss": 1.0991,
+      "step": 676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9957543452698325e-05,
+      "loss": 1.1699,
+      "step": 677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995741608142647e-05,
+      "loss": 0.9235,
+      "step": 678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995728851951702e-05,
+      "loss": 0.5803,
+      "step": 679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9957160766970776e-05,
+      "loss": 0.1788,
+      "step": 680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9957032823788564e-05,
+      "loss": 0.4068,
+      "step": 681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995690468997119e-05,
+      "loss": 1.1909,
+      "step": 682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995677636551947e-05,
+      "loss": 0.9389,
+      "step": 683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995664785043423e-05,
+      "loss": 0.9354,
+      "step": 684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995651914471626e-05,
+      "loss": 0.6715,
+      "step": 685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995639024836642e-05,
+      "loss": 0.3613,
+      "step": 686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995626116138549e-05,
+      "loss": 0.3665,
+      "step": 687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995613188377432e-05,
+      "loss": 1.2575,
+      "step": 688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9956002415533715e-05,
+      "loss": 0.3688,
+      "step": 689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995587275666451e-05,
+      "loss": 0.5994,
+      "step": 690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995574290716753e-05,
+      "loss": 1.268,
+      "step": 691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99556128670436e-05,
+      "loss": 0.2947,
+      "step": 692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995548263629354e-05,
+      "loss": 0.2899,
+      "step": 693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9955352214918185e-05,
+      "loss": 0.9178,
+      "step": 694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995522160291836e-05,
+      "loss": 0.6208,
+      "step": 695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9955090800294907e-05,
+      "loss": 0.6876,
+      "step": 696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995495980704865e-05,
+      "loss": 0.8101,
+      "step": 697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995482862318042e-05,
+      "loss": 0.9553,
+      "step": 698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9954697248691064e-05,
+      "loss": 0.2213,
+      "step": 699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99545656835814e-05,
+      "loss": 0.9857,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9954433927852284e-05,
+      "loss": 0.9924,
+      "step": 701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995430198150454e-05,
+      "loss": 0.4246,
+      "step": 702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995416984453902e-05,
+      "loss": 0.8086,
+      "step": 703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995403751695656e-05,
+      "loss": 0.737,
+      "step": 704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995390499875799e-05,
+      "loss": 0.2854,
+      "step": 705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995377228994417e-05,
+      "loss": 0.4219,
+      "step": 706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9953639390515936e-05,
+      "loss": 0.159,
+      "step": 707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9953506300474134e-05,
+      "loss": 0.5283,
+      "step": 708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995337301981962e-05,
+      "loss": 0.9114,
+      "step": 709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995323954855323e-05,
+      "loss": 0.4824,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995310588667581e-05,
+      "loss": 0.5147,
+      "step": 711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995297203418823e-05,
+      "loss": 0.5746,
+      "step": 712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995283799109133e-05,
+      "loss": 0.9441,
+      "step": 713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995270375738596e-05,
+      "loss": 0.4538,
+      "step": 714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995256933307298e-05,
+      "loss": 0.4202,
+      "step": 715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9952434718153244e-05,
+      "loss": 0.4775,
+      "step": 716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995229991262761e-05,
+      "loss": 0.9116,
+      "step": 717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995216491649693e-05,
+      "loss": 0.6272,
+      "step": 718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9952029729762066e-05,
+      "loss": 0.9807,
+      "step": 719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995189435242388e-05,
+      "loss": 0.6383,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995175878448323e-05,
+      "loss": 0.8616,
+      "step": 721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995162302594099e-05,
+      "loss": 1.1022,
+      "step": 722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9951487076798e-05,
+      "loss": 0.4044,
+      "step": 723
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9951350937055155e-05,
+      "loss": 0.5404,
+      "step": 724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995121460671331e-05,
+      "loss": 1.1615,
+      "step": 725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995107808577332e-05,
+      "loss": 0.6072,
+      "step": 726
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9950941374236064e-05,
+      "loss": 0.9487,
+      "step": 727
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9950804472102405e-05,
+      "loss": 0.6474,
+      "step": 728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995066737937323e-05,
+      "loss": 1.1497,
+      "step": 729
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99505300960494e-05,
+      "loss": 0.7856,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995039262213179e-05,
+      "loss": 1.1286,
+      "step": 731
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995025495762126e-05,
+      "loss": 1.1544,
+      "step": 732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995011710251872e-05,
+      "loss": 0.1973,
+      "step": 733
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9949979056825024e-05,
+      "loss": 1.2409,
+      "step": 734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.994984082054105e-05,
+      "loss": 0.2227,
+      "step": 735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.994970239366769e-05,
+      "loss": 0.8987,
+      "step": 736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.994956377620581e-05,
+      "loss": 0.5902,
+      "step": 737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9949424968156305e-05,
+      "loss": 1.4954,
+      "step": 738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994928596952005e-05,
+      "loss": 0.3285,
+      "step": 739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994914678029793e-05,
+      "loss": 0.9254,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994900740049084e-05,
+      "loss": 0.6972,
+      "step": 741
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9948867830099655e-05,
+      "loss": 0.6307,
+      "step": 742
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9948728069125275e-05,
+      "loss": 0.7056,
+      "step": 743
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994858811756857e-05,
+      "loss": 0.4869,
+      "step": 744
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994844797543045e-05,
+      "loss": 0.1569,
+      "step": 745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9948307642711805e-05,
+      "loss": 0.9653,
+      "step": 746
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994816711941351e-05,
+      "loss": 0.5782,
+      "step": 747
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994802640553649e-05,
+      "loss": 0.4263,
+      "step": 748
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994788550108161e-05,
+      "loss": 1.4481,
+      "step": 749
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994774440604979e-05,
+      "loss": 1.0927,
+      "step": 750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994760312044191e-05,
+      "loss": 0.5295,
+      "step": 751
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994746164425887e-05,
+      "loss": 1.6251,
+      "step": 752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994731997750159e-05,
+      "loss": 0.8541,
+      "step": 753
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9947178120170943e-05,
+      "loss": 1.2485,
+      "step": 754
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994703607226787e-05,
+      "loss": 0.1555,
+      "step": 755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994689383379323e-05,
+      "loss": 1.304,
+      "step": 756
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9946751404747964e-05,
+      "loss": 0.6735,
+      "step": 757
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994660878513295e-05,
+      "loss": 1.1749,
+      "step": 758
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994646597494912e-05,
+      "loss": 0.8336,
+      "step": 759
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9946322974197366e-05,
+      "loss": 1.3057,
+      "step": 760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994617978287862e-05,
+      "loss": 1.2463,
+      "step": 761
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994603640099376e-05,
+      "loss": 0.8107,
+      "step": 762
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9945892828543725e-05,
+      "loss": 1.2675,
+      "step": 763
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994574906552942e-05,
+      "loss": 1.3613,
+      "step": 764
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994560511195176e-05,
+      "loss": 1.1386,
+      "step": 765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9945460967811654e-05,
+      "loss": 0.9878,
+      "step": 766
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994531663311004e-05,
+      "loss": 0.839,
+      "step": 767
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994517210784781e-05,
+      "loss": 0.9265,
+      "step": 768
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.99450273920259e-05,
+      "loss": 0.8044,
+      "step": 769
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9944882485645236e-05,
+      "loss": 0.2424,
+      "step": 770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9944737388706716e-05,
+      "loss": 0.4956,
+      "step": 771
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9944592101211296e-05,
+      "loss": 0.6453,
+      "step": 772
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994444662315987e-05,
+      "loss": 1.1616,
+      "step": 773
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994430095455338e-05,
+      "loss": 0.9172,
+      "step": 774
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994415509539275e-05,
+      "loss": 1.275,
+      "step": 775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9944009045678907e-05,
+      "loss": 0.9774,
+      "step": 776
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994386280541279e-05,
+      "loss": 0.8119,
+      "step": 777
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994371637459531e-05,
+      "loss": 0.78,
+      "step": 778
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9943569753227425e-05,
+      "loss": 0.4432,
+      "step": 779
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9943422941310045e-05,
+      "loss": 0.9889,
+      "step": 780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9943275938844106e-05,
+      "loss": 0.8009,
+      "step": 781
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994312874583056e-05,
+      "loss": 0.6015,
+      "step": 782
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994298136227032e-05,
+      "loss": 1.2429,
+      "step": 783
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9942833788164356e-05,
+      "loss": 0.9151,
+      "step": 784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994268602351358e-05,
+      "loss": 0.3077,
+      "step": 785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9942538068318935e-05,
+      "loss": 0.8845,
+      "step": 786
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9942389922581375e-05,
+      "loss": 0.2527,
+      "step": 787
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9942241586301836e-05,
+      "loss": 0.6106,
+      "step": 788
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994209305948126e-05,
+      "loss": 0.8529,
+      "step": 789
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9941944342120595e-05,
+      "loss": 0.6385,
+      "step": 790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9941795434220784e-05,
+      "loss": 0.8685,
+      "step": 791
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994164633578278e-05,
+      "loss": 0.4938,
+      "step": 792
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9941497046807536e-05,
+      "loss": 1.0358,
+      "step": 793
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994134756729598e-05,
+      "loss": 1.0513,
+      "step": 794
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.99411978972491e-05,
+      "loss": 0.5238,
+      "step": 795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9941048036667804e-05,
+      "loss": 0.8088,
+      "step": 796
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994089798555308e-05,
+      "loss": 0.7233,
+      "step": 797
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994074774390587e-05,
+      "loss": 1.0202,
+      "step": 798
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994059731172713e-05,
+      "loss": 1.108,
+      "step": 799
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9940446689017825e-05,
+      "loss": 0.6646,
+      "step": 800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.99402958757789e-05,
+      "loss": 0.6832,
+      "step": 801
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9940144872011325e-05,
+      "loss": 1.2159,
+      "step": 802
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9939993677716055e-05,
+      "loss": 0.8498,
+      "step": 803
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993984229289405e-05,
+      "loss": 0.4032,
+      "step": 804
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993969071754629e-05,
+      "loss": 0.7459,
+      "step": 805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993953895167372e-05,
+      "loss": 0.2889,
+      "step": 806
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993938699527732e-05,
+      "loss": 1.0238,
+      "step": 807
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993923484835804e-05,
+      "loss": 0.9424,
+      "step": 808
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993908251091687e-05,
+      "loss": 0.7954,
+      "step": 809
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993892998295477e-05,
+      "loss": 1.1163,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.99387772644727e-05,
+      "loss": 1.0896,
+      "step": 811
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993862435547164e-05,
+      "loss": 0.7397,
+      "step": 812
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993847125595258e-05,
+      "loss": 0.4138,
+      "step": 813
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9938317965916456e-05,
+      "loss": 0.3417,
+      "step": 814
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9938164485364275e-05,
+      "loss": 0.2841,
+      "step": 815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9938010814297016e-05,
+      "loss": 0.4741,
+      "step": 816
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993785695271563e-05,
+      "loss": 0.2632,
+      "step": 817
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993770290062111e-05,
+      "loss": 0.5257,
+      "step": 818
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9937548658014446e-05,
+      "loss": 1.1948,
+      "step": 819
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993739422489661e-05,
+      "loss": 0.8593,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993723960126858e-05,
+      "loss": 0.9148,
+      "step": 821
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993708478713135e-05,
+      "loss": 0.5111,
+      "step": 822
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9936929782485904e-05,
+      "loss": 1.4134,
+      "step": 823
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9936774587333216e-05,
+      "loss": 0.7602,
+      "step": 824
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993661920167429e-05,
+      "loss": 1.3578,
+      "step": 825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993646362551011e-05,
+      "loss": 1.2169,
+      "step": 826
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993630785884167e-05,
+      "loss": 0.2906,
+      "step": 827
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993615190166994e-05,
+      "loss": 0.6531,
+      "step": 828
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9935995753995936e-05,
+      "loss": 1.164,
+      "step": 829
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993583941582064e-05,
+      "loss": 0.4303,
+      "step": 830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9935682887145046e-05,
+      "loss": 0.8911,
+      "step": 831
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993552616797017e-05,
+      "loss": 0.4636,
+      "step": 832
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993536925829698e-05,
+      "loss": 0.3854,
+      "step": 833
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993521215812649e-05,
+      "loss": 0.3658,
+      "step": 834
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.99350548674597e-05,
+      "loss": 0.5457,
+      "step": 835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993489738629761e-05,
+      "loss": 0.5916,
+      "step": 836
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993473971464122e-05,
+      "loss": 0.5056,
+      "step": 837
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993458185249153e-05,
+      "loss": 1.1208,
+      "step": 838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993442379984955e-05,
+      "loss": 1.0001,
+      "step": 839
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9934265556716285e-05,
+      "loss": 0.7269,
+      "step": 840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993410712309274e-05,
+      "loss": 1.1036,
+      "step": 841
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993394849897994e-05,
+      "loss": 0.8246,
+      "step": 842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993378968437886e-05,
+      "loss": 0.2627,
+      "step": 843
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9933630679290534e-05,
+      "loss": 0.6711,
+      "step": 844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993347148371597e-05,
+      "loss": 1.3401,
+      "step": 845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993331209765619e-05,
+      "loss": 0.4429,
+      "step": 846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993315252111219e-05,
+      "loss": 0.3356,
+      "step": 847
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9932992754084995e-05,
+      "loss": 0.8205,
+      "step": 848
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9932832796575626e-05,
+      "loss": 1.0614,
+      "step": 849
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993267264858509e-05,
+      "loss": 1.1596,
+      "step": 850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993251231011441e-05,
+      "loss": 0.3754,
+      "step": 851
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993235178116462e-05,
+      "loss": 1.0934,
+      "step": 852
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993219106173672e-05,
+      "loss": 0.5334,
+      "step": 853
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993203015183175e-05,
+      "loss": 0.6592,
+      "step": 854
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993186905145072e-05,
+      "loss": 1.2703,
+      "step": 855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9931707760594665e-05,
+      "loss": 0.6539,
+      "step": 856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9931546279264614e-05,
+      "loss": 0.7534,
+      "step": 857
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993138460746158e-05,
+      "loss": 0.4736,
+      "step": 858
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993122274518661e-05,
+      "loss": 0.5078,
+      "step": 859
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993106069244072e-05,
+      "loss": 0.5637,
+      "step": 860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993089844922494e-05,
+      "loss": 1.3652,
+      "step": 861
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9930736015540314e-05,
+      "loss": 0.5501,
+      "step": 862
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9930573391387875e-05,
+      "loss": 0.9075,
+      "step": 863
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9930410576768646e-05,
+      "loss": 0.9839,
+      "step": 864
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993024757168367e-05,
+      "loss": 0.376,
+      "step": 865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9930084376133986e-05,
+      "loss": 0.8899,
+      "step": 866
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9929920990120636e-05,
+      "loss": 1.2575,
+      "step": 867
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992975741364465e-05,
+      "loss": 0.3999,
+      "step": 868
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992959364670708e-05,
+      "loss": 1.0932,
+      "step": 869
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992942968930895e-05,
+      "loss": 1.2193,
+      "step": 870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9929265541451324e-05,
+      "loss": 0.4893,
+      "step": 871
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9929101203135236e-05,
+      "loss": 1.2102,
+      "step": 872
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992893667436174e-05,
+      "loss": 1.0117,
+      "step": 873
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992877195513187e-05,
+      "loss": 1.0674,
+      "step": 874
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992860704544668e-05,
+      "loss": 0.7328,
+      "step": 875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992844194530721e-05,
+      "loss": 0.4814,
+      "step": 876
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992827665471453e-05,
+      "loss": 1.2574,
+      "step": 877
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9928111173669696e-05,
+      "loss": 0.3544,
+      "step": 878
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992794550217373e-05,
+      "loss": 0.8544,
+      "step": 879
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9927779640227714e-05,
+      "loss": 0.4075,
+      "step": 880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992761358783269e-05,
+      "loss": 0.7004,
+      "step": 881
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9927447344989716e-05,
+      "loss": 1.0467,
+      "step": 882
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992728091169985e-05,
+      "loss": 0.2034,
+      "step": 883
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992711428796416e-05,
+      "loss": 0.5745,
+      "step": 884
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992694747378369e-05,
+      "loss": 0.2963,
+      "step": 885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9926780469159524e-05,
+      "loss": 1.6303,
+      "step": 886
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9926613274092705e-05,
+      "loss": 0.9505,
+      "step": 887
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9926445888584303e-05,
+      "loss": 1.1239,
+      "step": 888
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992627831263539e-05,
+      "loss": 0.6669,
+      "step": 889
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992611054624702e-05,
+      "loss": 0.934,
+      "step": 890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9925942589420267e-05,
+      "loss": 0.6273,
+      "step": 891
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9925774442156196e-05,
+      "loss": 1.3053,
+      "step": 892
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992560610445589e-05,
+      "loss": 1.1022,
+      "step": 893
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9925437576320406e-05,
+      "loss": 0.1624,
+      "step": 894
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992526885775083e-05,
+      "loss": 1.0971,
+      "step": 895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992509994874821e-05,
+      "loss": 0.4757,
+      "step": 896
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992493084931365e-05,
+      "loss": 0.4453,
+      "step": 897
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9924761559448205e-05,
+      "loss": 1.22,
+      "step": 898
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992459207915297e-05,
+      "loss": 1.5914,
+      "step": 899
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992442240842902e-05,
+      "loss": 0.6316,
+      "step": 900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992425254727742e-05,
+      "loss": 1.5475,
+      "step": 901
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992408249569926e-05,
+      "loss": 0.4901,
+      "step": 902
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992391225369563e-05,
+      "loss": 0.3896,
+      "step": 903
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99237418212676e-05,
+      "loss": 1.3218,
+      "step": 904
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9923571198416266e-05,
+      "loss": 1.0489,
+      "step": 905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99234003851427e-05,
+      "loss": 0.6329,
+      "step": 906
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9923229381448e-05,
+      "loss": 0.5065,
+      "step": 907
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992305818733326e-05,
+      "loss": 0.0689,
+      "step": 908
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992288680279955e-05,
+      "loss": 0.5334,
+      "step": 909
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992271522784798e-05,
+      "loss": 1.6699,
+      "step": 910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992254346247963e-05,
+      "loss": 0.7227,
+      "step": 911
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99223715066956e-05,
+      "loss": 0.5982,
+      "step": 912
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9922199360496976e-05,
+      "loss": 0.6213,
+      "step": 913
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992202702388486e-05,
+      "loss": 0.6863,
+      "step": 914
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992185449686034e-05,
+      "loss": 0.3016,
+      "step": 915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992168177942453e-05,
+      "loss": 0.4865,
+      "step": 916
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992150887157852e-05,
+      "loss": 0.7168,
+      "step": 917
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99213357733234e-05,
+      "loss": 1.0368,
+      "step": 918
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99211624846603e-05,
+      "loss": 0.933,
+      "step": 919
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992098900559029e-05,
+      "loss": 0.4797,
+      "step": 920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992081533611449e-05,
+      "loss": 1.3121,
+      "step": 921
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992064147623401e-05,
+      "loss": 0.9277,
+      "step": 922
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9920467425949945e-05,
+      "loss": 1.0576,
+      "step": 923
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992029318526341e-05,
+      "loss": 1.3692,
+      "step": 924
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99201187541755e-05,
+      "loss": 0.3829,
+      "step": 925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9919944132687354e-05,
+      "loss": 0.3077,
+      "step": 926
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.991976932080006e-05,
+      "loss": 0.8876,
+      "step": 927
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.991959431851473e-05,
+      "loss": 0.4811,
+      "step": 928
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.991941912583249e-05,
+      "loss": 0.337,
+      "step": 929
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.991924374275445e-05,
+      "loss": 1.2518,
+      "step": 930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9919068169281716e-05,
+      "loss": 0.619,
+      "step": 931
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9918892405415425e-05,
+      "loss": 0.4147,
+      "step": 932
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.991871645115668e-05,
+      "loss": 0.6999,
+      "step": 933
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99185403065066e-05,
+      "loss": 0.8617,
+      "step": 934
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.991836397146632e-05,
+      "loss": 2.0689,
+      "step": 935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991818744603694e-05,
+      "loss": 1.1669,
+      "step": 936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991801073021961e-05,
+      "loss": 0.3879,
+      "step": 937
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991783382401544e-05,
+      "loss": 0.4889,
+      "step": 938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991765672742555e-05,
+      "loss": 0.0959,
+      "step": 939
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9917479440451074e-05,
+      "loss": 1.3007,
+      "step": 940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991730196309315e-05,
+      "loss": 0.9245,
+      "step": 941
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991712429535288e-05,
+      "loss": 0.5462,
+      "step": 942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991694643723142e-05,
+      "loss": 0.5576,
+      "step": 943
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.99167683887299e-05,
+      "loss": 0.7575,
+      "step": 944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991659014984945e-05,
+      "loss": 0.7763,
+      "step": 945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9916411720591194e-05,
+      "loss": 0.2509,
+      "step": 946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991623310095627e-05,
+      "loss": 0.7057,
+      "step": 947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9916054290945824e-05,
+      "loss": 0.3836,
+      "step": 948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991587529056099e-05,
+      "loss": 0.635,
+      "step": 949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9915696099802906e-05,
+      "loss": 0.4033,
+      "step": 950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991551671867271e-05,
+      "loss": 0.2431,
+      "step": 951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991533714717155e-05,
+      "loss": 1.0649,
+      "step": 952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9915157385300564e-05,
+      "loss": 1.3552,
+      "step": 953
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991497743306089e-05,
+      "loss": 0.6536,
+      "step": 954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9914797290453686e-05,
+      "loss": 0.5932,
+      "step": 955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9914616957480095e-05,
+      "loss": 0.4215,
+      "step": 956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991443643414126e-05,
+      "loss": 1.0358,
+      "step": 957
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991425572043833e-05,
+      "loss": 1.0602,
+      "step": 958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9914074816372454e-05,
+      "loss": 0.8767,
+      "step": 959
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9913893721944785e-05,
+      "loss": 0.5317,
+      "step": 960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991371243715648e-05,
+      "loss": 1.2056,
+      "step": 961
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991353096200869e-05,
+      "loss": 0.6534,
+      "step": 962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9913349296502564e-05,
+      "loss": 0.9767,
+      "step": 963
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991316744063926e-05,
+      "loss": 0.9296,
+      "step": 964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991298539441995e-05,
+      "loss": 0.9538,
+      "step": 965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991280315784578e-05,
+      "loss": 1.4234,
+      "step": 966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9912620730917904e-05,
+      "loss": 1.4448,
+      "step": 967
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991243811363749e-05,
+      "loss": 0.3306,
+      "step": 968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991225530600569e-05,
+      "loss": 0.542,
+      "step": 969
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991207230802369e-05,
+      "loss": 0.579,
+      "step": 970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9911889119692635e-05,
+      "loss": 0.438,
+      "step": 971
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9911705741013694e-05,
+      "loss": 1.1109,
+      "step": 972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991152217198804e-05,
+      "loss": 1.5141,
+      "step": 973
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991133841261684e-05,
+      "loss": 0.6465,
+      "step": 974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991115446290126e-05,
+      "loss": 1.3102,
+      "step": 975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9910970322842475e-05,
+      "loss": 0.9687,
+      "step": 976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9910785992441645e-05,
+      "loss": 0.5502,
+      "step": 977
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991060147169995e-05,
+      "loss": 0.3456,
+      "step": 978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991041676061858e-05,
+      "loss": 1.4105,
+      "step": 979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991023185919868e-05,
+      "loss": 0.8106,
+      "step": 980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991004676744145e-05,
+      "loss": 0.4834,
+      "step": 981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990986148534806e-05,
+      "loss": 1.2451,
+      "step": 982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.99096760129197e-05,
+      "loss": 0.4089,
+      "step": 983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990949035015753e-05,
+      "loss": 1.3578,
+      "step": 984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990930449706274e-05,
+      "loss": 1.5009,
+      "step": 985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990911845363652e-05,
+      "loss": 1.1916,
+      "step": 986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990893221988005e-05,
+      "loss": 0.2532,
+      "step": 987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990874579579451e-05,
+      "loss": 1.0601,
+      "step": 988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990855918138108e-05,
+      "loss": 0.4567,
+      "step": 989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9908372376640966e-05,
+      "loss": 0.9463,
+      "step": 990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990818538157535e-05,
+      "loss": 1.0196,
+      "step": 991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990799819618541e-05,
+      "loss": 0.564,
+      "step": 992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9907810820472356e-05,
+      "loss": 0.7695,
+      "step": 993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990762325443737e-05,
+      "loss": 0.3217,
+      "step": 994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9907435498081656e-05,
+      "loss": 0.3681,
+      "step": 995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9907247551406383e-05,
+      "loss": 0.5123,
+      "step": 996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9907059414412776e-05,
+      "loss": 0.9781,
+      "step": 997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990687108710201e-05,
+      "loss": 0.8282,
+      "step": 998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.99066825694753e-05,
+      "loss": 1.0481,
+      "step": 999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990649386153384e-05,
+      "loss": 1.1435,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990630496327883e-05,
+      "loss": 1.1217,
+      "step": 1001
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990611587471147e-05,
+      "loss": 0.4725,
+      "step": 1002
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990592659583297e-05,
+      "loss": 0.2599,
+      "step": 1003
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9905737126644523e-05,
+      "loss": 0.2956,
+      "step": 1004
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9905547467147344e-05,
+      "loss": 0.2303,
+      "step": 1005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9905357617342636e-05,
+      "loss": 1.0216,
+      "step": 1006
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9905167577231615e-05,
+      "loss": 0.4012,
+      "step": 1007
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990497734681548e-05,
+      "loss": 0.9425,
+      "step": 1008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990478692609545e-05,
+      "loss": 1.1702,
+      "step": 1009
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990459631507272e-05,
+      "loss": 0.7712,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9904405513748524e-05,
+      "loss": 0.8091,
+      "step": 1011
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990421452212407e-05,
+      "loss": 0.548,
+      "step": 1012
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990402334020057e-05,
+      "loss": 1.1248,
+      "step": 1013
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9903831967979236e-05,
+      "loss": 1.0739,
+      "step": 1014
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990364040546129e-05,
+      "loss": 1.4044,
+      "step": 1015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990344865264796e-05,
+      "loss": 0.9397,
+      "step": 1016
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990325670954046e-05,
+      "loss": 0.6002,
+      "step": 1017
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990306457614e-05,
+      "loss": 2.0723,
+      "step": 1018
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9902872252447825e-05,
+      "loss": 0.3973,
+      "step": 1019
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990267973846514e-05,
+      "loss": 0.6684,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990248703419318e-05,
+      "loss": 0.9518,
+      "step": 1021
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990229413963316e-05,
+      "loss": 0.5254,
+      "step": 1022
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990210105478633e-05,
+      "loss": 0.9565,
+      "step": 1023
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9901907779653884e-05,
+      "loss": 0.8618,
+      "step": 1024
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990171431423709e-05,
+      "loss": 0.4014,
+      "step": 1025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9901520658537146e-05,
+      "loss": 0.6232,
+      "step": 1026
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990132681255531e-05,
+      "loss": 0.9021,
+      "step": 1027
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.99011327762928e-05,
+      "loss": 0.9832,
+      "step": 1028
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990093854975086e-05,
+      "loss": 1.5473,
+      "step": 1029
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990074413293073e-05,
+      "loss": 0.8174,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990054952583362e-05,
+      "loss": 0.5064,
+      "step": 1031
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9900354728460804e-05,
+      "loss": 0.4166,
+      "step": 1032
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.99001597408135e-05,
+      "loss": 0.8929,
+      "step": 1033
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.989996456289295e-05,
+      "loss": 0.7764,
+      "step": 1034
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.98997691947004e-05,
+      "loss": 2.2356,
+      "step": 1035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.98995736362371e-05,
+      "loss": 1.2962,
+      "step": 1036
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989937788750428e-05,
+      "loss": 0.4391,
+      "step": 1037
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.98991819485032e-05,
+      "loss": 0.9117,
+      "step": 1038
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.98989858192351e-05,
+      "loss": 0.2917,
+      "step": 1039
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989878949970122e-05,
+      "loss": 1.262,
+      "step": 1040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989859298990283e-05,
+      "loss": 0.941,
+      "step": 1041
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9898396289841154e-05,
+      "loss": 1.0029,
+      "step": 1042
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989819939951747e-05,
+      "loss": 0.2558,
+      "step": 1043
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989800231893302e-05,
+      "loss": 1.1906,
+      "step": 1044
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989780504808904e-05,
+      "loss": 0.599,
+      "step": 1045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9897607586986826e-05,
+      "loss": 0.9021,
+      "step": 1046
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.98974099356276e-05,
+      "loss": 0.8136,
+      "step": 1047
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9897212094012626e-05,
+      "loss": 1.3501,
+      "step": 1048
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989701406214317e-05,
+      "loss": 0.6223,
+      "step": 1049
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989681584002049e-05,
+      "loss": 1.1058,
+      "step": 1050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989661742764585e-05,
+      "loss": 0.3952,
+      "step": 1051
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9896418825020506e-05,
+      "loss": 1.1468,
+      "step": 1052
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989622003214573e-05,
+      "loss": 0.4558,
+      "step": 1053
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989602104902278e-05,
+      "loss": 1.1686,
+      "step": 1054
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9895821875652924e-05,
+      "loss": 1.2434,
+      "step": 1055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989562251203744e-05,
+      "loss": 1.3086,
+      "step": 1056
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9895422958177584e-05,
+      "loss": 1.0195,
+      "step": 1057
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989522321407462e-05,
+      "loss": 0.5461,
+      "step": 1058
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989502327972983e-05,
+      "loss": 0.6049,
+      "step": 1059
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989482315514449e-05,
+      "loss": 0.7353,
+      "step": 1060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9894622840319867e-05,
+      "loss": 0.9036,
+      "step": 1061
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9894422335257233e-05,
+      "loss": 1.1176,
+      "step": 1062
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989422163995787e-05,
+      "loss": 1.0178,
+      "step": 1063
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989402075442306e-05,
+      "loss": 0.9615,
+      "step": 1064
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9893819678654064e-05,
+      "loss": 1.0475,
+      "step": 1065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989361841265217e-05,
+      "loss": 0.8589,
+      "step": 1066
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989341695641867e-05,
+      "loss": 1.1158,
+      "step": 1067
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989321530995483e-05,
+      "loss": 0.3571,
+      "step": 1068
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9893013473261946e-05,
+      "loss": 0.4898,
+      "step": 1069
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9892811446341295e-05,
+      "loss": 0.2291,
+      "step": 1070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989260922919416e-05,
+      "loss": 1.3025,
+      "step": 1071
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989240682182183e-05,
+      "loss": 0.6572,
+      "step": 1072
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.98922042242256e-05,
+      "loss": 0.2266,
+      "step": 1073
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989200143640675e-05,
+      "loss": 1.293,
+      "step": 1074
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9891798458366576e-05,
+      "loss": 0.7097,
+      "step": 1075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989159529010636e-05,
+      "loss": 0.8102,
+      "step": 1076
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989139193162741e-05,
+      "loss": 0.5723,
+      "step": 1077
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989118838293102e-05,
+      "loss": 0.6829,
+      "step": 1078
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9890984644018465e-05,
+      "loss": 0.5598,
+      "step": 1079
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989078071489107e-05,
+      "loss": 0.6182,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9890576595550105e-05,
+      "loss": 1.3274,
+      "step": 1081
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989037228599688e-05,
+      "loss": 1.0231,
+      "step": 1082
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9890167786232707e-05,
+      "loss": 0.0848,
+      "step": 1083
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9889963096258874e-05,
+      "loss": 0.6395,
+      "step": 1084
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9889758216076684e-05,
+      "loss": 1.3573,
+      "step": 1085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9889553145687446e-05,
+      "loss": 0.7138,
+      "step": 1086
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988934788509246e-05,
+      "loss": 0.6657,
+      "step": 1087
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9889142434293034e-05,
+      "loss": 0.8916,
+      "step": 1088
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9888936793290476e-05,
+      "loss": 0.6092,
+      "step": 1089
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.98887309620861e-05,
+      "loss": 1.4271,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.98885249406812e-05,
+      "loss": 1.186,
+      "step": 1091
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988831872907711e-05,
+      "loss": 1.1247,
+      "step": 1092
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988811232727512e-05,
+      "loss": 0.9706,
+      "step": 1093
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988790573527656e-05,
+      "loss": 1.2722,
+      "step": 1094
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9887698953082726e-05,
+      "loss": 0.2973,
+      "step": 1095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988749198069496e-05,
+      "loss": 0.9726,
+      "step": 1096
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988728481811455e-05,
+      "loss": 0.4296,
+      "step": 1097
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988707746534284e-05,
+      "loss": 0.5022,
+      "step": 1098
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988686992238114e-05,
+      "loss": 0.5961,
+      "step": 1099
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988666218923077e-05,
+      "loss": 0.6511,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988645426589305e-05,
+      "loss": 1.2256,
+      "step": 1101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9886246152369296e-05,
+      "loss": 1.1772,
+      "step": 1102
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9886037848660846e-05,
+      "loss": 1.7857,
+      "step": 1103
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9885829354769024e-05,
+      "loss": 0.8413,
+      "step": 1104
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9885620670695144e-05,
+      "loss": 0.3054,
+      "step": 1105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9885411796440555e-05,
+      "loss": 0.9743,
+      "step": 1106
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9885202732006563e-05,
+      "loss": 0.4139,
+      "step": 1107
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988499347739452e-05,
+      "loss": 0.4964,
+      "step": 1108
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9884784032605736e-05,
+      "loss": 0.7091,
+      "step": 1109
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988457439764156e-05,
+      "loss": 0.9931,
+      "step": 1110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988436457250333e-05,
+      "loss": 0.8346,
+      "step": 1111
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9884154557192367e-05,
+      "loss": 0.5494,
+      "step": 1112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988394435171001e-05,
+      "loss": 1.3061,
+      "step": 1113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.98837339560576e-05,
+      "loss": 0.5401,
+      "step": 1114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9883523370236475e-05,
+      "loss": 1.0709,
+      "step": 1115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9883312594247974e-05,
+      "loss": 0.6027,
+      "step": 1116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988310162809344e-05,
+      "loss": 1.5028,
+      "step": 1117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9882890471774226e-05,
+      "loss": 0.3135,
+      "step": 1118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9882679125291654e-05,
+      "loss": 0.7944,
+      "step": 1119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988246758864708e-05,
+      "loss": 0.6967,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988225586184185e-05,
+      "loss": 1.0496,
+      "step": 1121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9882043944877326e-05,
+      "loss": 0.4604,
+      "step": 1122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9881831837754824e-05,
+      "loss": 0.6298,
+      "step": 1123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9881619540475716e-05,
+      "loss": 1.084,
+      "step": 1124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988140705304136e-05,
+      "loss": 0.7396,
+      "step": 1125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988119437545309e-05,
+      "loss": 1.3351,
+      "step": 1126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988098150771225e-05,
+      "loss": 0.6039,
+      "step": 1127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988076844982023e-05,
+      "loss": 0.2427,
+      "step": 1128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988055520177837e-05,
+      "loss": 0.4775,
+      "step": 1129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9880341763588014e-05,
+      "loss": 1.3579,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9880128135250534e-05,
+      "loss": 1.0938,
+      "step": 1131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.987991431676728e-05,
+      "loss": 0.6036,
+      "step": 1132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987970030813962e-05,
+      "loss": 0.8429,
+      "step": 1133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987948610936892e-05,
+      "loss": 0.6885,
+      "step": 1134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987927172045653e-05,
+      "loss": 0.5668,
+      "step": 1135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987905714140383e-05,
+      "loss": 1.3051,
+      "step": 1136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987884237221217e-05,
+      "loss": 0.3961,
+      "step": 1137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9878627412882924e-05,
+      "loss": 0.6899,
+      "step": 1138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987841226341746e-05,
+      "loss": 1.5431,
+      "step": 1139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987819692381715e-05,
+      "loss": 0.9564,
+      "step": 1140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987798139408335e-05,
+      "loss": 0.3516,
+      "step": 1141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987776567421746e-05,
+      "loss": 1.0985,
+      "step": 1142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9877549764220824e-05,
+      "loss": 0.6131,
+      "step": 1143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987733366409483e-05,
+      "loss": 0.5717,
+      "step": 1144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987711737384085e-05,
+      "loss": 0.5964,
+      "step": 1145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987690089346026e-05,
+      "loss": 1.8654,
+      "step": 1146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9876684222954445e-05,
+      "loss": 0.8197,
+      "step": 1147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9876467362324765e-05,
+      "loss": 1.0402,
+      "step": 1148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987625031157262e-05,
+      "loss": 1.0689,
+      "step": 1149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987603307069938e-05,
+      "loss": 0.9582,
+      "step": 1150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9875815639706436e-05,
+      "loss": 1.1618,
+      "step": 1151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9875598018595166e-05,
+      "loss": 0.4754,
+      "step": 1152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987538020736695e-05,
+      "loss": 0.6072,
+      "step": 1153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987516220602318e-05,
+      "loss": 0.8993,
+      "step": 1154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987494401456525e-05,
+      "loss": 1.3039,
+      "step": 1155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987472563299454e-05,
+      "loss": 0.3788,
+      "step": 1156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9874507061312436e-05,
+      "loss": 0.3224,
+      "step": 1157
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987428829952034e-05,
+      "loss": 1.7597,
+      "step": 1158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9874069347619637e-05,
+      "loss": 0.6044,
+      "step": 1159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987385020561172e-05,
+      "loss": 1.4585,
+      "step": 1160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987363087349798e-05,
+      "loss": 0.7983,
+      "step": 1161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987341135127982e-05,
+      "loss": 0.3511,
+      "step": 1162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987319163895863e-05,
+      "loss": 1.0326,
+      "step": 1163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987297173653583e-05,
+      "loss": 1.1401,
+      "step": 1164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9872751644012777e-05,
+      "loss": 0.7799,
+      "step": 1165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9872531361390915e-05,
+      "loss": 0.8379,
+      "step": 1166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987231088867162e-05,
+      "loss": 1.3704,
+      "step": 1167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9872090225856306e-05,
+      "loss": 1.2137,
+      "step": 1168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987186937294638e-05,
+      "loss": 0.7455,
+      "step": 1169
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9871648329943235e-05,
+      "loss": 0.3902,
+      "step": 1170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987142709684828e-05,
+      "loss": 1.2054,
+      "step": 1171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987120567366293e-05,
+      "loss": 0.3369,
+      "step": 1172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987098406038859e-05,
+      "loss": 0.7286,
+      "step": 1173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987076225702667e-05,
+      "loss": 0.5244,
+      "step": 1174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9870540263578586e-05,
+      "loss": 0.6615,
+      "step": 1175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987031808004574e-05,
+      "loss": 0.4507,
+      "step": 1176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987009570642956e-05,
+      "loss": 0.8982,
+      "step": 1177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9869873142731455e-05,
+      "loss": 0.8075,
+      "step": 1178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986965038895284e-05,
+      "loss": 1.0292,
+      "step": 1179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9869427445095126e-05,
+      "loss": 1.0694,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9869204311159746e-05,
+      "loss": 1.2786,
+      "step": 1181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9868980987148105e-05,
+      "loss": 1.1012,
+      "step": 1182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9868757473061634e-05,
+      "loss": 0.8826,
+      "step": 1183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986853376890176e-05,
+      "loss": 0.5006,
+      "step": 1184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986830987466989e-05,
+      "loss": 0.7524,
+      "step": 1185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986808579036747e-05,
+      "loss": 1.1605,
+      "step": 1186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986786151599591e-05,
+      "loss": 1.1111,
+      "step": 1187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9867637051556636e-05,
+      "loss": 0.333,
+      "step": 1188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986741239705108e-05,
+      "loss": 0.8858,
+      "step": 1189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986718755248068e-05,
+      "loss": 0.5438,
+      "step": 1190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9866962517846864e-05,
+      "loss": 0.3823,
+      "step": 1191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986673729315106e-05,
+      "loss": 1.0548,
+      "step": 1192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9866511878394685e-05,
+      "loss": 1.6192,
+      "step": 1193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.98662862735792e-05,
+      "loss": 1.1931,
+      "step": 1194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986606047870605e-05,
+      "loss": 0.353,
+      "step": 1195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986583449377663e-05,
+      "loss": 0.9614,
+      "step": 1196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986560831879241e-05,
+      "loss": 0.3753,
+      "step": 1197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9865381953754815e-05,
+      "loss": 1.5116,
+      "step": 1198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9865155398665294e-05,
+      "loss": 0.3428,
+      "step": 1199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986492865352528e-05,
+      "loss": 0.7692,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986470171833623e-05,
+      "loss": 0.2597,
+      "step": 1201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9864474593099576e-05,
+      "loss": 0.9104,
+      "step": 1202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986424727781677e-05,
+      "loss": 0.8567,
+      "step": 1203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986401977248925e-05,
+      "loss": 0.9326,
+      "step": 1204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9863792077118466e-05,
+      "loss": 1.2911,
+      "step": 1205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986356419170587e-05,
+      "loss": 0.9643,
+      "step": 1206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986333611625292e-05,
+      "loss": 0.4945,
+      "step": 1207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986310785076105e-05,
+      "loss": 0.7052,
+      "step": 1208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986287939523173e-05,
+      "loss": 1.0412,
+      "step": 1209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986265074966641e-05,
+      "loss": 0.982,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9862421914066534e-05,
+      "loss": 0.5036,
+      "step": 1211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986219288843356e-05,
+      "loss": 1.0965,
+      "step": 1212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9861963672768965e-05,
+      "loss": 0.4125,
+      "step": 1213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986173426707418e-05,
+      "loss": 0.7608,
+      "step": 1214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9861504671350684e-05,
+      "loss": 0.7943,
+      "step": 1215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9861274885599934e-05,
+      "loss": 0.6643,
+      "step": 1216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986104490982339e-05,
+      "loss": 0.6512,
+      "step": 1217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9860814744022505e-05,
+      "loss": 1.1138,
+      "step": 1218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986058438819877e-05,
+      "loss": 0.8373,
+      "step": 1219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986035384235363e-05,
+      "loss": 1.023,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986012310648855e-05,
+      "loss": 0.9845,
+      "step": 1221
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985989218060502e-05,
+      "loss": 0.908,
+      "step": 1222
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985966106470448e-05,
+      "loss": 0.4896,
+      "step": 1223
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985942975878843e-05,
+      "loss": 1.1295,
+      "step": 1224
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985919826285832e-05,
+      "loss": 0.5718,
+      "step": 1225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985896657691563e-05,
+      "loss": 1.528,
+      "step": 1226
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985873470096185e-05,
+      "loss": 1.1541,
+      "step": 1227
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985850263499842e-05,
+      "loss": 0.4132,
+      "step": 1228
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9858270379026855e-05,
+      "loss": 0.8125,
+      "step": 1229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.98580379330486e-05,
+      "loss": 1.1453,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985780529706516e-05,
+      "loss": 0.6032,
+      "step": 1231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985757247107801e-05,
+      "loss": 0.4604,
+      "step": 1232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9857339455088624e-05,
+      "loss": 1.1665,
+      "step": 1233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985710624909848e-05,
+      "loss": 1.0531,
+      "step": 1234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985687285310908e-05,
+      "loss": 0.9177,
+      "step": 1235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9856639267121895e-05,
+      "loss": 0.4208,
+      "step": 1236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985640549113842e-05,
+      "loss": 0.2211,
+      "step": 1237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9856171525160135e-05,
+      "loss": 0.302,
+      "step": 1238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9855937369188524e-05,
+      "loss": 0.6353,
+      "step": 1239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.98557030232251e-05,
+      "loss": 0.3606,
+      "step": 1240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9855468487271324e-05,
+      "loss": 0.6197,
+      "step": 1241
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9855233761328716e-05,
+      "loss": 0.7696,
+      "step": 1242
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985499884539875e-05,
+      "loss": 0.3138,
+      "step": 1243
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985476373948294e-05,
+      "loss": 0.3279,
+      "step": 1244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985452844358276e-05,
+      "loss": 0.7071,
+      "step": 1245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985429295769973e-05,
+      "loss": 1.2222,
+      "step": 1246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985405728183533e-05,
+      "loss": 1.096,
+      "step": 1247
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9853821415991064e-05,
+      "loss": 0.2738,
+      "step": 1248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9853585360168436e-05,
+      "loss": 0.7136,
+      "step": 1249
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985334911436894e-05,
+      "loss": 0.5939,
+      "step": 1250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985311267859411e-05,
+      "loss": 0.6351,
+      "step": 1251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.98528760528454e-05,
+      "loss": 1.6408,
+      "step": 1252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985263923712437e-05,
+      "loss": 0.846,
+      "step": 1253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985240223143248e-05,
+      "loss": 0.559,
+      "step": 1254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9852165035771265e-05,
+      "loss": 1.3103,
+      "step": 1255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9851927650142235e-05,
+      "loss": 1.171,
+      "step": 1256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985169007454689e-05,
+      "loss": 1.3802,
+      "step": 1257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985145230898674e-05,
+      "loss": 1.1768,
+      "step": 1258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985121435346331e-05,
+      "loss": 0.2226,
+      "step": 1259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.98509762079781e-05,
+      "loss": 1.1928,
+      "step": 1260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985073787253264e-05,
+      "loss": 0.7502,
+      "step": 1261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9850499347128437e-05,
+      "loss": 0.3657,
+      "step": 1262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9850260631767014e-05,
+      "loss": 0.6444,
+      "step": 1263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985002172644989e-05,
+      "loss": 0.8802,
+      "step": 1264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984978263117858e-05,
+      "loss": 0.568,
+      "step": 1265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9849543345954616e-05,
+      "loss": 0.4587,
+      "step": 1266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.98493038707795e-05,
+      "loss": 1.204,
+      "step": 1267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984906420565478e-05,
+      "loss": 0.2801,
+      "step": 1268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9848824350581964e-05,
+      "loss": 0.6,
+      "step": 1269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9848584305562587e-05,
+      "loss": 0.7783,
+      "step": 1270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984834407059818e-05,
+      "loss": 0.5533,
+      "step": 1271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9848103645690257e-05,
+      "loss": 0.6091,
+      "step": 1272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984786303084037e-05,
+      "loss": 0.407,
+      "step": 1273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984762222605001e-05,
+      "loss": 1.3683,
+      "step": 1274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984738123132076e-05,
+      "loss": 0.3727,
+      "step": 1275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984714004665412e-05,
+      "loss": 1.2181,
+      "step": 1276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984689867205165e-05,
+      "loss": 0.4414,
+      "step": 1277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984665710751486e-05,
+      "loss": 1.3013,
+      "step": 1278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.98464153530453e-05,
+      "loss": 0.2121,
+      "step": 1279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9846173408644516e-05,
+      "loss": 0.7102,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9845931274314026e-05,
+      "loss": 0.6573,
+      "step": 1281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984568895005539e-05,
+      "loss": 0.5891,
+      "step": 1282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984544643587014e-05,
+      "loss": 0.7248,
+      "step": 1283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984520373175982e-05,
+      "loss": 1.017,
+      "step": 1284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9844960837725984e-05,
+      "loss": 1.2089,
+      "step": 1285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984471775377016e-05,
+      "loss": 1.1071,
+      "step": 1286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984447447989391e-05,
+      "loss": 1.3618,
+      "step": 1287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9844231016098786e-05,
+      "loss": 1.0378,
+      "step": 1288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984398736238633e-05,
+      "loss": 0.9968,
+      "step": 1289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984374351875808e-05,
+      "loss": 0.6314,
+      "step": 1290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984349948521561e-05,
+      "loss": 0.7591,
+      "step": 1291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9843255261760456e-05,
+      "loss": 0.737,
+      "step": 1292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984301084839418e-05,
+      "loss": 0.2742,
+      "step": 1293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9842766245118334e-05,
+      "loss": 0.7152,
+      "step": 1294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9842521451934476e-05,
+      "loss": 0.4293,
+      "step": 1295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984227646884417e-05,
+      "loss": 0.6558,
+      "step": 1296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984203129584896e-05,
+      "loss": 0.6589,
+      "step": 1297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9841785932950426e-05,
+      "loss": 0.9688,
+      "step": 1298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984154038015011e-05,
+      "loss": 0.2537,
+      "step": 1299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984129463744959e-05,
+      "loss": 1.0172,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984104870485042e-05,
+      "loss": 0.9822,
+      "step": 1301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984080258235417e-05,
+      "loss": 1.304,
+      "step": 1302
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984055626996241e-05,
+      "loss": 1.0327,
+      "step": 1303
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984030976767669e-05,
+      "loss": 0.7037,
+      "step": 1304
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984006307549859e-05,
+      "loss": 0.416,
+      "step": 1305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9839816193429684e-05,
+      "loss": 1.271,
+      "step": 1306
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983956912147154e-05,
+      "loss": 0.5846,
+      "step": 1307
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983932185962574e-05,
+      "loss": 0.4476,
+      "step": 1308
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9839074407893835e-05,
+      "loss": 0.6944,
+      "step": 1309
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983882676627741e-05,
+      "loss": 0.4762,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9838578934778046e-05,
+      "loss": 1.2798,
+      "step": 1311
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9838330913397316e-05,
+      "loss": 1.4649,
+      "step": 1312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.98380827021368e-05,
+      "loss": 0.9865,
+      "step": 1313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983783430099808e-05,
+      "loss": 0.2177,
+      "step": 1314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983758570998273e-05,
+      "loss": 0.5438,
+      "step": 1315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9837336929092336e-05,
+      "loss": 0.2122,
+      "step": 1316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983708795832849e-05,
+      "loss": 0.6341,
+      "step": 1317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983683879769275e-05,
+      "loss": 0.7423,
+      "step": 1318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983658944718673e-05,
+      "loss": 1.2097,
+      "step": 1319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9836339906811994e-05,
+      "loss": 0.6911,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983609017657015e-05,
+      "loss": 0.4651,
+      "step": 1321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983584025646278e-05,
+      "loss": 0.3754,
+      "step": 1322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983559014649147e-05,
+      "loss": 0.2986,
+      "step": 1323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983533984665781e-05,
+      "loss": 0.7459,
+      "step": 1324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.98350893569634e-05,
+      "loss": 0.9162,
+      "step": 1325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983483867740983e-05,
+      "loss": 1.1588,
+      "step": 1326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983458780799871e-05,
+      "loss": 0.2967,
+      "step": 1327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9834336748731605e-05,
+      "loss": 1.1421,
+      "step": 1328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9834085499610135e-05,
+      "loss": 0.419,
+      "step": 1329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.98338340606359e-05,
+      "loss": 1.0198,
+      "step": 1330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9833582431810485e-05,
+      "loss": 0.9191,
+      "step": 1331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983333061313551e-05,
+      "loss": 0.3507,
+      "step": 1332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9833078604612566e-05,
+      "loss": 0.5406,
+      "step": 1333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9832826406243255e-05,
+      "loss": 0.4176,
+      "step": 1334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983257401802919e-05,
+      "loss": 0.9748,
+      "step": 1335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983232143997197e-05,
+      "loss": 1.2234,
+      "step": 1336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983206867207321e-05,
+      "loss": 0.98,
+      "step": 1337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983181571433451e-05,
+      "loss": 0.6652,
+      "step": 1338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983156256675748e-05,
+      "loss": 1.0509,
+      "step": 1339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983130922934374e-05,
+      "loss": 1.2525,
+      "step": 1340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983105570209489e-05,
+      "loss": 1.2173,
+      "step": 1341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9830801985012554e-05,
+      "loss": 0.6914,
+      "step": 1342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983054807809834e-05,
+      "loss": 1.4776,
+      "step": 1343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983029398135387e-05,
+      "loss": 0.8809,
+      "step": 1344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9830039694780755e-05,
+      "loss": 0.21,
+      "step": 1345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982978521838062e-05,
+      "loss": 1.771,
+      "step": 1346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982953055215508e-05,
+      "loss": 1.1672,
+      "step": 1347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982927569610575e-05,
+      "loss": 1.3271,
+      "step": 1348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982902065023425e-05,
+      "loss": 1.0076,
+      "step": 1349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9828765414542224e-05,
+      "loss": 1.2501,
+      "step": 1350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9828509989031275e-05,
+      "loss": 0.4928,
+      "step": 1351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982825437370303e-05,
+      "loss": 0.4342,
+      "step": 1352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9827998568559134e-05,
+      "loss": 1.0488,
+      "step": 1353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9827742573601196e-05,
+      "loss": 0.6044,
+      "step": 1354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982748638883085e-05,
+      "loss": 0.8611,
+      "step": 1355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982723001424973e-05,
+      "loss": 0.5924,
+      "step": 1356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9826973449859464e-05,
+      "loss": 1.0922,
+      "step": 1357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982671669566169e-05,
+      "loss": 1.445,
+      "step": 1358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982645975165803e-05,
+      "loss": 0.4924,
+      "step": 1359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982620261785013e-05,
+      "loss": 1.2243,
+      "step": 1360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9825945294239624e-05,
+      "loss": 0.294,
+      "step": 1361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982568778082815e-05,
+      "loss": 0.502,
+      "step": 1362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982543007761735e-05,
+      "loss": 1.02,
+      "step": 1363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9825172184608847e-05,
+      "loss": 0.9695,
+      "step": 1364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9824914101804296e-05,
+      "loss": 1.2078,
+      "step": 1365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982465582920535e-05,
+      "loss": 1.2938,
+      "step": 1366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982439736681363e-05,
+      "loss": 0.2824,
+      "step": 1367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982413871463079e-05,
+      "loss": 1.1618,
+      "step": 1368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982387987265848e-05,
+      "loss": 1.1872,
+      "step": 1369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9823620840898344e-05,
+      "loss": 0.6632,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982336161935203e-05,
+      "loss": 0.1458,
+      "step": 1371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982310220802119e-05,
+      "loss": 0.4226,
+      "step": 1372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982284260690747e-05,
+      "loss": 0.6113,
+      "step": 1373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982258281601252e-05,
+      "loss": 1.1409,
+      "step": 1374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9822322835338e-05,
+      "loss": 0.9909,
+      "step": 1375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982206266488557e-05,
+      "loss": 1.2272,
+      "step": 1376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9821802304656876e-05,
+      "loss": 1.1472,
+      "step": 1377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982154175465358e-05,
+      "loss": 0.65,
+      "step": 1378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982128101487733e-05,
+      "loss": 1.066,
+      "step": 1379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982102008532979e-05,
+      "loss": 0.9501,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982075896601263e-05,
+      "loss": 1.5551,
+      "step": 1381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.98204976569275e-05,
+      "loss": 0.9529,
+      "step": 1382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982023615807608e-05,
+      "loss": 1.534,
+      "step": 1383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9819974469460004e-05,
+      "loss": 1.1275,
+      "step": 1384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9819712591080955e-05,
+      "loss": 0.5649,
+      "step": 1385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981945052294061e-05,
+      "loss": 0.9163,
+      "step": 1386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9819188265040626e-05,
+      "loss": 0.3621,
+      "step": 1387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981892581738266e-05,
+      "loss": 0.5598,
+      "step": 1388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.98186631799684e-05,
+      "loss": 0.6739,
+      "step": 1389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9818400352799505e-05,
+      "loss": 1.2023,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9818137335877665e-05,
+      "loss": 0.6143,
+      "step": 1391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981787412920454e-05,
+      "loss": 1.1682,
+      "step": 1392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9817610732781805e-05,
+      "loss": 1.1159,
+      "step": 1393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981734714661114e-05,
+      "loss": 1.2014,
+      "step": 1394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981708337069422e-05,
+      "loss": 0.625,
+      "step": 1395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9816819405032715e-05,
+      "loss": 0.5495,
+      "step": 1396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9816555249628324e-05,
+      "loss": 0.4957,
+      "step": 1397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981629090448272e-05,
+      "loss": 1.1734,
+      "step": 1398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981602636959758e-05,
+      "loss": 1.5285,
+      "step": 1399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9815761644974584e-05,
+      "loss": 0.9433,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981549673061544e-05,
+      "loss": 0.6394,
+      "step": 1401
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9815231626521805e-05,
+      "loss": 0.4987,
+      "step": 1402
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9814966332695374e-05,
+      "loss": 0.9429,
+      "step": 1403
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981470084913785e-05,
+      "loss": 0.1641,
+      "step": 1404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981443517585091e-05,
+      "loss": 0.5173,
+      "step": 1405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981416931283624e-05,
+      "loss": 0.737,
+      "step": 1406
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981390326009555e-05,
+      "loss": 0.8649,
+      "step": 1407
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981363701763051e-05,
+      "loss": 1.3453,
+      "step": 1408
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981337058544283e-05,
+      "loss": 0.9963,
+      "step": 1409
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.98131039635342e-05,
+      "loss": 1.185,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9812837151906325e-05,
+      "loss": 0.341,
+      "step": 1411
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981257015056089e-05,
+      "loss": 1.0465,
+      "step": 1412
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.98123029594996e-05,
+      "loss": 0.3065,
+      "step": 1413
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9812035578724146e-05,
+      "loss": 0.3551,
+      "step": 1414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981176800823625e-05,
+      "loss": 0.9341,
+      "step": 1415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.98115002480376e-05,
+      "loss": 1.7064,
+      "step": 1416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981123229812991e-05,
+      "loss": 1.5043,
+      "step": 1417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9810964158514866e-05,
+      "loss": 0.9056,
+      "step": 1418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9810695829194196e-05,
+      "loss": 0.5721,
+      "step": 1419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.98104273101696e-05,
+      "loss": 0.7939,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9810158601442775e-05,
+      "loss": 0.7,
+      "step": 1421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9809889703015453e-05,
+      "loss": 0.9804,
+      "step": 1422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.980962061488932e-05,
+      "loss": 0.423,
+      "step": 1423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.980935133706611e-05,
+      "loss": 1.2736,
+      "step": 1424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.980908186954753e-05,
+      "loss": 1.3671,
+      "step": 1425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.980881221233529e-05,
+      "loss": 1.6594,
+      "step": 1426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.980854236543111e-05,
+      "loss": 1.369,
+      "step": 1427
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.98082723288367e-05,
+      "loss": 1.0586,
+      "step": 1428
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9808002102553795e-05,
+      "loss": 0.978,
+      "step": 1429
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9807731686584095e-05,
+      "loss": 0.8498,
+      "step": 1430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9807461080929336e-05,
+      "loss": 0.7775,
+      "step": 1431
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9807190285591224e-05,
+      "loss": 1.278,
+      "step": 1432
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.98069193005715e-05,
+      "loss": 0.9634,
+      "step": 1433
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980664812587189e-05,
+      "loss": 0.3161,
+      "step": 1434
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980637676149409e-05,
+      "loss": 1.0848,
+      "step": 1435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980610520743986e-05,
+      "loss": 0.314,
+      "step": 1436
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980583346371092e-05,
+      "loss": 0.43,
+      "step": 1437
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980556153030899e-05,
+      "loss": 1.035,
+      "step": 1438
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9805289407235794e-05,
+      "loss": 0.448,
+      "step": 1439
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9805017094493085e-05,
+      "loss": 0.5084,
+      "step": 1440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980474459208259e-05,
+      "loss": 0.4192,
+      "step": 1441
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980447190000603e-05,
+      "loss": 0.5493,
+      "step": 1442
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9804199018265144e-05,
+      "loss": 0.4352,
+      "step": 1443
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980392594686168e-05,
+      "loss": 1.202,
+      "step": 1444
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980365268579738e-05,
+      "loss": 0.4058,
+      "step": 1445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9803379235073964e-05,
+      "loss": 1.2179,
+      "step": 1446
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9803105594693176e-05,
+      "loss": 0.4104,
+      "step": 1447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980283176465676e-05,
+      "loss": 0.4885,
+      "step": 1448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980255774496647e-05,
+      "loss": 0.8567,
+      "step": 1449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980228353562403e-05,
+      "loss": 1.1628,
+      "step": 1450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980200913663121e-05,
+      "loss": 1.2568,
+      "step": 1451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980173454798973e-05,
+      "loss": 1.3571,
+      "step": 1452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980145976970134e-05,
+      "loss": 1.1845,
+      "step": 1453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9801184801767814e-05,
+      "loss": 0.4744,
+      "step": 1454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9800909644190874e-05,
+      "loss": 0.8657,
+      "step": 1455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980063429697229e-05,
+      "loss": 0.2679,
+      "step": 1456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9800358760113804e-05,
+      "loss": 1.4714,
+      "step": 1457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980008303361716e-05,
+      "loss": 0.615,
+      "step": 1458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9799807117484136e-05,
+      "loss": 0.3973,
+      "step": 1459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979953101171647e-05,
+      "loss": 0.2275,
+      "step": 1460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979925471631594e-05,
+      "loss": 1.2127,
+      "step": 1461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9798978231284265e-05,
+      "loss": 1.2615,
+      "step": 1462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9798701556623246e-05,
+      "loss": 0.5551,
+      "step": 1463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9798424692334616e-05,
+      "loss": 0.481,
+      "step": 1464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979814763842015e-05,
+      "loss": 1.4045,
+      "step": 1465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9797870394881614e-05,
+      "loss": 1.1194,
+      "step": 1466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979759296172077e-05,
+      "loss": 0.4279,
+      "step": 1467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979731533893937e-05,
+      "loss": 1.0279,
+      "step": 1468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979703752653919e-05,
+      "loss": 1.4037,
+      "step": 1469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9796759524522006e-05,
+      "loss": 0.7757,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9796481332889576e-05,
+      "loss": 0.5507,
+      "step": 1471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979620295164367e-05,
+      "loss": 0.9744,
+      "step": 1472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979592438078607e-05,
+      "loss": 0.5733,
+      "step": 1473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9795645620318545e-05,
+      "loss": 0.3869,
+      "step": 1474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979536667024285e-05,
+      "loss": 1.2466,
+      "step": 1475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979508753056079e-05,
+      "loss": 0.4342,
+      "step": 1476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979480820127413e-05,
+      "loss": 0.3171,
+      "step": 1477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9794528682384645e-05,
+      "loss": 0.6888,
+      "step": 1478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.97942489738941e-05,
+      "loss": 1.0982,
+      "step": 1479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.97939690758043e-05,
+      "loss": 1.3033,
+      "step": 1480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979368898811701e-05,
+      "loss": 1.2741,
+      "step": 1481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979340871083402e-05,
+      "loss": 0.8522,
+      "step": 1482
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979312824395711e-05,
+      "loss": 0.2461,
+      "step": 1483
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979284758748807e-05,
+      "loss": 0.8648,
+      "step": 1484
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9792566741428675e-05,
+      "loss": 1.0591,
+      "step": 1485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979228570578072e-05,
+      "loss": 0.8973,
+      "step": 1486
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9792004480546e-05,
+      "loss": 0.6757,
+      "step": 1487
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979172306572628e-05,
+      "loss": 0.8219,
+      "step": 1488
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9791441461323386e-05,
+      "loss": 0.9454,
+      "step": 1489
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979115966733909e-05,
+      "loss": 0.407,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9790877683775174e-05,
+      "loss": 0.9106,
+      "step": 1491
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9790595510633456e-05,
+      "loss": 0.3629,
+      "step": 1492
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9790313147915715e-05,
+      "loss": 0.6982,
+      "step": 1493
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979003059562377e-05,
+      "loss": 0.9565,
+      "step": 1494
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978974785375938e-05,
+      "loss": 0.5486,
+      "step": 1495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978946492232438e-05,
+      "loss": 0.9852,
+      "step": 1496
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978918180132056e-05,
+      "loss": 1.2247,
+      "step": 1497
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978889849074971e-05,
+      "loss": 1.9676,
+      "step": 1498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978861499061365e-05,
+      "loss": 0.5637,
+      "step": 1499
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978833130091417e-05,
+      "loss": 0.5143,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978804742165308e-05,
+      "loss": 1.1102,
+      "step": 1501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.97877633528322e-05,
+      "loss": 0.374,
+      "step": 1502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9787479094453314e-05,
+      "loss": 1.2437,
+      "step": 1503
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978719464651825e-05,
+      "loss": 1.2112,
+      "step": 1504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978691000902881e-05,
+      "loss": 1.9989,
+      "step": 1505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978662518198681e-05,
+      "loss": 0.4426,
+      "step": 1506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978634016539405e-05,
+      "loss": 0.8182,
+      "step": 1507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978605495925236e-05,
+      "loss": 1.0572,
+      "step": 1508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978576956356354e-05,
+      "loss": 1.3305,
+      "step": 1509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9785483978329416e-05,
+      "loss": 0.7135,
+      "step": 1510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.97851982035518e-05,
+      "loss": 0.3329,
+      "step": 1511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978491223923253e-05,
+      "loss": 0.5997,
+      "step": 1512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9784626085373394e-05,
+      "loss": 0.9828,
+      "step": 1513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978433974197623e-05,
+      "loss": 0.3034,
+      "step": 1514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978405320904286e-05,
+      "loss": 1.2557,
+      "step": 1515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9783766486575104e-05,
+      "loss": 1.1023,
+      "step": 1516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978347957457479e-05,
+      "loss": 1.0233,
+      "step": 1517
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978319247304374e-05,
+      "loss": 0.192,
+      "step": 1518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978290518198379e-05,
+      "loss": 1.1028,
+      "step": 1519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978261770139675e-05,
+      "loss": 0.6244,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978233003128447e-05,
+      "loss": 0.4375,
+      "step": 1521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978204217164877e-05,
+      "loss": 0.3215,
+      "step": 1522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9781754122491486e-05,
+      "loss": 0.4045,
+      "step": 1523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978146588381444e-05,
+      "loss": 0.57,
+      "step": 1524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978117745561948e-05,
+      "loss": 0.4967,
+      "step": 1525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978088883790843e-05,
+      "loss": 0.8931,
+      "step": 1526
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978060003068314e-05,
+      "loss": 0.8107,
+      "step": 1527
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978031103394544e-05,
+      "loss": 1.0542,
+      "step": 1528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978002184769717e-05,
+      "loss": 0.3544,
+      "step": 1529
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977973247194017e-05,
+      "loss": 0.9444,
+      "step": 1530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977944290667627e-05,
+      "loss": 0.4681,
+      "step": 1531
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977915315190733e-05,
+      "loss": 1.0468,
+      "step": 1532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9778863207635196e-05,
+      "loss": 1.656,
+      "step": 1533
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.97785730738617e-05,
+      "loss": 1.3162,
+      "step": 1534
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.97782827505887e-05,
+      "loss": 1.266,
+      "step": 1535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9777992237818026e-05,
+      "loss": 0.8591,
+      "step": 1536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977770153555154e-05,
+      "loss": 0.5086,
+      "step": 1537
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.97774106437911e-05,
+      "loss": 1.5029,
+      "step": 1538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977711956253853e-05,
+      "loss": 1.0147,
+      "step": 1539
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977682829179571e-05,
+      "loss": 0.5487,
+      "step": 1540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9776536831564484e-05,
+      "loss": 0.5364,
+      "step": 1541
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977624518184669e-05,
+      "loss": 0.6845,
+      "step": 1542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977595334264422e-05,
+      "loss": 0.0891,
+      "step": 1543
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977566131395889e-05,
+      "loss": 1.3858,
+      "step": 1544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977536909579259e-05,
+      "loss": 0.5517,
+      "step": 1545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9775076688147175e-05,
+      "loss": 0.6742,
+      "step": 1546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977478409102448e-05,
+      "loss": 0.3489,
+      "step": 1547
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.97744913044264e-05,
+      "loss": 1.0084,
+      "step": 1548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977419832835478e-05,
+      "loss": 0.7246,
+      "step": 1549
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977390516281149e-05,
+      "loss": 0.298,
+      "step": 1550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.97736118077984e-05,
+      "loss": 0.6454,
+      "step": 1551
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977331826331736e-05,
+      "loss": 0.8041,
+      "step": 1552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9773024529370246e-05,
+      "loss": 0.869,
+      "step": 1553
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977273060595895e-05,
+      "loss": 0.6331,
+      "step": 1554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9772436493085304e-05,
+      "loss": 0.0996,
+      "step": 1555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.97721421907512e-05,
+      "loss": 1.5555,
+      "step": 1556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9771847698958514e-05,
+      "loss": 1.0741,
+      "step": 1557
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9771553017709115e-05,
+      "loss": 1.0844,
+      "step": 1558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977125814700488e-05,
+      "loss": 1.0447,
+      "step": 1559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977096308684768e-05,
+      "loss": 0.5347,
+      "step": 1560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.97706678372394e-05,
+      "loss": 0.6057,
+      "step": 1561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9770372398181904e-05,
+      "loss": 0.6366,
+      "step": 1562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9770076769677094e-05,
+      "loss": 0.3858,
+      "step": 1563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976978095172684e-05,
+      "loss": 0.8321,
+      "step": 1564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9769484944333025e-05,
+      "loss": 0.8182,
+      "step": 1565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9769188747497534e-05,
+      "loss": 0.5166,
+      "step": 1566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976889236122224e-05,
+      "loss": 0.6648,
+      "step": 1567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976859578550905e-05,
+      "loss": 0.3111,
+      "step": 1568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976829902035984e-05,
+      "loss": 0.8296,
+      "step": 1569
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.97680020657765e-05,
+      "loss": 0.523,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9767704921760916e-05,
+      "loss": 0.854,
+      "step": 1571
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9767407588314986e-05,
+      "loss": 1.471,
+      "step": 1572
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976711006544059e-05,
+      "loss": 1.1114,
+      "step": 1573
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9766812353139635e-05,
+      "loss": 1.3511,
+      "step": 1574
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9766514451414015e-05,
+      "loss": 0.5587,
+      "step": 1575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976621636026561e-05,
+      "loss": 1.2171,
+      "step": 1576
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976591807969634e-05,
+      "loss": 0.6853,
+      "step": 1577
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976561960970808e-05,
+      "loss": 0.8367,
+      "step": 1578
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976532095030274e-05,
+      "loss": 0.7848,
+      "step": 1579
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9765022101482226e-05,
+      "loss": 0.4049,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9764723063248435e-05,
+      "loss": 0.9316,
+      "step": 1581
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976442383560327e-05,
+      "loss": 0.9644,
+      "step": 1582
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976412441854864e-05,
+      "loss": 0.2059,
+      "step": 1583
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976382481208643e-05,
+      "loss": 1.0027,
+      "step": 1584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976352501621857e-05,
+      "loss": 1.2819,
+      "step": 1585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9763225030946964e-05,
+      "loss": 0.4628,
+      "step": 1586
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976292485627351e-05,
+      "loss": 1.1785,
+      "step": 1587
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976262449220013e-05,
+      "loss": 0.4949,
+      "step": 1588
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976232393872872e-05,
+      "loss": 0.3875,
+      "step": 1589
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976202319586121e-05,
+      "loss": 0.7528,
+      "step": 1590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.97617222635995e-05,
+      "loss": 0.3443,
+      "step": 1591
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976142114194553e-05,
+      "loss": 0.9777,
+      "step": 1592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9761119830901176e-05,
+      "loss": 0.5877,
+      "step": 1593
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976081833046839e-05,
+      "loss": 0.391,
+      "step": 1594
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976051664064907e-05,
+      "loss": 1.1929,
+      "step": 1595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976021476144515e-05,
+      "loss": 0.7854,
+      "step": 1596
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975991269285854e-05,
+      "loss": 1.5092,
+      "step": 1597
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975961043489116e-05,
+      "loss": 0.3309,
+      "step": 1598
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9759307987544954e-05,
+      "loss": 1.1042,
+      "step": 1599
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9759005350821824e-05,
+      "loss": 0.3911,
+      "step": 1600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975870252472372e-05,
+      "loss": 1.4275,
+      "step": 1601
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9758399509252536e-05,
+      "loss": 0.7206,
+      "step": 1602
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975809630441022e-05,
+      "loss": 1.8159,
+      "step": 1603
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9757792910198705e-05,
+      "loss": 0.3662,
+      "step": 1604
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975748932661992e-05,
+      "loss": 1.221,
+      "step": 1605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975718555367579e-05,
+      "loss": 0.3864,
+      "step": 1606
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975688159136824e-05,
+      "loss": 0.3457,
+      "step": 1607
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9756577439699235e-05,
+      "loss": 1.1481,
+      "step": 1608
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975627309867068e-05,
+      "loss": 0.66,
+      "step": 1609
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975596856828452e-05,
+      "loss": 0.835,
+      "step": 1610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9755663848542705e-05,
+      "loss": 0.6892,
+      "step": 1611
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975535893944716e-05,
+      "loss": 0.3613,
+      "step": 1612
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975505384099983e-05,
+      "loss": 0.778,
+      "step": 1613
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9754748553202656e-05,
+      "loss": 1.2153,
+      "step": 1614
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9754443076057585e-05,
+      "loss": 0.2522,
+      "step": 1615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975413740956655e-05,
+      "loss": 0.6208,
+      "step": 1616
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9753831553731515e-05,
+      "loss": 0.4896,
+      "step": 1617
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975352550855441e-05,
+      "loss": 0.3359,
+      "step": 1618
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975321927403719e-05,
+      "loss": 1.0462,
+      "step": 1619
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975291285018179e-05,
+      "loss": 0.338,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.975260623699019e-05,
+      "loss": 0.7782,
+      "step": 1621
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9752299434464304e-05,
+      "loss": 1.2811,
+      "step": 1622
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9751992442606114e-05,
+      "loss": 0.9133,
+      "step": 1623
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9751685261417554e-05,
+      "loss": 0.5137,
+      "step": 1624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9751377890900596e-05,
+      "loss": 0.3997,
+      "step": 1625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975107033105717e-05,
+      "loss": 1.2322,
+      "step": 1626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975076258188927e-05,
+      "loss": 0.4079,
+      "step": 1627
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9750454643398824e-05,
+      "loss": 0.3107,
+      "step": 1628
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975014651558781e-05,
+      "loss": 1.2167,
+      "step": 1629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974983819845818e-05,
+      "loss": 0.3223,
+      "step": 1630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974952969201189e-05,
+      "loss": 1.3834,
+      "step": 1631
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9749220996250915e-05,
+      "loss": 1.122,
+      "step": 1632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9748912111177214e-05,
+      "loss": 0.3034,
+      "step": 1633
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9748603036792754e-05,
+      "loss": 0.5497,
+      "step": 1634
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97482937730995e-05,
+      "loss": 0.9487,
+      "step": 1635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974798432009942e-05,
+      "loss": 1.1578,
+      "step": 1636
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97476746777945e-05,
+      "loss": 0.9865,
+      "step": 1637
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974736484618667e-05,
+      "loss": 0.5843,
+      "step": 1638
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974705482527794e-05,
+      "loss": 0.9233,
+      "step": 1639
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974674461507027e-05,
+      "loss": 0.8369,
+      "step": 1640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9746434215565624e-05,
+      "loss": 1.2071,
+      "step": 1641
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9746123626766004e-05,
+      "loss": 0.7656,
+      "step": 1642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974581284867335e-05,
+      "loss": 1.2365,
+      "step": 1643
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974550188128967e-05,
+      "loss": 0.7545,
+      "step": 1644
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974519072461694e-05,
+      "loss": 0.6029,
+      "step": 1645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974487937865711e-05,
+      "loss": 0.6524,
+      "step": 1646
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97445678434122e-05,
+      "loss": 1.4184,
+      "step": 1647
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9744256118884174e-05,
+      "loss": 0.9958,
+      "step": 1648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974394420507501e-05,
+      "loss": 0.6584,
+      "step": 1649
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97436321019867e-05,
+      "loss": 1.3797,
+      "step": 1650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9743319809621236e-05,
+      "loss": 0.6361,
+      "step": 1651
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9743007327980596e-05,
+      "loss": 0.7146,
+      "step": 1652
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974269465706678e-05,
+      "loss": 0.7646,
+      "step": 1653
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974238179688177e-05,
+      "loss": 0.8582,
+      "step": 1654
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974206874742754e-05,
+      "loss": 1.0352,
+      "step": 1655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9741755508706116e-05,
+      "loss": 0.5479,
+      "step": 1656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974144208071947e-05,
+      "loss": 0.7961,
+      "step": 1657
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9741128463469594e-05,
+      "loss": 0.4908,
+      "step": 1658
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97408146569585e-05,
+      "loss": 1.0629,
+      "step": 1659
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974050066118817e-05,
+      "loss": 0.7654,
+      "step": 1660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974018647616061e-05,
+      "loss": 0.608,
+      "step": 1661
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973987210187782e-05,
+      "loss": 2.1207,
+      "step": 1662
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973955753834179e-05,
+      "loss": 0.6886,
+      "step": 1663
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9739242785554524e-05,
+      "loss": 0.8175,
+      "step": 1664
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973892784351805e-05,
+      "loss": 1.1177,
+      "step": 1665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973861271223434e-05,
+      "loss": 0.797,
+      "step": 1666
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9738297391705415e-05,
+      "loss": 0.5413,
+      "step": 1667
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973798188193327e-05,
+      "loss": 1.2156,
+      "step": 1668
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973766618291993e-05,
+      "loss": 0.4501,
+      "step": 1669
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973735029466739e-05,
+      "loss": 0.4866,
+      "step": 1670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9737034217177676e-05,
+      "loss": 0.9219,
+      "step": 1671
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973671795045277e-05,
+      "loss": 0.5281,
+      "step": 1672
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973640149449472e-05,
+      "loss": 1.1787,
+      "step": 1673
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973608484930552e-05,
+      "loss": 1.8833,
+      "step": 1674
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9735768014887184e-05,
+      "loss": 1.384,
+      "step": 1675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973545099124174e-05,
+      "loss": 0.4122,
+      "step": 1676
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973513377837119e-05,
+      "loss": 1.2693,
+      "step": 1677
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9734816376277554e-05,
+      "loss": 0.3397,
+      "step": 1678
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973449878496287e-05,
+      "loss": 0.9303,
+      "step": 1679
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973418100442915e-05,
+      "loss": 0.676,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973386303467841e-05,
+      "loss": 0.8367,
+      "step": 1681
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973354487571267e-05,
+      "loss": 0.5123,
+      "step": 1682
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9733226527533966e-05,
+      "loss": 0.8957,
+      "step": 1683
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9732907990144325e-05,
+      "loss": 1.4406,
+      "step": 1684
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9732589263545765e-05,
+      "loss": 0.3705,
+      "step": 1685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973227034774031e-05,
+      "loss": 1.2025,
+      "step": 1686
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973195124273001e-05,
+      "loss": 0.4599,
+      "step": 1687
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973163194851687e-05,
+      "loss": 1.2127,
+      "step": 1688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973131246510294e-05,
+      "loss": 0.1462,
+      "step": 1689
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973099279249025e-05,
+      "loss": 1.0677,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9730672930680836e-05,
+      "loss": 0.5677,
+      "step": 1691
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9730352879676725e-05,
+      "loss": 0.5824,
+      "step": 1692
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973003263947996e-05,
+      "loss": 1.0222,
+      "step": 1693
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9729712210092575e-05,
+      "loss": 0.2273,
+      "step": 1694
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972939159151661e-05,
+      "loss": 1.3486,
+      "step": 1695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972907078375411e-05,
+      "loss": 1.2736,
+      "step": 1696
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972874978680711e-05,
+      "loss": 1.1287,
+      "step": 1697
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972842860067766e-05,
+      "loss": 0.2947,
+      "step": 1698
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972810722536779e-05,
+      "loss": 1.2875,
+      "step": 1699
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972778566087957e-05,
+      "loss": 1.3454,
+      "step": 1700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972746390721502e-05,
+      "loss": 1.3171,
+      "step": 1701
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97271419643762e-05,
+      "loss": 0.8134,
+      "step": 1702
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972681983236515e-05,
+      "loss": 1.9027,
+      "step": 1703
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972649751118394e-05,
+      "loss": 0.9059,
+      "step": 1704
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97261750008346e-05,
+      "loss": 0.8212,
+      "step": 1705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9725852301319197e-05,
+      "loss": 1.1876,
+      "step": 1706
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972552941263978e-05,
+      "loss": 0.692,
+      "step": 1707
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972520633479839e-05,
+      "loss": 0.5219,
+      "step": 1708
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97248830677971e-05,
+      "loss": 0.9708,
+      "step": 1709
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972455961163797e-05,
+      "loss": 1.39,
+      "step": 1710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9724235966323046e-05,
+      "loss": 0.7811,
+      "step": 1711
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972391213185439e-05,
+      "loss": 0.5763,
+      "step": 1712
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972358810823407e-05,
+      "loss": 0.2155,
+      "step": 1713
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972326389546414e-05,
+      "loss": 0.7395,
+      "step": 1714
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972293949354666e-05,
+      "loss": 0.4732,
+      "step": 1715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972261490248371e-05,
+      "loss": 1.476,
+      "step": 1716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972229012227734e-05,
+      "loss": 0.6725,
+      "step": 1717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972196515292962e-05,
+      "loss": 1.0131,
+      "step": 1718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972163999444263e-05,
+      "loss": 0.9768,
+      "step": 1719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9721314646818426e-05,
+      "loss": 1.0049,
+      "step": 1720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972098911005908e-05,
+      "loss": 0.3555,
+      "step": 1721
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972066338416666e-05,
+      "loss": 0.2915,
+      "step": 1722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972033746914325e-05,
+      "loss": 0.1314,
+      "step": 1723
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.972001136499092e-05,
+      "loss": 0.5318,
+      "step": 1724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971968507171174e-05,
+      "loss": 0.6849,
+      "step": 1725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971935858930779e-05,
+      "loss": 0.6314,
+      "step": 1726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971903191778116e-05,
+      "loss": 0.4396,
+      "step": 1727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9718705057133905e-05,
+      "loss": 0.6359,
+      "step": 1728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9718378007368117e-05,
+      "loss": 0.5555,
+      "step": 1729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971805076848588e-05,
+      "loss": 0.8444,
+      "step": 1730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971772334048926e-05,
+      "loss": 0.3761,
+      "step": 1731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971739572338037e-05,
+      "loss": 1.2886,
+      "step": 1732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971706791716128e-05,
+      "loss": 1.5773,
+      "step": 1733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971673992183407e-05,
+      "loss": 1.0312,
+      "step": 1734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9716411737400836e-05,
+      "loss": 0.6728,
+      "step": 1735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971608336386365e-05,
+      "loss": 0.3699,
+      "step": 1736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971575480122462e-05,
+      "loss": 0.5057,
+      "step": 1737
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971542604948584e-05,
+      "loss": 1.6902,
+      "step": 1738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9715097108649376e-05,
+      "loss": 0.6212,
+      "step": 1739
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971476797871735e-05,
+      "loss": 0.8131,
+      "step": 1740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971443865969184e-05,
+      "loss": 1.0371,
+      "step": 1741
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971410915157495e-05,
+      "loss": 0.8871,
+      "step": 1742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971377945436877e-05,
+      "loss": 0.4749,
+      "step": 1743
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971344956807541e-05,
+      "loss": 1.3856,
+      "step": 1744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971311949269695e-05,
+      "loss": 0.8585,
+      "step": 1745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9712789228235506e-05,
+      "loss": 0.8252,
+      "step": 1746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971245877469317e-05,
+      "loss": 0.9857,
+      "step": 1747
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971212813207205e-05,
+      "loss": 0.899,
+      "step": 1748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971179730037425e-05,
+      "loss": 0.6917,
+      "step": 1749
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9711466279601875e-05,
+      "loss": 0.5349,
+      "step": 1750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971113506975704e-05,
+      "loss": 0.5645,
+      "step": 1751
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971080367084183e-05,
+      "loss": 0.5357,
+      "step": 1752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971047208285838e-05,
+      "loss": 1.3394,
+      "step": 1753
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971014030580879e-05,
+      "loss": 0.5505,
+      "step": 1754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970980833969516e-05,
+      "loss": 0.5869,
+      "step": 1755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970947618451962e-05,
+      "loss": 0.4634,
+      "step": 1756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970914384028426e-05,
+      "loss": 1.0956,
+      "step": 1757
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9708811306991225e-05,
+      "loss": 1.302,
+      "step": 1758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970847858464262e-05,
+      "loss": 0.6571,
+      "step": 1759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9708145673240545e-05,
+      "loss": 0.5773,
+      "step": 1760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9707812572787136e-05,
+      "loss": 1.7692,
+      "step": 1761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970747928328451e-05,
+      "loss": 1.4014,
+      "step": 1762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970714580473479e-05,
+      "loss": 0.3864,
+      "step": 1763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9706812137140104e-05,
+      "loss": 1.1486,
+      "step": 1764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970647828050256e-05,
+      "loss": 1.6944,
+      "step": 1765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970614423482428e-05,
+      "loss": 0.8864,
+      "step": 1766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970581000010741e-05,
+      "loss": 0.8034,
+      "step": 1767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970547557635406e-05,
+      "loss": 1.378,
+      "step": 1768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9705140963566366e-05,
+      "loss": 0.8189,
+      "step": 1769
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970480616174645e-05,
+      "loss": 0.8623,
+      "step": 1770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970447117089645e-05,
+      "loss": 1.4196,
+      "step": 1771
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9704135991018505e-05,
+      "loss": 0.3533,
+      "step": 1772
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970380062211473e-05,
+      "loss": 1.1769,
+      "step": 1773
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970346506418727e-05,
+      "loss": 1.2751,
+      "step": 1774
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970312931723826e-05,
+      "loss": 2.772,
+      "step": 1775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9702793381269826e-05,
+      "loss": 0.492,
+      "step": 1776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9702457256284125e-05,
+      "loss": 1.2609,
+      "step": 1777
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970212094228328e-05,
+      "loss": 0.8036,
+      "step": 1778
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970178443926944e-05,
+      "loss": 1.4552,
+      "step": 1779
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9701447747244734e-05,
+      "loss": 0.1135,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970111086621133e-05,
+      "loss": 0.5206,
+      "step": 1781
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9700773796171345e-05,
+      "loss": 1.5116,
+      "step": 1782
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9700436537126936e-05,
+      "loss": 1.0096,
+      "step": 1783
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9700099089080244e-05,
+      "loss": 1.0872,
+      "step": 1784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9699761452033424e-05,
+      "loss": 0.5911,
+      "step": 1785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9699423625988616e-05,
+      "loss": 0.9611,
+      "step": 1786
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969908561094797e-05,
+      "loss": 0.5215,
+      "step": 1787
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969874740691365e-05,
+      "loss": 0.6323,
+      "step": 1788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9698409013887805e-05,
+      "loss": 1.1345,
+      "step": 1789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969807043187257e-05,
+      "loss": 1.1169,
+      "step": 1790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9697731660870116e-05,
+      "loss": 0.4197,
+      "step": 1791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.96973927008826e-05,
+      "loss": 0.4604,
+      "step": 1792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9697053551912173e-05,
+      "loss": 1.4395,
+      "step": 1793
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9696714213960996e-05,
+      "loss": 0.2961,
+      "step": 1794
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9696374687031224e-05,
+      "loss": 0.634,
+      "step": 1795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9696034971125014e-05,
+      "loss": 0.4418,
+      "step": 1796
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969569506624455e-05,
+      "loss": 1.0772,
+      "step": 1797
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969535497239197e-05,
+      "loss": 0.8826,
+      "step": 1798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969501468956944e-05,
+      "loss": 1.3446,
+      "step": 1799
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969467421777915e-05,
+      "loss": 0.812,
+      "step": 1800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969433355702324e-05,
+      "loss": 0.5114,
+      "step": 1801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969399270730389e-05,
+      "loss": 1.2678,
+      "step": 1802
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9693651668623265e-05,
+      "loss": 0.6259,
+      "step": 1803
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9693310440983534e-05,
+      "loss": 2.0559,
+      "step": 1804
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969296902438687e-05,
+      "loss": 1.2513,
+      "step": 1805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969262741883546e-05,
+      "loss": 0.5999,
+      "step": 1806
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969228562433145e-05,
+      "loss": 0.4483,
+      "step": 1807
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969194364087704e-05,
+      "loss": 0.5862,
+      "step": 1808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9691601468474386e-05,
+      "loss": 1.0727,
+      "step": 1809
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969125910712569e-05,
+      "loss": 0.5698,
+      "step": 1810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969091655683311e-05,
+      "loss": 0.5367,
+      "step": 1811
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969057381759882e-05,
+      "loss": 1.3601,
+      "step": 1812
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969023088942503e-05,
+      "loss": 0.5868,
+      "step": 1813
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.968988777231389e-05,
+      "loss": 0.7297,
+      "step": 1814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.968954446626761e-05,
+      "loss": 0.5578,
+      "step": 1815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.968920097128836e-05,
+      "loss": 0.476,
+      "step": 1816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9688857287378336e-05,
+      "loss": 0.9506,
+      "step": 1817
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9688513414539705e-05,
+      "loss": 0.9572,
+      "step": 1818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.968816935277468e-05,
+      "loss": 0.398,
+      "step": 1819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.968782510208543e-05,
+      "loss": 0.2044,
+      "step": 1820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.968748066247416e-05,
+      "loss": 0.3861,
+      "step": 1821
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968713603394306e-05,
+      "loss": 0.3498,
+      "step": 1822
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968679121649432e-05,
+      "loss": 0.9798,
+      "step": 1823
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968644621013013e-05,
+      "loss": 1.0584,
+      "step": 1824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968610101485269e-05,
+      "loss": 1.4247,
+      "step": 1825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9685755630664196e-05,
+      "loss": 1.4017,
+      "step": 1826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968541005756684e-05,
+      "loss": 0.9637,
+      "step": 1827
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968506429556283e-05,
+      "loss": 0.4174,
+      "step": 1828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9684718344654376e-05,
+      "loss": 0.3928,
+      "step": 1829
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968437220484366e-05,
+      "loss": 1.15,
+      "step": 1830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968402587613289e-05,
+      "loss": 0.9854,
+      "step": 1831
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9683679358524274e-05,
+      "loss": 0.748,
+      "step": 1832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9683332652020016e-05,
+      "loss": 0.2382,
+      "step": 1833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9682985756622314e-05,
+      "loss": 1.0285,
+      "step": 1834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968263867233339e-05,
+      "loss": 0.7253,
+      "step": 1835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9682291399155445e-05,
+      "loss": 1.0218,
+      "step": 1836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968194393709068e-05,
+      "loss": 0.4681,
+      "step": 1837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968159628614132e-05,
+      "loss": 1.1544,
+      "step": 1838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968124844630958e-05,
+      "loss": 0.9235,
+      "step": 1839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968090041759766e-05,
+      "loss": 0.5093,
+      "step": 1840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9680552200007784e-05,
+      "loss": 0.9123,
+      "step": 1841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968020379354216e-05,
+      "loss": 0.6413,
+      "step": 1842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9679855198203014e-05,
+      "loss": 0.3583,
+      "step": 1843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967950641399256e-05,
+      "loss": 0.3314,
+      "step": 1844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967915744091302e-05,
+      "loss": 1.4601,
+      "step": 1845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9678808278966605e-05,
+      "loss": 0.6474,
+      "step": 1846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967845892815555e-05,
+      "loss": 0.9649,
+      "step": 1847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967810938848207e-05,
+      "loss": 0.6192,
+      "step": 1848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967775965994839e-05,
+      "loss": 1.2495,
+      "step": 1849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967740974255674e-05,
+      "loss": 0.2614,
+      "step": 1850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967705963630935e-05,
+      "loss": 0.5897,
+      "step": 1851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9676709341208434e-05,
+      "loss": 0.9869,
+      "step": 1852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967635885725624e-05,
+      "loss": 1.766,
+      "step": 1853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9676008184454975e-05,
+      "loss": 1.3281,
+      "step": 1854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967565732280689e-05,
+      "loss": 1.4757,
+      "step": 1855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9675306272314205e-05,
+      "loss": 1.2879,
+      "step": 1856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9674955032979164e-05,
+      "loss": 1.0235,
+      "step": 1857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9674603604804e-05,
+      "loss": 0.8651,
+      "step": 1858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967425198779095e-05,
+      "loss": 2.2621,
+      "step": 1859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967390018194224e-05,
+      "loss": 1.1933,
+      "step": 1860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967354818726012e-05,
+      "loss": 1.2548,
+      "step": 1861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967319600374684e-05,
+      "loss": 1.1124,
+      "step": 1862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967284363140462e-05,
+      "loss": 0.7509,
+      "step": 1863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967249107023571e-05,
+      "loss": 0.8679,
+      "step": 1864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967213832024235e-05,
+      "loss": 0.3401,
+      "step": 1865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.96717853814268e-05,
+      "loss": 0.3983,
+      "step": 1866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967143225379129e-05,
+      "loss": 0.4677,
+      "step": 1867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967107893733807e-05,
+      "loss": 1.1463,
+      "step": 1868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.96707254320694e-05,
+      "loss": 1.109,
+      "step": 1869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967037173798751e-05,
+      "loss": 1.2674,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9670017855094664e-05,
+      "loss": 0.7873,
+      "step": 1871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966966378339312e-05,
+      "loss": 1.0374,
+      "step": 1872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966930952288512e-05,
+      "loss": 1.5313,
+      "step": 1873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966895507357292e-05,
+      "loss": 0.4504,
+      "step": 1874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966860043545877e-05,
+      "loss": 1.2439,
+      "step": 1875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966824560854494e-05,
+      "loss": 1.1837,
+      "step": 1876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9667890592833675e-05,
+      "loss": 1.2827,
+      "step": 1877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966753538832725e-05,
+      "loss": 0.9065,
+      "step": 1878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9667179995027905e-05,
+      "loss": 0.6224,
+      "step": 1879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966682441293792e-05,
+      "loss": 0.3902,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9666468642059554e-05,
+      "loss": 0.5243,
+      "step": 1881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966611268239505e-05,
+      "loss": 0.6128,
+      "step": 1882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966575653394671e-05,
+      "loss": 0.6561,
+      "step": 1883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966540019671677e-05,
+      "loss": 1.3146,
+      "step": 1884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966504367070751e-05,
+      "loss": 0.4909,
+      "step": 1885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.96646869559212e-05,
+      "loss": 0.485,
+      "step": 1886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966433005236011e-05,
+      "loss": 0.6364,
+      "step": 1887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.96639729600265e-05,
+      "loss": 1.1639,
+      "step": 1888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9663615678922654e-05,
+      "loss": 0.8256,
+      "step": 1889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966325820905085e-05,
+      "loss": 1.5423,
+      "step": 1890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966290055041335e-05,
+      "loss": 0.9053,
+      "step": 1891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966254270301243e-05,
+      "loss": 0.3285,
+      "step": 1892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966218466685037e-05,
+      "loss": 1.1316,
+      "step": 1893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9661826441929455e-05,
+      "loss": 1.3753,
+      "step": 1894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9661468028251975e-05,
+      "loss": 0.5123,
+      "step": 1895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9661109425820175e-05,
+      "loss": 1.2879,
+      "step": 1896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9660750634636366e-05,
+      "loss": 0.8628,
+      "step": 1897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966039165470283e-05,
+      "loss": 1.0976,
+      "step": 1898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9660032486021836e-05,
+      "loss": 0.438,
+      "step": 1899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.965967312859568e-05,
+      "loss": 1.4286,
+      "step": 1900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.965931358242666e-05,
+      "loss": 0.4039,
+      "step": 1901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9658953847517034e-05,
+      "loss": 0.94,
+      "step": 1902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9658593923869114e-05,
+      "loss": 0.9372,
+      "step": 1903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9658233811485185e-05,
+      "loss": 0.5633,
+      "step": 1904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.965787351036754e-05,
+      "loss": 0.2773,
+      "step": 1905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.965751302051847e-05,
+      "loss": 0.8922,
+      "step": 1906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9657152341940266e-05,
+      "loss": 1.583,
+      "step": 1907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.965679147463524e-05,
+      "loss": 0.8569,
+      "step": 1908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9656430418605664e-05,
+      "loss": 0.6852,
+      "step": 1909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9656069173853846e-05,
+      "loss": 0.6526,
+      "step": 1910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.965570774038209e-05,
+      "loss": 0.7228,
+      "step": 1911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9655346118192684e-05,
+      "loss": 0.9298,
+      "step": 1912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.965498430728794e-05,
+      "loss": 1.0741,
+      "step": 1913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9654622307670166e-05,
+      "loss": 0.9284,
+      "step": 1914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.965426011934165e-05,
+      "loss": 1.0235,
+      "step": 1915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.96538977423047e-05,
+      "loss": 0.5426,
+      "step": 1916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9653535176561636e-05,
+      "loss": 0.2982,
+      "step": 1917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9653172422114746e-05,
+      "loss": 0.6485,
+      "step": 1918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.965280947896635e-05,
+      "loss": 0.7931,
+      "step": 1919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9652446347118755e-05,
+      "loss": 0.7334,
+      "step": 1920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.965208302657427e-05,
+      "loss": 1.4552,
+      "step": 1921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9651719517335214e-05,
+      "loss": 0.5567,
+      "step": 1922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.965135581940389e-05,
+      "loss": 1.0262,
+      "step": 1923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9650991932782616e-05,
+      "loss": 1.3212,
+      "step": 1924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9650627857473714e-05,
+      "loss": 1.0592,
+      "step": 1925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.965026359347949e-05,
+      "loss": 0.9559,
+      "step": 1926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964989914080227e-05,
+      "loss": 0.3213,
+      "step": 1927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964953449944438e-05,
+      "loss": 1.4297,
+      "step": 1928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9649169669408126e-05,
+      "loss": 0.2667,
+      "step": 1929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9648804650695834e-05,
+      "loss": 0.4263,
+      "step": 1930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964843944330982e-05,
+      "loss": 1.3483,
+      "step": 1931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964807404725243e-05,
+      "loss": 0.3982,
+      "step": 1932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964770846252597e-05,
+      "loss": 0.2619,
+      "step": 1933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9647342689132774e-05,
+      "loss": 1.4046,
+      "step": 1934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9646976727075167e-05,
+      "loss": 1.1999,
+      "step": 1935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964661057635548e-05,
+      "loss": 0.1572,
+      "step": 1936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964624423697603e-05,
+      "loss": 1.6078,
+      "step": 1937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9645877708939175e-05,
+      "loss": 1.2117,
+      "step": 1938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964551099224722e-05,
+      "loss": 0.7846,
+      "step": 1939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9645144086902514e-05,
+      "loss": 0.3061,
+      "step": 1940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964477699290739e-05,
+      "loss": 0.4388,
+      "step": 1941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964440971026419e-05,
+      "loss": 1.1041,
+      "step": 1942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964404223897524e-05,
+      "loss": 0.9268,
+      "step": 1943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964367457904287e-05,
+      "loss": 0.8486,
+      "step": 1944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9643306730469445e-05,
+      "loss": 0.9467,
+      "step": 1945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964293869325729e-05,
+      "loss": 0.2279,
+      "step": 1946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9642570467408754e-05,
+      "loss": 1.0068,
+      "step": 1947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9642202052926164e-05,
+      "loss": 1.4953,
+      "step": 1948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964183344981189e-05,
+      "loss": 1.4955,
+      "step": 1949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9641464658068255e-05,
+      "loss": 0.5641,
+      "step": 1950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9641095677697624e-05,
+      "loss": 0.3844,
+      "step": 1951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964072650870233e-05,
+      "loss": 1.1599,
+      "step": 1952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9640357151084726e-05,
+      "loss": 0.6373,
+      "step": 1953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963998760484717e-05,
+      "loss": 0.572,
+      "step": 1954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9639617869992e-05,
+      "loss": 1.1169,
+      "step": 1955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963924794652159e-05,
+      "loss": 1.1567,
+      "step": 1956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963887783443827e-05,
+      "loss": 0.4087,
+      "step": 1957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963850753374442e-05,
+      "loss": 0.4372,
+      "step": 1958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963813704444237e-05,
+      "loss": 1.3178,
+      "step": 1959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9637766366534494e-05,
+      "loss": 0.9917,
+      "step": 1960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963739550002315e-05,
+      "loss": 1.1084,
+      "step": 1961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.96370244449107e-05,
+      "loss": 1.0593,
+      "step": 1962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963665320119949e-05,
+      "loss": 1.485,
+      "step": 1963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.96362817688919e-05,
+      "loss": 0.5759,
+      "step": 1964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9635910147990286e-05,
+      "loss": 0.2528,
+      "step": 1965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963553833849702e-05,
+      "loss": 1.1895,
+      "step": 1966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963516634041445e-05,
+      "loss": 1.013,
+      "step": 1967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963479415374497e-05,
+      "loss": 0.8271,
+      "step": 1968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963442177849092e-05,
+      "loss": 1.1133,
+      "step": 1969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963404921465469e-05,
+      "loss": 1.2698,
+      "step": 1970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9633676462238645e-05,
+      "loss": 0.35,
+      "step": 1971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9633303521245155e-05,
+      "loss": 0.7272,
+      "step": 1972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963293039167659e-05,
+      "loss": 1.0201,
+      "step": 1973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963255707353533e-05,
+      "loss": 0.4115,
+      "step": 1974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963218356682376e-05,
+      "loss": 1.143,
+      "step": 1975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9631809871544236e-05,
+      "loss": 0.814,
+      "step": 1976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963143598769915e-05,
+      "loss": 0.4163,
+      "step": 1977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963106191529088e-05,
+      "loss": 0.5441,
+      "step": 1978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9630687654321794e-05,
+      "loss": 0.9346,
+      "step": 1979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963031320479429e-05,
+      "loss": 0.6711,
+      "step": 1980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962993856671075e-05,
+      "loss": 0.9891,
+      "step": 1981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9629563740073545e-05,
+      "loss": 0.5022,
+      "step": 1982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9629188724885066e-05,
+      "loss": 1.4512,
+      "step": 1983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962881352114772e-05,
+      "loss": 1.269,
+      "step": 1984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9628438128863856e-05,
+      "loss": 0.6845,
+      "step": 1985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962806254803589e-05,
+      "loss": 1.4855,
+      "step": 1986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.96276867786662e-05,
+      "loss": 0.6426,
+      "step": 1987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962731082075719e-05,
+      "loss": 1.1786,
+      "step": 1988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962693467431124e-05,
+      "loss": 0.6797,
+      "step": 1989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962655833933075e-05,
+      "loss": 0.6506,
+      "step": 1990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962618181581811e-05,
+      "loss": 0.6533,
+      "step": 1991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962580510377573e-05,
+      "loss": 1.2443,
+      "step": 1992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9625428203205986e-05,
+      "loss": 0.307,
+      "step": 1993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9625051114111295e-05,
+      "loss": 1.0541,
+      "step": 1994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962467383649404e-05,
+      "loss": 0.4458,
+      "step": 1995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962429637035663e-05,
+      "loss": 0.3133,
+      "step": 1996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962391871570148e-05,
+      "loss": 1.1313,
+      "step": 1997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9623540872530975e-05,
+      "loss": 1.0685,
+      "step": 1998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9623162840847524e-05,
+      "loss": 0.659,
+      "step": 1999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962278462065353e-05,
+      "loss": 1.1681,
+      "step": 2000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.96224062119514e-05,
+      "loss": 0.7629,
+      "step": 2001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962202761474355e-05,
+      "loss": 0.2156,
+      "step": 2002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962164882903239e-05,
+      "loss": 0.4361,
+      "step": 2003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9621269854820324e-05,
+      "loss": 0.4392,
+      "step": 2004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9620890692109764e-05,
+      "loss": 0.6316,
+      "step": 2005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962051134090313e-05,
+      "loss": 0.3045,
+      "step": 2006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962013180120282e-05,
+      "loss": 0.9845,
+      "step": 2007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.961975207301126e-05,
+      "loss": 0.755,
+      "step": 2008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.961937215633088e-05,
+      "loss": 0.218,
+      "step": 2009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.961899205116407e-05,
+      "loss": 0.9749,
+      "step": 2010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.961861175751326e-05,
+      "loss": 1.2353,
+      "step": 2011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9618231275380876e-05,
+      "loss": 1.1462,
+      "step": 2012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9617850604769334e-05,
+      "loss": 0.3959,
+      "step": 2013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.961746974568107e-05,
+      "loss": 1.19,
+      "step": 2014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9617088698118486e-05,
+      "loss": 1.2067,
+      "step": 2015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.961670746208401e-05,
+      "loss": 1.4503,
+      "step": 2016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9616326037580085e-05,
+      "loss": 0.6058,
+      "step": 2017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9615944424609126e-05,
+      "loss": 0.5747,
+      "step": 2018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961556262317356e-05,
+      "loss": 0.259,
+      "step": 2019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961518063327582e-05,
+      "loss": 0.9001,
+      "step": 2020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961479845491834e-05,
+      "loss": 1.545,
+      "step": 2021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961441608810355e-05,
+      "loss": 0.4146,
+      "step": 2022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9614033532833874e-05,
+      "loss": 0.3089,
+      "step": 2023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9613650789111755e-05,
+      "loss": 0.469,
+      "step": 2024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961326785693963e-05,
+      "loss": 0.1887,
+      "step": 2025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961288473631994e-05,
+      "loss": 0.691,
+      "step": 2026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961250142725511e-05,
+      "loss": 0.7155,
+      "step": 2027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961211792974759e-05,
+      "loss": 0.5071,
+      "step": 2028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961173424379981e-05,
+      "loss": 0.4681,
+      "step": 2029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961135036941422e-05,
+      "loss": 0.4226,
+      "step": 2030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9610966306593275e-05,
+      "loss": 0.8446,
+      "step": 2031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961058205533939e-05,
+      "loss": 1.4307,
+      "step": 2032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961019761565503e-05,
+      "loss": 0.9168,
+      "step": 2033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9609812987542634e-05,
+      "loss": 0.969,
+      "step": 2034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960942817100466e-05,
+      "loss": 1.3414,
+      "step": 2035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960904316604354e-05,
+      "loss": 1.0957,
+      "step": 2036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9608657972661736e-05,
+      "loss": 0.8218,
+      "step": 2037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.96082725908617e-05,
+      "loss": 0.3387,
+      "step": 2038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960788702064588e-05,
+      "loss": 0.9256,
+      "step": 2039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960750126201672e-05,
+      "loss": 0.392,
+      "step": 2040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.96071153149767e-05,
+      "loss": 0.3964,
+      "step": 2041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960672917952825e-05,
+      "loss": 0.7331,
+      "step": 2042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960634285567384e-05,
+      "loss": 0.9969,
+      "step": 2043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960595634341594e-05,
+      "loss": 1.0108,
+      "step": 2044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960556964275698e-05,
+      "loss": 0.3684,
+      "step": 2045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960518275369944e-05,
+      "loss": 1.0517,
+      "step": 2046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960479567624579e-05,
+      "loss": 0.3014,
+      "step": 2047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960440841039847e-05,
+      "loss": 0.4952,
+      "step": 2048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9604020956159965e-05,
+      "loss": 0.4108,
+      "step": 2049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9603633313532725e-05,
+      "loss": 0.6244,
+      "step": 2050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960324548251923e-05,
+      "loss": 0.5555,
+      "step": 2051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9602857463121946e-05,
+      "loss": 1.1432,
+      "step": 2052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960246925534333e-05,
+      "loss": 0.9252,
+      "step": 2053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960208085918587e-05,
+      "loss": 0.4488,
+      "step": 2054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9601692274652026e-05,
+      "loss": 0.8799,
+      "step": 2055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9601303501744265e-05,
+      "loss": 0.5533,
+      "step": 2056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9600914540465085e-05,
+      "loss": 0.4137,
+      "step": 2057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960052539081693e-05,
+      "loss": 0.7381,
+      "step": 2058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.96001360528023e-05,
+      "loss": 0.3563,
+      "step": 2059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959974652642366e-05,
+      "loss": 0.9662,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959935681168351e-05,
+      "loss": 1.5496,
+      "step": 2061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9598966908584294e-05,
+      "loss": 1.221,
+      "step": 2062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959857681712852e-05,
+      "loss": 0.2996,
+      "step": 2063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959818653731866e-05,
+      "loss": 0.3,
+      "step": 2064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.95977960691572e-05,
+      "loss": 2.2458,
+      "step": 2065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959740541264663e-05,
+      "loss": 0.9467,
+      "step": 2066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9597014567789424e-05,
+      "loss": 1.102,
+      "step": 2067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959662353458809e-05,
+      "loss": 0.5661,
+      "step": 2068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9596232313045086e-05,
+      "loss": 1.0639,
+      "step": 2069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9595840903162926e-05,
+      "loss": 0.4242,
+      "step": 2070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.95954493049441e-05,
+      "loss": 1.2749,
+      "step": 2071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959505751839109e-05,
+      "loss": 0.7759,
+      "step": 2072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959466554350639e-05,
+      "loss": 1.6358,
+      "step": 2073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.95942733802925e-05,
+      "loss": 1.2132,
+      "step": 2074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.95938810287519e-05,
+      "loss": 0.5293,
+      "step": 2075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9593488488887114e-05,
+      "loss": 1.5332,
+      "step": 2076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959309576070062e-05,
+      "loss": 0.5561,
+      "step": 2077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959270284419492e-05,
+      "loss": 0.4297,
+      "step": 2078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9592309739372525e-05,
+      "loss": 0.4421,
+      "step": 2079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959191644623592e-05,
+      "loss": 1.2804,
+      "step": 2080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9591522964787624e-05,
+      "loss": 0.3882,
+      "step": 2081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959112929503012e-05,
+      "loss": 0.8107,
+      "step": 2082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9590735436965935e-05,
+      "loss": 0.2377,
+      "step": 2083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9590341390597565e-05,
+      "loss": 1.4244,
+      "step": 2084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958994715592752e-05,
+      "loss": 0.3204,
+      "step": 2085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958955273295831e-05,
+      "loss": 0.2708,
+      "step": 2086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958915812169244e-05,
+      "loss": 0.3905,
+      "step": 2087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9588763322132423e-05,
+      "loss": 0.3961,
+      "step": 2088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958836833428078e-05,
+      "loss": 0.7883,
+      "step": 2089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9587973158140007e-05,
+      "loss": 0.7464,
+      "step": 2090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9587577793712625e-05,
+      "loss": 0.4365,
+      "step": 2091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9587182241001163e-05,
+      "loss": 1.231,
+      "step": 2092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958678650000813e-05,
+      "loss": 1.4578,
+      "step": 2093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958639057073604e-05,
+      "loss": 1.534,
+      "step": 2094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958599445318741e-05,
+      "loss": 1.3917,
+      "step": 2095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9585598147364775e-05,
+      "loss": 0.2313,
+      "step": 2096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958520165327064e-05,
+      "loss": 0.8112,
+      "step": 2097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958480497090755e-05,
+      "loss": 0.503,
+      "step": 2098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958440810027801e-05,
+      "loss": 0.688,
+      "step": 2099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958401104138455e-05,
+      "loss": 0.9918,
+      "step": 2100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958361379422969e-05,
+      "loss": 1.2182,
+      "step": 2101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958321635881597e-05,
+      "loss": 0.6189,
+      "step": 2102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958281873514592e-05,
+      "loss": 0.5386,
+      "step": 2103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9582420923222065e-05,
+      "loss": 0.7895,
+      "step": 2104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958202292304694e-05,
+      "loss": 0.3653,
+      "step": 2105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958162473462306e-05,
+      "loss": 0.6737,
+      "step": 2106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958122635795299e-05,
+      "loss": 1.5867,
+      "step": 2107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958082779303924e-05,
+      "loss": 0.4472,
+      "step": 2108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958042903988436e-05,
+      "loss": 2.1094,
+      "step": 2109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958003009849088e-05,
+      "loss": 0.1871,
+      "step": 2110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9579630968861334e-05,
+      "loss": 1.2405,
+      "step": 2111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9579231650998276e-05,
+      "loss": 0.8308,
+      "step": 2112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9578832144904244e-05,
+      "loss": 0.6184,
+      "step": 2113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.957843245058178e-05,
+      "loss": 1.3217,
+      "step": 2114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.957803256803341e-05,
+      "loss": 0.3041,
+      "step": 2115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.95776324972617e-05,
+      "loss": 0.7624,
+      "step": 2116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.957723223826919e-05,
+      "loss": 0.8077,
+      "step": 2117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957683179105843e-05,
+      "loss": 0.898,
+      "step": 2118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957643115563195e-05,
+      "loss": 1.7453,
+      "step": 2119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957603033199232e-05,
+      "loss": 0.5137,
+      "step": 2120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957562932014209e-05,
+      "loss": 0.3131,
+      "step": 2121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957522812008381e-05,
+      "loss": 0.2675,
+      "step": 2122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9574826731820016e-05,
+      "loss": 0.315,
+      "step": 2123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957442515535329e-05,
+      "loss": 1.3645,
+      "step": 2124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957402339068616e-05,
+      "loss": 0.8875,
+      "step": 2125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.95736214378212e-05,
+      "loss": 0.5362,
+      "step": 2126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957321929676097e-05,
+      "loss": 0.8996,
+      "step": 2127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957281696750801e-05,
+      "loss": 0.3836,
+      "step": 2128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9572414450064896e-05,
+      "loss": 0.3687,
+      "step": 2129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957201174443419e-05,
+      "loss": 1.0334,
+      "step": 2130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957160885061845e-05,
+      "loss": 1.0832,
+      "step": 2131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957120576862024e-05,
+      "loss": 0.1588,
+      "step": 2132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9570802498442136e-05,
+      "loss": 0.5114,
+      "step": 2133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957039904008667e-05,
+      "loss": 1.0364,
+      "step": 2134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9569995393556464e-05,
+      "loss": 0.6059,
+      "step": 2135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956959155885404e-05,
+      "loss": 0.9489,
+      "step": 2136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956918753598199e-05,
+      "loss": 0.9418,
+      "step": 2137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9568783324942863e-05,
+      "loss": 0.7637,
+      "step": 2138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956837892573926e-05,
+      "loss": 0.6017,
+      "step": 2139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9567974338373743e-05,
+      "loss": 0.9605,
+      "step": 2140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9567569562848884e-05,
+      "loss": 1.0633,
+      "step": 2141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956716459916726e-05,
+      "loss": 0.4589,
+      "step": 2142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956675944733145e-05,
+      "loss": 1.3311,
+      "step": 2143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956635410734403e-05,
+      "loss": 0.4576,
+      "step": 2144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9565948579207583e-05,
+      "loss": 1.1327,
+      "step": 2145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956554286292468e-05,
+      "loss": 1.196,
+      "step": 2146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956513695849791e-05,
+      "loss": 1.9873,
+      "step": 2147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9564730865929864e-05,
+      "loss": 1.2422,
+      "step": 2148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956432458522311e-05,
+      "loss": 0.9793,
+      "step": 2149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956391811638024e-05,
+      "loss": 0.5048,
+      "step": 2150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956351145940385e-05,
+      "loss": 1.1631,
+      "step": 2151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9563104614296506e-05,
+      "loss": 0.8936,
+      "step": 2152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9562697581060824e-05,
+      "loss": 0.3804,
+      "step": 2153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9562290359699375e-05,
+      "loss": 0.4076,
+      "step": 2154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956188295021476e-05,
+      "loss": 1.1775,
+      "step": 2155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9561475352609564e-05,
+      "loss": 1.2598,
+      "step": 2156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956106756688638e-05,
+      "loss": 0.3611,
+      "step": 2157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9560659593047816e-05,
+      "loss": 0.3868,
+      "step": 2158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956025143109646e-05,
+      "loss": 0.608,
+      "step": 2159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9559843081034906e-05,
+      "loss": 0.7251,
+      "step": 2160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955943454286576e-05,
+      "loss": 0.8895,
+      "step": 2161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955902581659161e-05,
+      "loss": 1.3419,
+      "step": 2162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955861690221507e-05,
+      "loss": 0.4093,
+      "step": 2163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9558207799738735e-05,
+      "loss": 1.1101,
+      "step": 2164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955779850916521e-05,
+      "loss": 0.35,
+      "step": 2165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.95573890304971e-05,
+      "loss": 0.7798,
+      "step": 2166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955697936373701e-05,
+      "loss": 0.3641,
+      "step": 2167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955656950888755e-05,
+      "loss": 1.0015,
+      "step": 2168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955615946595132e-05,
+      "loss": 1.0058,
+      "step": 2169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955574923493094e-05,
+      "loss": 1.3674,
+      "step": 2170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955533881582901e-05,
+      "loss": 0.8125,
+      "step": 2171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955492820864815e-05,
+      "loss": 0.9914,
+      "step": 2172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955451741339097e-05,
+      "loss": 0.8793,
+      "step": 2173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9554106430060084e-05,
+      "loss": 1.4865,
+      "step": 2174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955369525865811e-05,
+      "loss": 0.4402,
+      "step": 2175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955328389918766e-05,
+      "loss": 0.4447,
+      "step": 2176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955287235165135e-05,
+      "loss": 0.3639,
+      "step": 2177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955246061605181e-05,
+      "loss": 0.7062,
+      "step": 2178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955204869239165e-05,
+      "loss": 1.1076,
+      "step": 2179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9551636580673496e-05,
+      "loss": 1.1117,
+      "step": 2180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955122428089997e-05,
+      "loss": 0.347,
+      "step": 2181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955081179307369e-05,
+      "loss": 1.0347,
+      "step": 2182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955039911719729e-05,
+      "loss": 1.5676,
+      "step": 2183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954998625327339e-05,
+      "loss": 0.4467,
+      "step": 2184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954957320130461e-05,
+      "loss": 0.469,
+      "step": 2185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.95491599612936e-05,
+      "loss": 0.6529,
+      "step": 2186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954874653324297e-05,
+      "loss": 0.0643,
+      "step": 2187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9548332917155364e-05,
+      "loss": 0.8411,
+      "step": 2188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954791911303341e-05,
+      "loss": 0.7156,
+      "step": 2189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9547505120879725e-05,
+      "loss": 0.3546,
+      "step": 2190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9547090940696973e-05,
+      "loss": 1.4946,
+      "step": 2191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954667657248777e-05,
+      "loss": 0.6065,
+      "step": 2192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954626201625475e-05,
+      "loss": 0.7337,
+      "step": 2193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9545847272000565e-05,
+      "loss": 0.324,
+      "step": 2194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954543233972785e-05,
+      "loss": 0.9723,
+      "step": 2195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954501721943924e-05,
+      "loss": 0.8342,
+      "step": 2196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954460191113738e-05,
+      "loss": 1.2135,
+      "step": 2197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9544186414824906e-05,
+      "loss": 1.2936,
+      "step": 2198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954377073050448e-05,
+      "loss": 0.183,
+      "step": 2199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954335485817873e-05,
+      "loss": 1.1609,
+      "step": 2200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954293879785031e-05,
+      "loss": 0.9869,
+      "step": 2201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954252254952186e-05,
+      "loss": 0.9749,
+      "step": 2202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954210611319606e-05,
+      "loss": 1.0105,
+      "step": 2203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9541689488875504e-05,
+      "loss": 1.124,
+      "step": 2204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954127267656289e-05,
+      "loss": 0.9871,
+      "step": 2205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9540855676260854e-05,
+      "loss": 0.9169,
+      "step": 2206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954043848797205e-05,
+      "loss": 1.1402,
+      "step": 2207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954002111169913e-05,
+      "loss": 0.9834,
+      "step": 2208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.953960354744475e-05,
+      "loss": 0.2368,
+      "step": 2209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.953918579521158e-05,
+      "loss": 0.2871,
+      "step": 2210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.953876785500226e-05,
+      "loss": 1.3669,
+      "step": 2211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.953834972681946e-05,
+      "loss": 1.3795,
+      "step": 2212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.953793141066585e-05,
+      "loss": 0.9443,
+      "step": 2213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.953751290654407e-05,
+      "loss": 0.7215,
+      "step": 2214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9537094214456796e-05,
+      "loss": 0.8953,
+      "step": 2215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953667533440669e-05,
+      "loss": 0.3258,
+      "step": 2216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953625626639642e-05,
+      "loss": 0.7486,
+      "step": 2217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953583701042865e-05,
+      "loss": 0.3926,
+      "step": 2218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9535417566506056e-05,
+      "loss": 1.6139,
+      "step": 2219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9534997934631296e-05,
+      "loss": 0.2754,
+      "step": 2220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9534578114807046e-05,
+      "loss": 0.4352,
+      "step": 2221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953415810703598e-05,
+      "loss": 0.2472,
+      "step": 2222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953373791132076e-05,
+      "loss": 0.49,
+      "step": 2223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953331752766407e-05,
+      "loss": 1.022,
+      "step": 2224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953289695606858e-05,
+      "loss": 1.6402,
+      "step": 2225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953247619653697e-05,
+      "loss": 0.7377,
+      "step": 2226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953205524907192e-05,
+      "loss": 0.4341,
+      "step": 2227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9531634113676095e-05,
+      "loss": 1.3683,
+      "step": 2228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9531212790352196e-05,
+      "loss": 0.8495,
+      "step": 2229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953079127910289e-05,
+      "loss": 1.0899,
+      "step": 2230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953036957993086e-05,
+      "loss": 1.1959,
+      "step": 2231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952994769283879e-05,
+      "loss": 0.4037,
+      "step": 2232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952952561782938e-05,
+      "loss": 1.4337,
+      "step": 2233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952910335490528e-05,
+      "loss": 1.2475,
+      "step": 2234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9528680904069215e-05,
+      "loss": 1.426,
+      "step": 2235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952825826532385e-05,
+      "loss": 1.3955,
+      "step": 2236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952783543867189e-05,
+      "loss": 0.7031,
+      "step": 2237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952741242411602e-05,
+      "loss": 0.9248,
+      "step": 2238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952698922165892e-05,
+      "loss": 0.4671,
+      "step": 2239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9526565831303306e-05,
+      "loss": 0.9758,
+      "step": 2240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9526142253051846e-05,
+      "loss": 1.3609,
+      "step": 2241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9525718486907255e-05,
+      "loss": 0.5554,
+      "step": 2242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952529453287223e-05,
+      "loss": 0.4166,
+      "step": 2243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9524870390949454e-05,
+      "loss": 1.0726,
+      "step": 2244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952444606114164e-05,
+      "loss": 0.9166,
+      "step": 2245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952402154345148e-05,
+      "loss": 0.9977,
+      "step": 2246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952359683788168e-05,
+      "loss": 0.6456,
+      "step": 2247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952317194443494e-05,
+      "loss": 0.3907,
+      "step": 2248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952274686311397e-05,
+      "loss": 0.542,
+      "step": 2249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952232159392145e-05,
+      "loss": 1.4145,
+      "step": 2250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952189613686013e-05,
+      "loss": 0.7416,
+      "step": 2251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952147049193269e-05,
+      "loss": 1.193,
+      "step": 2252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952104465914183e-05,
+      "loss": 0.2797,
+      "step": 2253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9520618638490274e-05,
+      "loss": 0.8197,
+      "step": 2254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952019242998074e-05,
+      "loss": 0.4167,
+      "step": 2255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951976603361593e-05,
+      "loss": 0.9374,
+      "step": 2256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951933944939856e-05,
+      "loss": 1.0152,
+      "step": 2257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9518912677331336e-05,
+      "loss": 0.8476,
+      "step": 2258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9518485717416986e-05,
+      "loss": 0.7759,
+      "step": 2259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9518058569658226e-05,
+      "loss": 0.8008,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951763123405777e-05,
+      "loss": 0.2715,
+      "step": 2261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9517203710618327e-05,
+      "loss": 0.5598,
+      "step": 2262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951677599934264e-05,
+      "loss": 0.6027,
+      "step": 2263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9516348100233416e-05,
+      "loss": 0.4162,
+      "step": 2264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9515920013293374e-05,
+      "loss": 0.3498,
+      "step": 2265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9515491738525254e-05,
+      "loss": 1.4992,
+      "step": 2266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951506327593177e-05,
+      "loss": 0.8462,
+      "step": 2267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951463462551565e-05,
+      "loss": 0.865,
+      "step": 2268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951420578727963e-05,
+      "loss": 1.3235,
+      "step": 2269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951377676122643e-05,
+      "loss": 0.7111,
+      "step": 2270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951334754735878e-05,
+      "loss": 0.7908,
+      "step": 2271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951291814567941e-05,
+      "loss": 0.8306,
+      "step": 2272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9512488556191056e-05,
+      "loss": 0.3779,
+      "step": 2273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951205877889646e-05,
+      "loss": 0.9137,
+      "step": 2274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951162881379835e-05,
+      "loss": 0.7706,
+      "step": 2275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9511198660899444e-05,
+      "loss": 1.3054,
+      "step": 2276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9510768320202515e-05,
+      "loss": 0.6213,
+      "step": 2277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951033779171027e-05,
+      "loss": 0.7545,
+      "step": 2278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9509907075425475e-05,
+      "loss": 0.7413,
+      "step": 2279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9509476171350846e-05,
+      "loss": 0.5276,
+      "step": 2280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9509045079489135e-05,
+      "loss": 0.9164,
+      "step": 2281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.95086137998431e-05,
+      "loss": 0.6579,
+      "step": 2282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950818233241546e-05,
+      "loss": 0.6903,
+      "step": 2283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950775067720898e-05,
+      "loss": 0.8204,
+      "step": 2284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950731883422639e-05,
+      "loss": 0.9505,
+      "step": 2285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950688680347046e-05,
+      "loss": 1.2598,
+      "step": 2286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950645458494393e-05,
+      "loss": 0.5954,
+      "step": 2287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950602217864954e-05,
+      "loss": 0.8422,
+      "step": 2288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950558958459005e-05,
+      "loss": 1.491,
+      "step": 2289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950515680276821e-05,
+      "loss": 1.3455,
+      "step": 2290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950472383318678e-05,
+      "loss": 1.0867,
+      "step": 2291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9504290675848515e-05,
+      "loss": 1.3041,
+      "step": 2292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950385733075616e-05,
+      "loss": 0.6913,
+      "step": 2293
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950342379791249e-05,
+      "loss": 1.1731,
+      "step": 2294
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950299007732025e-05,
+      "loss": 0.984,
+      "step": 2295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950255616898221e-05,
+      "loss": 1.2216,
+      "step": 2296
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950212207290111e-05,
+      "loss": 1.2232,
+      "step": 2297
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950168778907974e-05,
+      "loss": 0.0649,
+      "step": 2298
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950125331752085e-05,
+      "loss": 0.5823,
+      "step": 2299
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950081865822721e-05,
+      "loss": 0.8986,
+      "step": 2300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950038381120159e-05,
+      "loss": 1.0603,
+      "step": 2301
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.949994877644673e-05,
+      "loss": 0.9665,
+      "step": 2302
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9499513553965433e-05,
+      "loss": 0.3427,
+      "step": 2303
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9499078143760445e-05,
+      "loss": 0.5818,
+      "step": 2304
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.949864254583455e-05,
+      "loss": 1.01,
+      "step": 2305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.949820676019052e-05,
+      "loss": 0.4451,
+      "step": 2306
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.949777078683112e-05,
+      "loss": 0.2554,
+      "step": 2307
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9497334625759125e-05,
+      "loss": 0.9999,
+      "step": 2308
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.949689827697731e-05,
+      "loss": 0.7491,
+      "step": 2309
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.949646174048847e-05,
+      "loss": 0.4519,
+      "step": 2310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.949602501629536e-05,
+      "loss": 1.385,
+      "step": 2311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.949558810440076e-05,
+      "loss": 1.1231,
+      "step": 2312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9495151004807466e-05,
+      "loss": 0.8794,
+      "step": 2313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.949471371751826e-05,
+      "loss": 0.6092,
+      "step": 2314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.94942762425359e-05,
+      "loss": 1.1576,
+      "step": 2315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.94938385798632e-05,
+      "loss": 0.4212,
+      "step": 2316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9493400729502927e-05,
+      "loss": 1.2769,
+      "step": 2317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.949296269145787e-05,
+      "loss": 1.2863,
+      "step": 2318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.949252446573082e-05,
+      "loss": 0.9786,
+      "step": 2319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.949208605232457e-05,
+      "loss": 0.9793,
+      "step": 2320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.949164745124188e-05,
+      "loss": 1.0596,
+      "step": 2321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.949120866248559e-05,
+      "loss": 1.2119,
+      "step": 2322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9490769686058454e-05,
+      "loss": 0.7602,
+      "step": 2323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.949033052196328e-05,
+      "loss": 0.5408,
+      "step": 2324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948989117020287e-05,
+      "loss": 0.3976,
+      "step": 2325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948945163078e-05,
+      "loss": 0.3997,
+      "step": 2326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948901190369749e-05,
+      "loss": 0.5914,
+      "step": 2327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9488571988958114e-05,
+      "loss": 0.7604,
+      "step": 2328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948813188656469e-05,
+      "loss": 0.2832,
+      "step": 2329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948769159652001e-05,
+      "loss": 0.3986,
+      "step": 2330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9487251118826886e-05,
+      "loss": 1.3259,
+      "step": 2331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9486810453488106e-05,
+      "loss": 0.56,
+      "step": 2332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9486369600506483e-05,
+      "loss": 0.5352,
+      "step": 2333
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948592855988482e-05,
+      "loss": 0.4722,
+      "step": 2334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9485487331625924e-05,
+      "loss": 0.2505,
+      "step": 2335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9485045915732605e-05,
+      "loss": 1.2138,
+      "step": 2336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948460431220767e-05,
+      "loss": 0.8894,
+      "step": 2337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9484162521053935e-05,
+      "loss": 1.1538,
+      "step": 2338
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.94837205422742e-05,
+      "loss": 1.9655,
+      "step": 2339
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9483278375871284e-05,
+      "loss": 0.4408,
+      "step": 2340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9482836021848006e-05,
+      "loss": 1.1423,
+      "step": 2341
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948239348020717e-05,
+      "loss": 0.0983,
+      "step": 2342
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.94819507509516e-05,
+      "loss": 0.5239,
+      "step": 2343
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9481507834084106e-05,
+      "loss": 0.4176,
+      "step": 2344
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948106472960751e-05,
+      "loss": 1.2511,
+      "step": 2345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948062143752464e-05,
+      "loss": 0.5899,
+      "step": 2346
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948017795783831e-05,
+      "loss": 0.5648,
+      "step": 2347
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947973429055135e-05,
+      "loss": 1.3003,
+      "step": 2348
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947929043566656e-05,
+      "loss": 0.5513,
+      "step": 2349
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947884639318679e-05,
+      "loss": 0.4099,
+      "step": 2350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947840216311486e-05,
+      "loss": 0.6228,
+      "step": 2351
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9477957745453576e-05,
+      "loss": 0.9686,
+      "step": 2352
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947751314020579e-05,
+      "loss": 0.43,
+      "step": 2353
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947706834737433e-05,
+      "loss": 0.7032,
+      "step": 2354
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947662336696202e-05,
+      "loss": 0.6318,
+      "step": 2355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947617819897168e-05,
+      "loss": 1.0296,
+      "step": 2356
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9475732843406174e-05,
+      "loss": 0.9949,
+      "step": 2357
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947528730026831e-05,
+      "loss": 0.3617,
+      "step": 2358
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9474841569560926e-05,
+      "loss": 1.3906,
+      "step": 2359
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947439565128686e-05,
+      "loss": 1.2923,
+      "step": 2360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947394954544896e-05,
+      "loss": 0.9679,
+      "step": 2361
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9473503252050056e-05,
+      "loss": 1.1516,
+      "step": 2362
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947305677109299e-05,
+      "loss": 0.5929,
+      "step": 2363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947261010258061e-05,
+      "loss": 0.644,
+      "step": 2364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947216324651574e-05,
+      "loss": 0.5442,
+      "step": 2365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947171620290124e-05,
+      "loss": 0.3594,
+      "step": 2366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947126897173995e-05,
+      "loss": 0.4067,
+      "step": 2367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947082155303471e-05,
+      "loss": 0.4321,
+      "step": 2368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9470373946788375e-05,
+      "loss": 0.2963,
+      "step": 2369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946992615300379e-05,
+      "loss": 0.9423,
+      "step": 2370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946947817168381e-05,
+      "loss": 0.4505,
+      "step": 2371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9469030002831284e-05,
+      "loss": 0.4056,
+      "step": 2372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946858164644906e-05,
+      "loss": 1.2519,
+      "step": 2373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946813310253999e-05,
+      "loss": 1.0287,
+      "step": 2374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946768437110693e-05,
+      "loss": 0.7399,
+      "step": 2375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946723545215274e-05,
+      "loss": 1.2247,
+      "step": 2376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946678634568027e-05,
+      "loss": 0.5443,
+      "step": 2377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946633705169239e-05,
+      "loss": 0.4095,
+      "step": 2378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9465887570191945e-05,
+      "loss": 0.9901,
+      "step": 2379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9465437901181795e-05,
+      "loss": 0.9318,
+      "step": 2380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9464988044664814e-05,
+      "loss": 0.2323,
+      "step": 2381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946453800064386e-05,
+      "loss": 0.8907,
+      "step": 2382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946408776912179e-05,
+      "loss": 0.8393,
+      "step": 2383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9463637350101484e-05,
+      "loss": 0.3813,
+      "step": 2384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946318674358578e-05,
+      "loss": 0.9322,
+      "step": 2385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946273594957758e-05,
+      "loss": 0.4559,
+      "step": 2386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9462284968079727e-05,
+      "loss": 0.3484,
+      "step": 2387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946183379909511e-05,
+      "loss": 0.3934,
+      "step": 2388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9461382442626584e-05,
+      "loss": 0.7687,
+      "step": 2389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946093089867703e-05,
+      "loss": 1.3251,
+      "step": 2390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946047916724932e-05,
+      "loss": 1.1644,
+      "step": 2391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9460027248346323e-05,
+      "loss": 0.6371,
+      "step": 2392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.945957514197092e-05,
+      "loss": 0.377,
+      "step": 2393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9459122848125994e-05,
+      "loss": 0.9156,
+      "step": 2394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.945867036681441e-05,
+      "loss": 0.5349,
+      "step": 2395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9458217698039065e-05,
+      "loss": 0.6317,
+      "step": 2396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9457764841802824e-05,
+      "loss": 0.7928,
+      "step": 2397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9457311798108576e-05,
+      "loss": 1.4559,
+      "step": 2398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.945685856695919e-05,
+      "loss": 0.9948,
+      "step": 2399
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.945640514835757e-05,
+      "loss": 0.2423,
+      "step": 2400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.94559515423066e-05,
+      "loss": 1.5212,
+      "step": 2401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.945549774880915e-05,
+      "loss": 0.771,
+      "step": 2402
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.945504376786812e-05,
+      "loss": 0.5958,
+      "step": 2403
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9454589599486404e-05,
+      "loss": 0.476,
+      "step": 2404
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9454135243666876e-05,
+      "loss": 0.585,
+      "step": 2405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.945368070041244e-05,
+      "loss": 2.3113,
+      "step": 2406
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9453225969725985e-05,
+      "loss": 2.065,
+      "step": 2407
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9452771051610404e-05,
+      "loss": 0.6814,
+      "step": 2408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.945231594606859e-05,
+      "loss": 1.164,
+      "step": 2409
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.945186065310344e-05,
+      "loss": 0.5236,
+      "step": 2410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9451405172717856e-05,
+      "loss": 0.6364,
+      "step": 2411
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.945094950491474e-05,
+      "loss": 0.8572,
+      "step": 2412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9450493649696975e-05,
+      "loss": 0.8268,
+      "step": 2413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.945003760706747e-05,
+      "loss": 1.0304,
+      "step": 2414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944958137702914e-05,
+      "loss": 0.5459,
+      "step": 2415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944912495958487e-05,
+      "loss": 0.5306,
+      "step": 2416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944866835473758e-05,
+      "loss": 0.9092,
+      "step": 2417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944821156249016e-05,
+      "loss": 0.4681,
+      "step": 2418
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9447754582845524e-05,
+      "loss": 0.5539,
+      "step": 2419
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9447297415806584e-05,
+      "loss": 0.3933,
+      "step": 2420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944684006137624e-05,
+      "loss": 1.0416,
+      "step": 2421
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944638251955741e-05,
+      "loss": 0.2931,
+      "step": 2422
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9445924790353004e-05,
+      "loss": 1.4223,
+      "step": 2423
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944546687376594e-05,
+      "loss": 0.4252,
+      "step": 2424
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944500876979912e-05,
+      "loss": 0.9621,
+      "step": 2425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944455047845547e-05,
+      "loss": 0.2691,
+      "step": 2426
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9444091999737895e-05,
+      "loss": 0.7919,
+      "step": 2427
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9443633333649325e-05,
+      "loss": 1.3725,
+      "step": 2428
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944317448019268e-05,
+      "loss": 1.4259,
+      "step": 2429
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9442715439370864e-05,
+      "loss": 0.7938,
+      "step": 2430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944225621118681e-05,
+      "loss": 0.9755,
+      "step": 2431
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944179679564344e-05,
+      "loss": 1.4638,
+      "step": 2432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9441337192743666e-05,
+      "loss": 1.3943,
+      "step": 2433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944087740249043e-05,
+      "loss": 1.7849,
+      "step": 2434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9440417424886655e-05,
+      "loss": 0.8695,
+      "step": 2435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943995725993526e-05,
+      "loss": 0.4596,
+      "step": 2436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9439496907639166e-05,
+      "loss": 0.7793,
+      "step": 2437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943903636800132e-05,
+      "loss": 1.1398,
+      "step": 2438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943857564102464e-05,
+      "loss": 1.1037,
+      "step": 2439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943811472671208e-05,
+      "loss": 0.2739,
+      "step": 2440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943765362506654e-05,
+      "loss": 1.152,
+      "step": 2441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943719233609098e-05,
+      "loss": 1.4002,
+      "step": 2442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943673085978832e-05,
+      "loss": 0.3812,
+      "step": 2443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943626919616152e-05,
+      "loss": 0.8041,
+      "step": 2444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9435807345213485e-05,
+      "loss": 0.6315,
+      "step": 2445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9435345306947176e-05,
+      "loss": 0.3944,
+      "step": 2446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943488308136552e-05,
+      "loss": 1.0586,
+      "step": 2447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943442066847147e-05,
+      "loss": 0.7982,
+      "step": 2448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943395806826796e-05,
+      "loss": 0.3336,
+      "step": 2449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943349528075795e-05,
+      "loss": 1.1432,
+      "step": 2450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943303230594436e-05,
+      "loss": 1.0904,
+      "step": 2451
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9432569143830165e-05,
+      "loss": 1.3246,
+      "step": 2452
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9432105794418286e-05,
+      "loss": 0.3778,
+      "step": 2453
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943164225771168e-05,
+      "loss": 1.0095,
+      "step": 2454
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9431178533713306e-05,
+      "loss": 0.4772,
+      "step": 2455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.94307146224261e-05,
+      "loss": 0.3371,
+      "step": 2456
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9430250523853024e-05,
+      "loss": 0.8545,
+      "step": 2457
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9429786237997034e-05,
+      "loss": 0.7143,
+      "step": 2458
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942932176486108e-05,
+      "loss": 1.2538,
+      "step": 2459
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942885710444812e-05,
+      "loss": 0.4512,
+      "step": 2460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.94283922567611e-05,
+      "loss": 1.2608,
+      "step": 2461
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9427927221803e-05,
+      "loss": 0.5212,
+      "step": 2462
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9427461999576744e-05,
+      "loss": 1.7004,
+      "step": 2463
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9426996590085334e-05,
+      "loss": 0.3162,
+      "step": 2464
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9426530993331705e-05,
+      "loss": 1.3919,
+      "step": 2465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942606520931883e-05,
+      "loss": 0.5451,
+      "step": 2466
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9425599238049665e-05,
+      "loss": 0.7366,
+      "step": 2467
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942513307952719e-05,
+      "loss": 0.3898,
+      "step": 2468
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942466673375435e-05,
+      "loss": 0.4571,
+      "step": 2469
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9424200200734136e-05,
+      "loss": 0.6988,
+      "step": 2470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.94237334804695e-05,
+      "loss": 1.0383,
+      "step": 2471
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9423266572963414e-05,
+      "loss": 0.9577,
+      "step": 2472
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942279947821886e-05,
+      "loss": 1.1399,
+      "step": 2473
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9422332196238794e-05,
+      "loss": 1.4027,
+      "step": 2474
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942186472702621e-05,
+      "loss": 1.0917,
+      "step": 2475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942139707058407e-05,
+      "loss": 0.7248,
+      "step": 2476
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942092922691535e-05,
+      "loss": 0.5327,
+      "step": 2477
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942046119602302e-05,
+      "loss": 0.2194,
+      "step": 2478
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9419992977910076e-05,
+      "loss": 0.6992,
+      "step": 2479
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.941952457257948e-05,
+      "loss": 1.1535,
+      "step": 2480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.941905598003423e-05,
+      "loss": 0.6142,
+      "step": 2481
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.941858720027729e-05,
+      "loss": 1.3552,
+      "step": 2482
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.941811823331166e-05,
+      "loss": 0.4111,
+      "step": 2483
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.941764907914032e-05,
+      "loss": 0.7859,
+      "step": 2484
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.941717973776625e-05,
+      "loss": 0.8723,
+      "step": 2485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9416710209192436e-05,
+      "loss": 0.459,
+      "step": 2486
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.941624049342187e-05,
+      "loss": 0.4722,
+      "step": 2487
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9415770590457546e-05,
+      "loss": 0.5883,
+      "step": 2488
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9415300500302436e-05,
+      "loss": 1.0449,
+      "step": 2489
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.941483022295955e-05,
+      "loss": 0.379,
+      "step": 2490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.941435975843187e-05,
+      "loss": 0.4455,
+      "step": 2491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.94138891067224e-05,
+      "loss": 1.1085,
+      "step": 2492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9413418267834134e-05,
+      "loss": 0.8996,
+      "step": 2493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9412947241770054e-05,
+      "loss": 0.2513,
+      "step": 2494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9412476028533174e-05,
+      "loss": 0.2914,
+      "step": 2495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.941200462812648e-05,
+      "loss": 1.319,
+      "step": 2496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9411533040552985e-05,
+      "loss": 0.6258,
+      "step": 2497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9411061265815674e-05,
+      "loss": 0.632,
+      "step": 2498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9410589303917556e-05,
+      "loss": 1.1774,
+      "step": 2499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.941011715486165e-05,
+      "loss": 1.4072,
+      "step": 2500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9409644818650935e-05,
+      "loss": 0.9624,
+      "step": 2501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.940917229528843e-05,
+      "loss": 1.5124,
+      "step": 2502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.940869958477714e-05,
+      "loss": 1.4012,
+      "step": 2503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9408226687120075e-05,
+      "loss": 0.3687,
+      "step": 2504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.940775360232024e-05,
+      "loss": 0.4577,
+      "step": 2505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.940728033038065e-05,
+      "loss": 1.2154,
+      "step": 2506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9406806871304315e-05,
+      "loss": 0.9035,
+      "step": 2507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.940633322509425e-05,
+      "loss": 0.6519,
+      "step": 2508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9405859391753464e-05,
+      "loss": 0.6558,
+      "step": 2509
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9405385371284984e-05,
+      "loss": 0.3812,
+      "step": 2510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.940491116369181e-05,
+      "loss": 0.3227,
+      "step": 2511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940443676897697e-05,
+      "loss": 0.4254,
+      "step": 2512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940396218714347e-05,
+      "loss": 0.765,
+      "step": 2513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940348741819436e-05,
+      "loss": 1.15,
+      "step": 2514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9403012462132624e-05,
+      "loss": 0.9,
+      "step": 2515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940253731896131e-05,
+      "loss": 0.508,
+      "step": 2516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940206198868344e-05,
+      "loss": 0.6364,
+      "step": 2517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9401586471302026e-05,
+      "loss": 1.0135,
+      "step": 2518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9401110766820096e-05,
+      "loss": 0.4942,
+      "step": 2519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9400634875240684e-05,
+      "loss": 0.4724,
+      "step": 2520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9400158796566825e-05,
+      "loss": 1.7387,
+      "step": 2521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9399682530801535e-05,
+      "loss": 0.8669,
+      "step": 2522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939920607794785e-05,
+      "loss": 0.5612,
+      "step": 2523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.93987294380088e-05,
+      "loss": 0.7745,
+      "step": 2524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9398252610987415e-05,
+      "loss": 0.5662,
+      "step": 2525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939777559688674e-05,
+      "loss": 1.0908,
+      "step": 2526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.93972983957098e-05,
+      "loss": 0.8797,
+      "step": 2527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939682100745965e-05,
+      "loss": 0.6313,
+      "step": 2528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939634343213929e-05,
+      "loss": 1.2741,
+      "step": 2529
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.93958656697518e-05,
+      "loss": 0.7943,
+      "step": 2530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9395387720300205e-05,
+      "loss": 1.0341,
+      "step": 2531
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9394909583787536e-05,
+      "loss": 0.3303,
+      "step": 2532
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939443126021685e-05,
+      "loss": 0.3907,
+      "step": 2533
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939395274959119e-05,
+      "loss": 0.3209,
+      "step": 2534
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939347405191359e-05,
+      "loss": 1.2374,
+      "step": 2535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9392995167187104e-05,
+      "loss": 1.4128,
+      "step": 2536
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939251609541478e-05,
+      "loss": 1.1299,
+      "step": 2537
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9392036836599665e-05,
+      "loss": 0.6629,
+      "step": 2538
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939155739074481e-05,
+      "loss": 0.7476,
+      "step": 2539
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939107775785326e-05,
+      "loss": 0.78,
+      "step": 2540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939059793792808e-05,
+      "loss": 0.2316,
+      "step": 2541
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939011793097231e-05,
+      "loss": 1.1474,
+      "step": 2542
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9389637736989015e-05,
+      "loss": 0.5111,
+      "step": 2543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9389157355981235e-05,
+      "loss": 1.0598,
+      "step": 2544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9388676787952044e-05,
+      "loss": 1.5312,
+      "step": 2545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9388196032904497e-05,
+      "loss": 0.5879,
+      "step": 2546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938771509084164e-05,
+      "loss": 0.8647,
+      "step": 2547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938723396176656e-05,
+      "loss": 0.9654,
+      "step": 2548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9386752645682286e-05,
+      "loss": 1.0643,
+      "step": 2549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9386271142591917e-05,
+      "loss": 0.4193,
+      "step": 2550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938578945249848e-05,
+      "loss": 0.3744,
+      "step": 2551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938530757540506e-05,
+      "loss": 0.688,
+      "step": 2552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938482551131473e-05,
+      "loss": 0.5685,
+      "step": 2553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938434326023054e-05,
+      "loss": 0.8555,
+      "step": 2554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938386082215558e-05,
+      "loss": 1.0364,
+      "step": 2555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938337819709289e-05,
+      "loss": 1.0913,
+      "step": 2556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9382895385045575e-05,
+      "loss": 0.5181,
+      "step": 2557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938241238601669e-05,
+      "loss": 0.6151,
+      "step": 2558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938192920000931e-05,
+      "loss": 1.0,
+      "step": 2559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.93814458270265e-05,
+      "loss": 0.7786,
+      "step": 2560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938096226707135e-05,
+      "loss": 0.6949,
+      "step": 2561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938047852014694e-05,
+      "loss": 0.8818,
+      "step": 2562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937999458625634e-05,
+      "loss": 0.5558,
+      "step": 2563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937951046540262e-05,
+      "loss": 0.836,
+      "step": 2564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937902615758888e-05,
+      "loss": 0.7766,
+      "step": 2565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9378541662818195e-05,
+      "loss": 0.7394,
+      "step": 2566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9378056981093644e-05,
+      "loss": 0.3822,
+      "step": 2567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937757211241832e-05,
+      "loss": 0.2547,
+      "step": 2568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9377087056795296e-05,
+      "loss": 1.2686,
+      "step": 2569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937660181422767e-05,
+      "loss": 0.5114,
+      "step": 2570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937611638471852e-05,
+      "loss": 1.1328,
+      "step": 2571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937563076827095e-05,
+      "loss": 0.6315,
+      "step": 2572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9375144964888035e-05,
+      "loss": 0.8797,
+      "step": 2573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937465897457287e-05,
+      "loss": 0.1253,
+      "step": 2574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9374172797328554e-05,
+      "loss": 1.3569,
+      "step": 2575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937368643315818e-05,
+      "loss": 0.5742,
+      "step": 2576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937319988206483e-05,
+      "loss": 0.5999,
+      "step": 2577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937271314405162e-05,
+      "loss": 1.2341,
+      "step": 2578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9372226219121635e-05,
+      "loss": 0.5089,
+      "step": 2579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937173910727797e-05,
+      "loss": 0.1828,
+      "step": 2580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937125180852374e-05,
+      "loss": 1.1294,
+      "step": 2581
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937076432286203e-05,
+      "loss": 0.99,
+      "step": 2582
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937027665029595e-05,
+      "loss": 1.1303,
+      "step": 2583
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.93697887908286e-05,
+      "loss": 1.1076,
+      "step": 2584
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.936930074446309e-05,
+      "loss": 1.3479,
+      "step": 2585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.936881251120252e-05,
+      "loss": 0.7112,
+      "step": 2586
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.936832409105001e-05,
+      "loss": 0.635,
+      "step": 2587
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.936783548400865e-05,
+      "loss": 0.5022,
+      "step": 2588
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.936734669008156e-05,
+      "loss": 0.6093,
+      "step": 2589
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9366857709271846e-05,
+      "loss": 1.2218,
+      "step": 2590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9366368541582614e-05,
+      "loss": 0.6383,
+      "step": 2591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9365879187017e-05,
+      "loss": 1.3004,
+      "step": 2592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.936538964557809e-05,
+      "loss": 0.5613,
+      "step": 2593
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9364899917269016e-05,
+      "loss": 1.2688,
+      "step": 2594
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.936441000209288e-05,
+      "loss": 0.871,
+      "step": 2595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.936391990005282e-05,
+      "loss": 0.6956,
+      "step": 2596
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9363429611151945e-05,
+      "loss": 0.8977,
+      "step": 2597
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.936293913539337e-05,
+      "loss": 0.8609,
+      "step": 2598
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9362448472780226e-05,
+      "loss": 0.8019,
+      "step": 2599
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.936195762331562e-05,
+      "loss": 1.2905,
+      "step": 2600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.93614665870027e-05,
+      "loss": 0.8435,
+      "step": 2601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.936097536384456e-05,
+      "loss": 1.3059,
+      "step": 2602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.936048395384435e-05,
+      "loss": 1.6494,
+      "step": 2603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9359992357005196e-05,
+      "loss": 0.8236,
+      "step": 2604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9359500573330215e-05,
+      "loss": 0.4517,
+      "step": 2605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9359008602822534e-05,
+      "loss": 0.5884,
+      "step": 2606
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9358516445485305e-05,
+      "loss": 0.7392,
+      "step": 2607
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.935802410132163e-05,
+      "loss": 0.4826,
+      "step": 2608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9357531570334667e-05,
+      "loss": 1.5875,
+      "step": 2609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935703885252754e-05,
+      "loss": 0.3892,
+      "step": 2610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935654594790339e-05,
+      "loss": 0.8682,
+      "step": 2611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9356052856465345e-05,
+      "loss": 0.6221,
+      "step": 2612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9355559578216544e-05,
+      "loss": 0.4781,
+      "step": 2613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935506611316013e-05,
+      "loss": 1.0887,
+      "step": 2614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935457246129925e-05,
+      "loss": 0.6299,
+      "step": 2615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9354078622637034e-05,
+      "loss": 0.831,
+      "step": 2616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935358459717663e-05,
+      "loss": 0.8885,
+      "step": 2617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935309038492117e-05,
+      "loss": 0.9677,
+      "step": 2618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935259598587382e-05,
+      "loss": 1.1161,
+      "step": 2619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.93521014000377e-05,
+      "loss": 0.3266,
+      "step": 2620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935160662741599e-05,
+      "loss": 1.3458,
+      "step": 2621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935111166801182e-05,
+      "loss": 1.2525,
+      "step": 2622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935061652182834e-05,
+      "loss": 0.4709,
+      "step": 2623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935012118886869e-05,
+      "loss": 0.9347,
+      "step": 2624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9349625669136045e-05,
+      "loss": 1.0765,
+      "step": 2625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934912996263355e-05,
+      "loss": 0.4198,
+      "step": 2626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934863406936435e-05,
+      "loss": 0.3387,
+      "step": 2627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934813798933162e-05,
+      "loss": 0.8128,
+      "step": 2628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.93476417225385e-05,
+      "loss": 0.386,
+      "step": 2629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9347145268988136e-05,
+      "loss": 0.3452,
+      "step": 2630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934664862868372e-05,
+      "loss": 0.6041,
+      "step": 2631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9346151801628394e-05,
+      "loss": 0.2839,
+      "step": 2632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9345654787825315e-05,
+      "loss": 1.3644,
+      "step": 2633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9345157587277654e-05,
+      "loss": 0.7991,
+      "step": 2634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934466019998858e-05,
+      "loss": 1.184,
+      "step": 2635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9344162625961254e-05,
+      "loss": 1.3064,
+      "step": 2636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934366486519883e-05,
+      "loss": 1.1136,
+      "step": 2637
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934316691770448e-05,
+      "loss": 0.7883,
+      "step": 2638
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9342668783481396e-05,
+      "loss": 0.6127,
+      "step": 2639
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934217046253272e-05,
+      "loss": 0.4445,
+      "step": 2640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934167195486164e-05,
+      "loss": 1.2362,
+      "step": 2641
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934117326047132e-05,
+      "loss": 0.8911,
+      "step": 2642
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934067437936493e-05,
+      "loss": 1.1379,
+      "step": 2643
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934017531154565e-05,
+      "loss": 0.3782,
+      "step": 2644
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9339676057016666e-05,
+      "loss": 0.9548,
+      "step": 2645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933917661578113e-05,
+      "loss": 1.3717,
+      "step": 2646
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933867698784224e-05,
+      "loss": 0.4404,
+      "step": 2647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9338177173203185e-05,
+      "loss": 0.254,
+      "step": 2648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933767717186712e-05,
+      "loss": 0.4807,
+      "step": 2649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933717698383724e-05,
+      "loss": 0.3947,
+      "step": 2650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9336676609116725e-05,
+      "loss": 0.345,
+      "step": 2651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933617604770876e-05,
+      "loss": 1.328,
+      "step": 2652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933567529961653e-05,
+      "loss": 1.644,
+      "step": 2653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9335174364843216e-05,
+      "loss": 0.5056,
+      "step": 2654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933467324339203e-05,
+      "loss": 1.2747,
+      "step": 2655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9334171935266126e-05,
+      "loss": 1.108,
+      "step": 2656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9333670440468715e-05,
+      "loss": 0.5009,
+      "step": 2657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9333168759002984e-05,
+      "loss": 0.9502,
+      "step": 2658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9332666890872136e-05,
+      "loss": 1.2955,
+      "step": 2659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933216483607934e-05,
+      "loss": 0.5542,
+      "step": 2660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933166259462782e-05,
+      "loss": 0.903,
+      "step": 2661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933116016652075e-05,
+      "loss": 0.2534,
+      "step": 2662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933065755176133e-05,
+      "loss": 0.6614,
+      "step": 2663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933015475035277e-05,
+      "loss": 0.7016,
+      "step": 2664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9329651762298266e-05,
+      "loss": 1.5718,
+      "step": 2665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9329148587601005e-05,
+      "loss": 1.3658,
+      "step": 2666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932864522626421e-05,
+      "loss": 0.5261,
+      "step": 2667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9328141678291066e-05,
+      "loss": 0.7204,
+      "step": 2668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9327637943684786e-05,
+      "loss": 1.6531,
+      "step": 2669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932713402244858e-05,
+      "loss": 1.0488,
+      "step": 2670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932662991458565e-05,
+      "loss": 0.7499,
+      "step": 2671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9326125620099196e-05,
+      "loss": 0.2094,
+      "step": 2672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9325621138992446e-05,
+      "loss": 0.5847,
+      "step": 2673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932511647126859e-05,
+      "loss": 0.9503,
+      "step": 2674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932461161693086e-05,
+      "loss": 0.3795,
+      "step": 2675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9324106575982445e-05,
+      "loss": 1.3708,
+      "step": 2676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9323601348426574e-05,
+      "loss": 0.4262,
+      "step": 2677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932309593426646e-05,
+      "loss": 0.6167,
+      "step": 2678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932259033350532e-05,
+      "loss": 0.9506,
+      "step": 2679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932208454614637e-05,
+      "loss": 0.8584,
+      "step": 2680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9321578572192824e-05,
+      "loss": 0.3826,
+      "step": 2681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9321072411647916e-05,
+      "loss": 0.6214,
+      "step": 2682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932056606451484e-05,
+      "loss": 0.5605,
+      "step": 2683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932005953079686e-05,
+      "loss": 0.4148,
+      "step": 2684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.931955281049715e-05,
+      "loss": 0.8016,
+      "step": 2685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.931904590361898e-05,
+      "loss": 0.9365,
+      "step": 2686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.931853881016555e-05,
+      "loss": 0.9676,
+      "step": 2687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.931803153014009e-05,
+      "loss": 0.7064,
+      "step": 2688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.931752406354583e-05,
+      "loss": 0.5693,
+      "step": 2689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9317016410386e-05,
+      "loss": 0.3687,
+      "step": 2690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.931650857066383e-05,
+      "loss": 0.8766,
+      "step": 2691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.931600054438255e-05,
+      "loss": 0.373,
+      "step": 2692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.93154923315454e-05,
+      "loss": 0.5874,
+      "step": 2693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.93149839321556e-05,
+      "loss": 0.7334,
+      "step": 2694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9314475346216404e-05,
+      "loss": 0.6727,
+      "step": 2695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.931396657373104e-05,
+      "loss": 0.6347,
+      "step": 2696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9313457614702736e-05,
+      "loss": 0.5914,
+      "step": 2697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9312948469134745e-05,
+      "loss": 0.9814,
+      "step": 2698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.93124391370303e-05,
+      "loss": 1.0294,
+      "step": 2699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.931192961839264e-05,
+      "loss": 0.7637,
+      "step": 2700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.931141991322501e-05,
+      "loss": 0.7036,
+      "step": 2701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9310910021530665e-05,
+      "loss": 0.4761,
+      "step": 2702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.931039994331282e-05,
+      "loss": 1.2104,
+      "step": 2703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.930988967857476e-05,
+      "loss": 0.2889,
+      "step": 2704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.93093792273197e-05,
+      "loss": 0.8682,
+      "step": 2705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9308868589550906e-05,
+      "loss": 1.0814,
+      "step": 2706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.930835776527162e-05,
+      "loss": 0.7622,
+      "step": 2707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.930784675448509e-05,
+      "loss": 1.4267,
+      "step": 2708
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930733555719458e-05,
+      "loss": 0.795,
+      "step": 2709
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9306824173403337e-05,
+      "loss": 0.3121,
+      "step": 2710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.93063126031146e-05,
+      "loss": 1.3382,
+      "step": 2711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9305800846331655e-05,
+      "loss": 0.4616,
+      "step": 2712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9305288903057736e-05,
+      "loss": 1.2397,
+      "step": 2713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930477677329611e-05,
+      "loss": 1.1048,
+      "step": 2714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930426445705002e-05,
+      "loss": 0.5483,
+      "step": 2715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930375195432275e-05,
+      "loss": 0.8359,
+      "step": 2716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930323926511755e-05,
+      "loss": 0.3722,
+      "step": 2717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930272638943768e-05,
+      "loss": 0.8784,
+      "step": 2718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930221332728641e-05,
+      "loss": 0.5223,
+      "step": 2719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930170007866699e-05,
+      "loss": 0.8484,
+      "step": 2720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930118664358272e-05,
+      "loss": 1.3115,
+      "step": 2721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930067302203682e-05,
+      "loss": 0.7204,
+      "step": 2722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9300159214032595e-05,
+      "loss": 0.774,
+      "step": 2723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.92996452195733e-05,
+      "loss": 0.5616,
+      "step": 2724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929913103866221e-05,
+      "loss": 0.2959,
+      "step": 2725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.92986166713026e-05,
+      "loss": 1.1404,
+      "step": 2726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929810211749774e-05,
+      "loss": 1.476,
+      "step": 2727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9297587377250894e-05,
+      "loss": 1.35,
+      "step": 2728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929707245056536e-05,
+      "loss": 0.4244,
+      "step": 2729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9296557337444394e-05,
+      "loss": 0.829,
+      "step": 2730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9296042037891285e-05,
+      "loss": 1.1587,
+      "step": 2731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929552655190931e-05,
+      "loss": 0.8993,
+      "step": 2732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929501087950175e-05,
+      "loss": 0.9386,
+      "step": 2733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929449502067188e-05,
+      "loss": 0.2318,
+      "step": 2734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929397897542299e-05,
+      "loss": 0.3798,
+      "step": 2735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929346274375836e-05,
+      "loss": 0.4701,
+      "step": 2736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929294632568128e-05,
+      "loss": 1.2082,
+      "step": 2737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929242972119503e-05,
+      "loss": 1.2994,
+      "step": 2738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.92919129303029e-05,
+      "loss": 0.3465,
+      "step": 2739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9291395953008186e-05,
+      "loss": 0.5373,
+      "step": 2740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929087878931416e-05,
+      "loss": 1.2316,
+      "step": 2741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9290361439224135e-05,
+      "loss": 1.1344,
+      "step": 2742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928984390274139e-05,
+      "loss": 0.5771,
+      "step": 2743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9289326179869214e-05,
+      "loss": 0.7135,
+      "step": 2744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928880827061092e-05,
+      "loss": 1.198,
+      "step": 2745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9288290174969785e-05,
+      "loss": 0.3488,
+      "step": 2746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928777189294911e-05,
+      "loss": 1.2442,
+      "step": 2747
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.92872534245522e-05,
+      "loss": 1.5295,
+      "step": 2748
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928673476978235e-05,
+      "loss": 0.3148,
+      "step": 2749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9286215928642874e-05,
+      "loss": 1.3636,
+      "step": 2750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9285696901137044e-05,
+      "loss": 1.1427,
+      "step": 2751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9285177687268185e-05,
+      "loss": 0.3014,
+      "step": 2752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.92846582870396e-05,
+      "loss": 1.1612,
+      "step": 2753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928413870045458e-05,
+      "loss": 1.7744,
+      "step": 2754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9283618927516444e-05,
+      "loss": 1.3985,
+      "step": 2755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9283098968228505e-05,
+      "loss": 0.7399,
+      "step": 2756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9282578822594055e-05,
+      "loss": 0.9183,
+      "step": 2757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928205849061642e-05,
+      "loss": 1.4553,
+      "step": 2758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.92815379722989e-05,
+      "loss": 0.3375,
+      "step": 2759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928101726764482e-05,
+      "loss": 1.4664,
+      "step": 2760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928049637665748e-05,
+      "loss": 1.0055,
+      "step": 2761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.92799752993402e-05,
+      "loss": 0.247,
+      "step": 2762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927945403569629e-05,
+      "loss": 0.9494,
+      "step": 2763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9278932585729084e-05,
+      "loss": 0.8962,
+      "step": 2764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927841094944188e-05,
+      "loss": 0.6279,
+      "step": 2765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9277889126838e-05,
+      "loss": 1.1228,
+      "step": 2766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927736711792079e-05,
+      "loss": 0.5425,
+      "step": 2767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9276844922693544e-05,
+      "loss": 0.8397,
+      "step": 2768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927632254115959e-05,
+      "loss": 0.6224,
+      "step": 2769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927579997332226e-05,
+      "loss": 1.4062,
+      "step": 2770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927527721918488e-05,
+      "loss": 1.0868,
+      "step": 2771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9274754278750775e-05,
+      "loss": 0.9379,
+      "step": 2772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9274231152023265e-05,
+      "loss": 0.3872,
+      "step": 2773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9273707839005684e-05,
+      "loss": 0.4311,
+      "step": 2774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927318433970137e-05,
+      "loss": 1.2329,
+      "step": 2775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927266065411363e-05,
+      "loss": 0.2967,
+      "step": 2776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9272136782245826e-05,
+      "loss": 0.526,
+      "step": 2777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9271612724101276e-05,
+      "loss": 0.5499,
+      "step": 2778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927108847968332e-05,
+      "loss": 0.9568,
+      "step": 2779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.92705640489953e-05,
+      "loss": 0.957,
+      "step": 2780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9270039432040525e-05,
+      "loss": 0.6194,
+      "step": 2781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.926951462882237e-05,
+      "loss": 0.4289,
+      "step": 2782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9268989639344156e-05,
+      "loss": 0.9951,
+      "step": 2783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.926846446360923e-05,
+      "loss": 1.0308,
+      "step": 2784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.926793910162092e-05,
+      "loss": 0.8851,
+      "step": 2785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9267413553382586e-05,
+      "loss": 0.5885,
+      "step": 2786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9266887818897567e-05,
+      "loss": 0.8935,
+      "step": 2787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9266361898169206e-05,
+      "loss": 0.4752,
+      "step": 2788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.926583579120085e-05,
+      "loss": 0.5286,
+      "step": 2789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.926530949799585e-05,
+      "loss": 0.8765,
+      "step": 2790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.926478301855755e-05,
+      "loss": 0.2945,
+      "step": 2791
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.926425635288931e-05,
+      "loss": 0.4005,
+      "step": 2792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9263729500994475e-05,
+      "loss": 0.7594,
+      "step": 2793
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.926320246287639e-05,
+      "loss": 1.1652,
+      "step": 2794
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.926267523853842e-05,
+      "loss": 1.0239,
+      "step": 2795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.926214782798393e-05,
+      "loss": 1.2178,
+      "step": 2796
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.926162023121625e-05,
+      "loss": 1.3344,
+      "step": 2797
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.926109244823875e-05,
+      "loss": 1.4422,
+      "step": 2798
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9260564479054785e-05,
+      "loss": 0.5753,
+      "step": 2799
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.926003632366773e-05,
+      "loss": 0.954,
+      "step": 2800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.925950798208093e-05,
+      "loss": 1.0034,
+      "step": 2801
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.925897945429775e-05,
+      "loss": 0.3456,
+      "step": 2802
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9258450740321567e-05,
+      "loss": 0.6127,
+      "step": 2803
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9257921840155715e-05,
+      "loss": 0.4492,
+      "step": 2804
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.925739275380359e-05,
+      "loss": 0.9029,
+      "step": 2805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9256863481268546e-05,
+      "loss": 0.6804,
+      "step": 2806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.925633402255395e-05,
+      "loss": 1.4224,
+      "step": 2807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.925580437766318e-05,
+      "loss": 0.7975,
+      "step": 2808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.92552745465996e-05,
+      "loss": 0.5334,
+      "step": 2809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.925474452936658e-05,
+      "loss": 1.1681,
+      "step": 2810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.925421432596749e-05,
+      "loss": 0.7034,
+      "step": 2811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9253683936405716e-05,
+      "loss": 1.7508,
+      "step": 2812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.925315336068462e-05,
+      "loss": 0.7097,
+      "step": 2813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9252622598807593e-05,
+      "loss": 0.4401,
+      "step": 2814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9252091650778e-05,
+      "loss": 0.7476,
+      "step": 2815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.925156051659922e-05,
+      "loss": 1.0557,
+      "step": 2816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9251029196274636e-05,
+      "loss": 1.6201,
+      "step": 2817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.925049768980764e-05,
+      "loss": 0.9322,
+      "step": 2818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924996599720159e-05,
+      "loss": 1.6787,
+      "step": 2819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924943411845989e-05,
+      "loss": 0.4933,
+      "step": 2820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924890205358593e-05,
+      "loss": 1.4084,
+      "step": 2821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9248369802583066e-05,
+      "loss": 0.1279,
+      "step": 2822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924783736545471e-05,
+      "loss": 0.9762,
+      "step": 2823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924730474220424e-05,
+      "loss": 0.3286,
+      "step": 2824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924677193283506e-05,
+      "loss": 0.4072,
+      "step": 2825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924623893735054e-05,
+      "loss": 0.693,
+      "step": 2826
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924570575575408e-05,
+      "loss": 0.2016,
+      "step": 2827
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924517238804908e-05,
+      "loss": 0.4246,
+      "step": 2828
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9244638834238924e-05,
+      "loss": 1.0796,
+      "step": 2829
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9244105094327017e-05,
+      "loss": 0.5491,
+      "step": 2830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9243571168316746e-05,
+      "loss": 0.9373,
+      "step": 2831
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924303705621151e-05,
+      "loss": 0.916,
+      "step": 2832
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924250275801472e-05,
+      "loss": 0.3933,
+      "step": 2833
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924196827372976e-05,
+      "loss": 1.198,
+      "step": 2834
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924143360336004e-05,
+      "loss": 0.3912,
+      "step": 2835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924089874690896e-05,
+      "loss": 1.0365,
+      "step": 2836
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924036370437992e-05,
+      "loss": 1.036,
+      "step": 2837
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923982847577634e-05,
+      "loss": 0.9326,
+      "step": 2838
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923929306110161e-05,
+      "loss": 1.3797,
+      "step": 2839
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9238757460359136e-05,
+      "loss": 0.9531,
+      "step": 2840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923822167355234e-05,
+      "loss": 0.4707,
+      "step": 2841
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923768570068462e-05,
+      "loss": 1.4376,
+      "step": 2842
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9237149541759396e-05,
+      "loss": 2.0054,
+      "step": 2843
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923661319678006e-05,
+      "loss": 0.6978,
+      "step": 2844
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923607666575006e-05,
+      "loss": 0.7265,
+      "step": 2845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923553994867278e-05,
+      "loss": 0.8122,
+      "step": 2846
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923500304555165e-05,
+      "loss": 1.5095,
+      "step": 2847
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923446595639008e-05,
+      "loss": 0.2637,
+      "step": 2848
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923392868119148e-05,
+      "loss": 0.4773,
+      "step": 2849
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923339121995929e-05,
+      "loss": 0.2558,
+      "step": 2850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923285357269692e-05,
+      "loss": 1.096,
+      "step": 2851
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923231573940778e-05,
+      "loss": 1.2387,
+      "step": 2852
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923177772009531e-05,
+      "loss": 0.7443,
+      "step": 2853
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923123951476293e-05,
+      "loss": 0.3131,
+      "step": 2854
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923070112341406e-05,
+      "loss": 1.7054,
+      "step": 2855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923016254605213e-05,
+      "loss": 1.2221,
+      "step": 2856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9229623782680563e-05,
+      "loss": 0.5848,
+      "step": 2857
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922908483330278e-05,
+      "loss": 0.5196,
+      "step": 2858
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9228545697922237e-05,
+      "loss": 0.2584,
+      "step": 2859
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922800637654234e-05,
+      "loss": 0.6905,
+      "step": 2860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922746686916652e-05,
+      "loss": 1.3144,
+      "step": 2861
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922692717579824e-05,
+      "loss": 0.7933,
+      "step": 2862
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9226387296440895e-05,
+      "loss": 0.7884,
+      "step": 2863
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922584723109796e-05,
+      "loss": 0.8868,
+      "step": 2864
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9225306979772836e-05,
+      "loss": 0.4059,
+      "step": 2865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922476654246898e-05,
+      "loss": 0.6388,
+      "step": 2866
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922422591918983e-05,
+      "loss": 0.6643,
+      "step": 2867
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922368510993882e-05,
+      "loss": 0.5397,
+      "step": 2868
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922314411471939e-05,
+      "loss": 1.9526,
+      "step": 2869
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9222602933535e-05,
+      "loss": 1.3379,
+      "step": 2870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9222061566389075e-05,
+      "loss": 1.0582,
+      "step": 2871
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9221520013285065e-05,
+      "loss": 1.4439,
+      "step": 2872
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922097827422642e-05,
+      "loss": 0.1558,
+      "step": 2873
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9220436349216585e-05,
+      "loss": 0.5585,
+      "step": 2874
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.921989423825901e-05,
+      "loss": 0.9535,
+      "step": 2875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9219351941357144e-05,
+      "loss": 0.5393,
+      "step": 2876
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.921880945851444e-05,
+      "loss": 0.647,
+      "step": 2877
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.921826678973435e-05,
+      "loss": 0.4546,
+      "step": 2878
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.921772393502032e-05,
+      "loss": 0.3729,
+      "step": 2879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.921718089437581e-05,
+      "loss": 0.7666,
+      "step": 2880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.921663766780427e-05,
+      "loss": 0.7895,
+      "step": 2881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.921609425530918e-05,
+      "loss": 0.2832,
+      "step": 2882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9215550656893965e-05,
+      "loss": 0.7428,
+      "step": 2883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9215006872562105e-05,
+      "loss": 0.4901,
+      "step": 2884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9214462902317056e-05,
+      "loss": 0.5311,
+      "step": 2885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9213918746162275e-05,
+      "loss": 0.4672,
+      "step": 2886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.921337440410123e-05,
+      "loss": 0.4962,
+      "step": 2887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.921282987613739e-05,
+      "loss": 0.714,
+      "step": 2888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.92122851622742e-05,
+      "loss": 0.9423,
+      "step": 2889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.921174026251515e-05,
+      "loss": 0.9459,
+      "step": 2890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.92111951768637e-05,
+      "loss": 0.7184,
+      "step": 2891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9210649905323314e-05,
+      "loss": 0.9499,
+      "step": 2892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.921010444789746e-05,
+      "loss": 0.5797,
+      "step": 2893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.920955880458962e-05,
+      "loss": 0.6779,
+      "step": 2894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.920901297540326e-05,
+      "loss": 1.0592,
+      "step": 2895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.920846696034184e-05,
+      "loss": 0.9393,
+      "step": 2896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.920792075940887e-05,
+      "loss": 0.5555,
+      "step": 2897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.920737437260778e-05,
+      "loss": 0.3911,
+      "step": 2898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.920682779994208e-05,
+      "loss": 0.8527,
+      "step": 2899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.920628104141524e-05,
+      "loss": 0.476,
+      "step": 2900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.920573409703074e-05,
+      "loss": 1.3307,
+      "step": 2901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9205186966792065e-05,
+      "loss": 0.5184,
+      "step": 2902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.920463965070268e-05,
+      "loss": 0.4153,
+      "step": 2903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.920409214876608e-05,
+      "loss": 0.392,
+      "step": 2904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9203544460985745e-05,
+      "loss": 0.7283,
+      "step": 2905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.920299658736516e-05,
+      "loss": 0.7322,
+      "step": 2906
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.920244852790782e-05,
+      "loss": 0.5474,
+      "step": 2907
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.920190028261721e-05,
+      "loss": 0.4088,
+      "step": 2908
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.920135185149681e-05,
+      "loss": 1.005,
+      "step": 2909
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9200803234550124e-05,
+      "loss": 0.8917,
+      "step": 2910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9200254431780626e-05,
+      "loss": 1.3583,
+      "step": 2911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919970544319182e-05,
+      "loss": 0.9509,
+      "step": 2912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.91991562687872e-05,
+      "loss": 0.7563,
+      "step": 2913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919860690857025e-05,
+      "loss": 1.3739,
+      "step": 2914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919805736254447e-05,
+      "loss": 0.9318,
+      "step": 2915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919750763071337e-05,
+      "loss": 1.1346,
+      "step": 2916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919695771308044e-05,
+      "loss": 0.6463,
+      "step": 2917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919640760964917e-05,
+      "loss": 1.083,
+      "step": 2918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919585732042307e-05,
+      "loss": 0.6667,
+      "step": 2919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919530684540565e-05,
+      "loss": 0.541,
+      "step": 2920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919475618460039e-05,
+      "loss": 1.3594,
+      "step": 2921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919420533801082e-05,
+      "loss": 1.1744,
+      "step": 2922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919365430564043e-05,
+      "loss": 0.3907,
+      "step": 2923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9193103087492726e-05,
+      "loss": 1.0834,
+      "step": 2924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919255168357122e-05,
+      "loss": 1.0398,
+      "step": 2925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9192000093879425e-05,
+      "loss": 0.532,
+      "step": 2926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919144831842084e-05,
+      "loss": 1.3751,
+      "step": 2927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919089635719899e-05,
+      "loss": 1.4594,
+      "step": 2928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919034421021738e-05,
+      "loss": 0.7526,
+      "step": 2929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918979187747952e-05,
+      "loss": 1.0144,
+      "step": 2930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918923935898893e-05,
+      "loss": 0.5603,
+      "step": 2931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918868665474912e-05,
+      "loss": 0.7797,
+      "step": 2932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9188133764763626e-05,
+      "loss": 1.1483,
+      "step": 2933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918758068903594e-05,
+      "loss": 0.8425,
+      "step": 2934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9187027427569596e-05,
+      "loss": 1.1235,
+      "step": 2935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918647398036811e-05,
+      "loss": 1.4581,
+      "step": 2936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918592034743501e-05,
+      "loss": 0.5403,
+      "step": 2937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918536652877382e-05,
+      "loss": 0.3219,
+      "step": 2938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9184812524388064e-05,
+      "loss": 0.2147,
+      "step": 2939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9184258334281255e-05,
+      "loss": 0.1997,
+      "step": 2940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918370395845693e-05,
+      "loss": 1.1778,
+      "step": 2941
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9183149396918616e-05,
+      "loss": 0.455,
+      "step": 2942
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9182594649669834e-05,
+      "loss": 0.9209,
+      "step": 2943
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9182039716714127e-05,
+      "loss": 1.2356,
+      "step": 2944
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9181484598055024e-05,
+      "loss": 0.9291,
+      "step": 2945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9180929293696044e-05,
+      "loss": 0.749,
+      "step": 2946
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918037380364074e-05,
+      "loss": 1.2336,
+      "step": 2947
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917981812789264e-05,
+      "loss": 0.4786,
+      "step": 2948
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917926226645528e-05,
+      "loss": 1.2246,
+      "step": 2949
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917870621933219e-05,
+      "loss": 0.7015,
+      "step": 2950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.91781499865269e-05,
+      "loss": 0.9072,
+      "step": 2951
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917759356804298e-05,
+      "loss": 1.2744,
+      "step": 2952
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917703696388395e-05,
+      "loss": 0.4463,
+      "step": 2953
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9176480174053355e-05,
+      "loss": 0.4358,
+      "step": 2954
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917592319855474e-05,
+      "loss": 0.4953,
+      "step": 2955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9175366037391656e-05,
+      "loss": 0.8504,
+      "step": 2956
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9174808690567635e-05,
+      "loss": 1.3182,
+      "step": 2957
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917425115808624e-05,
+      "loss": 1.1662,
+      "step": 2958
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9173693439950986e-05,
+      "loss": 1.1064,
+      "step": 2959
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9173135536165466e-05,
+      "loss": 0.4998,
+      "step": 2960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.91725774467332e-05,
+      "loss": 0.6833,
+      "step": 2961
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917201917165775e-05,
+      "loss": 1.3538,
+      "step": 2962
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917146071094267e-05,
+      "loss": 0.2013,
+      "step": 2963
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917090206459152e-05,
+      "loss": 0.3491,
+      "step": 2964
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917034323260783e-05,
+      "loss": 1.3042,
+      "step": 2965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916978421499518e-05,
+      "loss": 0.4853,
+      "step": 2966
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916922501175712e-05,
+      "loss": 0.4612,
+      "step": 2967
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916866562289721e-05,
+      "loss": 1.3429,
+      "step": 2968
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916810604841901e-05,
+      "loss": 0.2759,
+      "step": 2969
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916754628832607e-05,
+      "loss": 0.212,
+      "step": 2970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916698634262197e-05,
+      "loss": 0.5836,
+      "step": 2971
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916642621131026e-05,
+      "loss": 1.2393,
+      "step": 2972
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916586589439451e-05,
+      "loss": 1.1625,
+      "step": 2973
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916530539187828e-05,
+      "loss": 1.1571,
+      "step": 2974
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916474470376515e-05,
+      "loss": 0.8985,
+      "step": 2975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916418383005868e-05,
+      "loss": 1.3267,
+      "step": 2976
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916362277076243e-05,
+      "loss": 0.4606,
+      "step": 2977
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916306152587998e-05,
+      "loss": 0.3659,
+      "step": 2978
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916250009541491e-05,
+      "loss": 0.5498,
+      "step": 2979
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9161938479370786e-05,
+      "loss": 0.9286,
+      "step": 2980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916137667775116e-05,
+      "loss": 0.9443,
+      "step": 2981
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916081469055964e-05,
+      "loss": 0.6,
+      "step": 2982
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916025251779979e-05,
+      "loss": 0.475,
+      "step": 2983
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9159690159475184e-05,
+      "loss": 1.767,
+      "step": 2984
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9159127615589395e-05,
+      "loss": 1.5092,
+      "step": 2985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.915856488614602e-05,
+      "loss": 0.8064,
+      "step": 2986
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.915800197114863e-05,
+      "loss": 0.9938,
+      "step": 2987
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.91574388706008e-05,
+      "loss": 0.849,
+      "step": 2988
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9156875584506125e-05,
+      "loss": 1.6894,
+      "step": 2989
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.915631211286819e-05,
+      "loss": 1.5108,
+      "step": 2990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.915574845569058e-05,
+      "loss": 1.2392,
+      "step": 2991
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.915518461297687e-05,
+      "loss": 0.4963,
+      "step": 2992
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.915462058473066e-05,
+      "loss": 0.9816,
+      "step": 2993
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9154056370955534e-05,
+      "loss": 1.2292,
+      "step": 2994
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.915349197165507e-05,
+      "loss": 0.9534,
+      "step": 2995
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9152927386832894e-05,
+      "loss": 0.9726,
+      "step": 2996
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.915236261649258e-05,
+      "loss": 1.3728,
+      "step": 2997
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.915179766063771e-05,
+      "loss": 0.6356,
+      "step": 2998
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9151232519271886e-05,
+      "loss": 1.3041,
+      "step": 2999
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9150667192398715e-05,
+      "loss": 0.2374,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.915010168002179e-05,
+      "loss": 1.088,
+      "step": 3001
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.91495359821447e-05,
+      "loss": 0.8862,
+      "step": 3002
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.914897009877106e-05,
+      "loss": 1.3785,
+      "step": 3003
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914840402990445e-05,
+      "loss": 0.5526,
+      "step": 3004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9147837775548494e-05,
+      "loss": 0.4009,
+      "step": 3005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914727133570679e-05,
+      "loss": 1.3364,
+      "step": 3006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9146704710382936e-05,
+      "loss": 1.1238,
+      "step": 3007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914613789958054e-05,
+      "loss": 0.5144,
+      "step": 3008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914557090330321e-05,
+      "loss": 0.6594,
+      "step": 3009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9145003721554556e-05,
+      "loss": 0.9201,
+      "step": 3010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914443635433818e-05,
+      "loss": 0.1461,
+      "step": 3011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914386880165771e-05,
+      "loss": 1.1804,
+      "step": 3012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914330106351674e-05,
+      "loss": 1.3649,
+      "step": 3013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914273313991889e-05,
+      "loss": 1.2584,
+      "step": 3014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914216503086777e-05,
+      "loss": 1.1371,
+      "step": 3015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9141596736366994e-05,
+      "loss": 1.1143,
+      "step": 3016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914102825642019e-05,
+      "loss": 0.8118,
+      "step": 3017
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914045959103097e-05,
+      "loss": 0.5805,
+      "step": 3018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913989074020294e-05,
+      "loss": 0.4873,
+      "step": 3019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913932170393974e-05,
+      "loss": 1.5788,
+      "step": 3020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913875248224498e-05,
+      "loss": 0.9594,
+      "step": 3021
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913818307512228e-05,
+      "loss": 1.2666,
+      "step": 3022
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913761348257527e-05,
+      "loss": 1.2405,
+      "step": 3023
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9137043704607566e-05,
+      "loss": 1.2411,
+      "step": 3024
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9136473741222816e-05,
+      "loss": 1.494,
+      "step": 3025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913590359242462e-05,
+      "loss": 1.2451,
+      "step": 3026
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913533325821662e-05,
+      "loss": 0.7439,
+      "step": 3027
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913476273860244e-05,
+      "loss": 1.1388,
+      "step": 3028
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9134192033585715e-05,
+      "loss": 0.7016,
+      "step": 3029
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913362114317008e-05,
+      "loss": 1.0882,
+      "step": 3030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913305006735916e-05,
+      "loss": 1.5777,
+      "step": 3031
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9132478806156584e-05,
+      "loss": 1.4504,
+      "step": 3032
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913190735956601e-05,
+      "loss": 0.6834,
+      "step": 3033
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9131335727591046e-05,
+      "loss": 0.3708,
+      "step": 3034
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913076391023535e-05,
+      "loss": 1.2969,
+      "step": 3035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913019190750256e-05,
+      "loss": 0.5318,
+      "step": 3036
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9129619719396306e-05,
+      "loss": 0.6766,
+      "step": 3037
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9129047345920224e-05,
+      "loss": 0.8981,
+      "step": 3038
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.912847478707798e-05,
+      "loss": 0.8921,
+      "step": 3039
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9127902042873195e-05,
+      "loss": 0.7779,
+      "step": 3040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.912732911330952e-05,
+      "loss": 0.8777,
+      "step": 3041
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9126755998390614e-05,
+      "loss": 0.6166,
+      "step": 3042
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9126182698120104e-05,
+      "loss": 0.5984,
+      "step": 3043
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.912560921250165e-05,
+      "loss": 0.4443,
+      "step": 3044
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9125035541538897e-05,
+      "loss": 1.0342,
+      "step": 3045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.91244616852355e-05,
+      "loss": 1.4234,
+      "step": 3046
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.912388764359511e-05,
+      "loss": 0.6511,
+      "step": 3047
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9123313416621374e-05,
+      "loss": 0.1923,
+      "step": 3048
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.912273900431796e-05,
+      "loss": 1.0033,
+      "step": 3049
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.912216440668851e-05,
+      "loss": 0.8234,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9121589623736675e-05,
+      "loss": 0.364,
+      "step": 3051
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.912101465546613e-05,
+      "loss": 1.0198,
+      "step": 3052
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9120439501880524e-05,
+      "loss": 0.9399,
+      "step": 3053
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9119864162983516e-05,
+      "loss": 1.3259,
+      "step": 3054
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911928863877878e-05,
+      "loss": 1.3955,
+      "step": 3055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911871292926996e-05,
+      "loss": 1.1932,
+      "step": 3056
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911813703446073e-05,
+      "loss": 1.1828,
+      "step": 3057
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9117560954354754e-05,
+      "loss": 0.4412,
+      "step": 3058
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9116984688955694e-05,
+      "loss": 1.1149,
+      "step": 3059
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911640823826722e-05,
+      "loss": 0.3981,
+      "step": 3060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911583160229301e-05,
+      "loss": 0.6181,
+      "step": 3061
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9115254781036714e-05,
+      "loss": 1.0785,
+      "step": 3062
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911467777450201e-05,
+      "loss": 1.1694,
+      "step": 3063
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911410058269258e-05,
+      "loss": 0.5186,
+      "step": 3064
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911352320561209e-05,
+      "loss": 1.2296,
+      "step": 3065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911294564326421e-05,
+      "loss": 0.2419,
+      "step": 3066
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911236789565262e-05,
+      "loss": 0.6656,
+      "step": 3067
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9111789962781e-05,
+      "loss": 1.0874,
+      "step": 3068
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9111211844653016e-05,
+      "loss": 0.8867,
+      "step": 3069
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9110633541272356e-05,
+      "loss": 0.4455,
+      "step": 3070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.91100550526427e-05,
+      "loss": 1.3425,
+      "step": 3071
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9109476378767725e-05,
+      "loss": 1.2011,
+      "step": 3072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9108897519651126e-05,
+      "loss": 1.4782,
+      "step": 3073
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.910831847529657e-05,
+      "loss": 1.5563,
+      "step": 3074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.910773924570774e-05,
+      "loss": 0.5571,
+      "step": 3075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.910715983088834e-05,
+      "loss": 1.087,
+      "step": 3076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9106580230842044e-05,
+      "loss": 1.0741,
+      "step": 3077
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9106000445572554e-05,
+      "loss": 0.8211,
+      "step": 3078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.910542047508354e-05,
+      "loss": 1.5822,
+      "step": 3079
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.91048403193787e-05,
+      "loss": 0.721,
+      "step": 3080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.910425997846174e-05,
+      "loss": 1.2061,
+      "step": 3081
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9103679452336326e-05,
+      "loss": 0.5612,
+      "step": 3082
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9103098741006175e-05,
+      "loss": 1.1252,
+      "step": 3083
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.910251784447498e-05,
+      "loss": 1.2018,
+      "step": 3084
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.910193676274643e-05,
+      "loss": 1.0212,
+      "step": 3085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.910135549582422e-05,
+      "loss": 0.9368,
+      "step": 3086
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.910077404371206e-05,
+      "loss": 0.829,
+      "step": 3087
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9100192406413644e-05,
+      "loss": 1.3476,
+      "step": 3088
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.909961058393267e-05,
+      "loss": 1.1211,
+      "step": 3089
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.909902857627284e-05,
+      "loss": 0.8948,
+      "step": 3090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.909844638343787e-05,
+      "loss": 1.4558,
+      "step": 3091
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9097864005431464e-05,
+      "loss": 1.1084,
+      "step": 3092
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.909728144225732e-05,
+      "loss": 1.228,
+      "step": 3093
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.909669869391913e-05,
+      "loss": 1.0005,
+      "step": 3094
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.909611576042062e-05,
+      "loss": 0.7323,
+      "step": 3095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.909553264176551e-05,
+      "loss": 0.4748,
+      "step": 3096
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9094949337957496e-05,
+      "loss": 0.217,
+      "step": 3097
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.909436584900029e-05,
+      "loss": 0.3674,
+      "step": 3098
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.90937821748976e-05,
+      "loss": 0.4721,
+      "step": 3099
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9093198315653163e-05,
+      "loss": 1.0575,
+      "step": 3100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.909261427127067e-05,
+      "loss": 0.5585,
+      "step": 3101
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.909203004175385e-05,
+      "loss": 0.7611,
+      "step": 3102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.909144562710642e-05,
+      "loss": 0.3245,
+      "step": 3103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.909086102733209e-05,
+      "loss": 0.7421,
+      "step": 3104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.909027624243459e-05,
+      "loss": 1.1548,
+      "step": 3105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.908969127241764e-05,
+      "loss": 1.1382,
+      "step": 3106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9089106117284965e-05,
+      "loss": 1.3959,
+      "step": 3107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.908852077704029e-05,
+      "loss": 0.434,
+      "step": 3108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.908793525168732e-05,
+      "loss": 1.1544,
+      "step": 3109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.90873495412298e-05,
+      "loss": 1.088,
+      "step": 3110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9086763645671454e-05,
+      "loss": 0.3729,
+      "step": 3111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9086177565016005e-05,
+      "loss": 0.1659,
+      "step": 3112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9085591299267194e-05,
+      "loss": 0.7377,
+      "step": 3113
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.908500484842874e-05,
+      "loss": 0.8454,
+      "step": 3114
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.908441821250438e-05,
+      "loss": 1.091,
+      "step": 3115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.908383139149786e-05,
+      "loss": 1.4637,
+      "step": 3116
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9083244385412884e-05,
+      "loss": 0.3594,
+      "step": 3117
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9082657194253215e-05,
+      "loss": 0.7178,
+      "step": 3118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9082069818022574e-05,
+      "loss": 0.7441,
+      "step": 3119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9081482256724706e-05,
+      "loss": 0.648,
+      "step": 3120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.908089451036335e-05,
+      "loss": 1.4934,
+      "step": 3121
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9080306578942236e-05,
+      "loss": 1.2529,
+      "step": 3122
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.907971846246512e-05,
+      "loss": 1.028,
+      "step": 3123
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9079130160935735e-05,
+      "loss": 0.5571,
+      "step": 3124
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.907854167435783e-05,
+      "loss": 0.467,
+      "step": 3125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.907795300273514e-05,
+      "loss": 1.0174,
+      "step": 3126
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9077364146071434e-05,
+      "loss": 0.3982,
+      "step": 3127
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.907677510437044e-05,
+      "loss": 0.6749,
+      "step": 3128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.907618587763591e-05,
+      "loss": 0.26,
+      "step": 3129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.907559646587159e-05,
+      "loss": 1.3655,
+      "step": 3130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9075006869081224e-05,
+      "loss": 0.5134,
+      "step": 3131
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9074417087268596e-05,
+      "loss": 0.7376,
+      "step": 3132
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.907382712043742e-05,
+      "loss": 0.4792,
+      "step": 3133
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9073236968591486e-05,
+      "loss": 1.0223,
+      "step": 3134
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9072646631734526e-05,
+      "loss": 1.7371,
+      "step": 3135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.90720561098703e-05,
+      "loss": 1.1529,
+      "step": 3136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9071465403002565e-05,
+      "loss": 1.8652,
+      "step": 3137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.907087451113509e-05,
+      "loss": 0.6044,
+      "step": 3138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9070283434271626e-05,
+      "loss": 0.9829,
+      "step": 3139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906969217241594e-05,
+      "loss": 0.6484,
+      "step": 3140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906910072557179e-05,
+      "loss": 0.4032,
+      "step": 3141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906850909374294e-05,
+      "loss": 0.6202,
+      "step": 3142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9067917276933164e-05,
+      "loss": 1.2752,
+      "step": 3143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906732527514621e-05,
+      "loss": 0.5075,
+      "step": 3144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906673308838586e-05,
+      "loss": 1.4452,
+      "step": 3145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906614071665587e-05,
+      "loss": 1.2531,
+      "step": 3146
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906554815996003e-05,
+      "loss": 1.2336,
+      "step": 3147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906495541830209e-05,
+      "loss": 0.2719,
+      "step": 3148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906436249168584e-05,
+      "loss": 0.8961,
+      "step": 3149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906376938011504e-05,
+      "loss": 1.121,
+      "step": 3150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906317608359346e-05,
+      "loss": 0.5384,
+      "step": 3151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906258260212489e-05,
+      "loss": 0.8803,
+      "step": 3152
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.90619889357131e-05,
+      "loss": 0.8505,
+      "step": 3153
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906139508436187e-05,
+      "loss": 0.5178,
+      "step": 3154
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906080104807498e-05,
+      "loss": 0.686,
+      "step": 3155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.90602068268562e-05,
+      "loss": 1.7751,
+      "step": 3156
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9059612420709324e-05,
+      "loss": 1.4534,
+      "step": 3157
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905901782963813e-05,
+      "loss": 0.9519,
+      "step": 3158
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905842305364639e-05,
+      "loss": 1.1705,
+      "step": 3159
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905782809273791e-05,
+      "loss": 0.7248,
+      "step": 3160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905723294691647e-05,
+      "loss": 0.614,
+      "step": 3161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905663761618584e-05,
+      "loss": 1.3354,
+      "step": 3162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905604210054983e-05,
+      "loss": 0.944,
+      "step": 3163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905544640001223e-05,
+      "loss": 1.4183,
+      "step": 3164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9054850514576804e-05,
+      "loss": 0.1765,
+      "step": 3165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905425444424738e-05,
+      "loss": 0.5379,
+      "step": 3166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905365818902773e-05,
+      "loss": 1.2533,
+      "step": 3167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9053061748921635e-05,
+      "loss": 0.7301,
+      "step": 3168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905246512393293e-05,
+      "loss": 0.6845,
+      "step": 3169
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905186831406538e-05,
+      "loss": 1.4804,
+      "step": 3170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905127131932279e-05,
+      "loss": 0.624,
+      "step": 3171
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905067413970896e-05,
+      "loss": 0.5738,
+      "step": 3172
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9050076775227694e-05,
+      "loss": 1.8128,
+      "step": 3173
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.904947922588278e-05,
+      "loss": 0.9165,
+      "step": 3174
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.904888149167805e-05,
+      "loss": 0.1917,
+      "step": 3175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.904828357261728e-05,
+      "loss": 0.6092,
+      "step": 3176
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9047685468704285e-05,
+      "loss": 1.4007,
+      "step": 3177
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.904708717994286e-05,
+      "loss": 0.712,
+      "step": 3178
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9046488706336836e-05,
+      "loss": 0.8811,
+      "step": 3179
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9045890047890005e-05,
+      "loss": 0.323,
+      "step": 3180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.904529120460617e-05,
+      "loss": 0.9542,
+      "step": 3181
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9044692176489166e-05,
+      "loss": 0.4523,
+      "step": 3182
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.904409296354278e-05,
+      "loss": 1.2917,
+      "step": 3183
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.904349356577083e-05,
+      "loss": 1.1141,
+      "step": 3184
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9042893983177154e-05,
+      "loss": 1.5931,
+      "step": 3185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9042294215765535e-05,
+      "loss": 1.8676,
+      "step": 3186
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.904169426353981e-05,
+      "loss": 0.9426,
+      "step": 3187
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9041094126503784e-05,
+      "loss": 0.5241,
+      "step": 3188
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.904049380466129e-05,
+      "loss": 0.5904,
+      "step": 3189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.903989329801614e-05,
+      "loss": 0.4892,
+      "step": 3190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.903929260657215e-05,
+      "loss": 0.3979,
+      "step": 3191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9038691730333156e-05,
+      "loss": 1.6905,
+      "step": 3192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.903809066930297e-05,
+      "loss": 0.8208,
+      "step": 3193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9037489423485427e-05,
+      "loss": 1.7912,
+      "step": 3194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9036887992884345e-05,
+      "loss": 0.9327,
+      "step": 3195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9036286377503554e-05,
+      "loss": 0.9727,
+      "step": 3196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9035684577346884e-05,
+      "loss": 0.1443,
+      "step": 3197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9035082592418154e-05,
+      "loss": 0.599,
+      "step": 3198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9034480422721215e-05,
+      "loss": 0.8083,
+      "step": 3199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.903387806825988e-05,
+      "loss": 1.5149,
+      "step": 3200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9033275529037996e-05,
+      "loss": 0.4109,
+      "step": 3201
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9032672805059386e-05,
+      "loss": 0.6354,
+      "step": 3202
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.90320698963279e-05,
+      "loss": 0.3985,
+      "step": 3203
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.903146680284736e-05,
+      "loss": 1.0418,
+      "step": 3204
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.903086352462161e-05,
+      "loss": 1.495,
+      "step": 3205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.903026006165448e-05,
+      "loss": 0.6177,
+      "step": 3206
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.902965641394983e-05,
+      "loss": 0.754,
+      "step": 3207
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9029052581511485e-05,
+      "loss": 0.5661,
+      "step": 3208
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.90284485643433e-05,
+      "loss": 0.2879,
+      "step": 3209
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.90278443624491e-05,
+      "loss": 1.2565,
+      "step": 3210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9027239975832746e-05,
+      "loss": 1.1427,
+      "step": 3211
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.902663540449808e-05,
+      "loss": 0.458,
+      "step": 3212
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.902603064844895e-05,
+      "loss": 0.496,
+      "step": 3213
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9025425707689205e-05,
+      "loss": 0.8751,
+      "step": 3214
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.902482058222268e-05,
+      "loss": 0.6154,
+      "step": 3215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.902421527205325e-05,
+      "loss": 0.5362,
+      "step": 3216
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9023609777184754e-05,
+      "loss": 1.0153,
+      "step": 3217
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.902300409762104e-05,
+      "loss": 1.0864,
+      "step": 3218
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.902239823336598e-05,
+      "loss": 0.203,
+      "step": 3219
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.902179218442341e-05,
+      "loss": 1.5964,
+      "step": 3220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9021185950797196e-05,
+      "loss": 0.6535,
+      "step": 3221
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.90205795324912e-05,
+      "loss": 0.3327,
+      "step": 3222
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9019972929509275e-05,
+      "loss": 1.3292,
+      "step": 3223
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.901936614185527e-05,
+      "loss": 0.794,
+      "step": 3224
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.901875916953307e-05,
+      "loss": 1.209,
+      "step": 3225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.901815201254653e-05,
+      "loss": 0.5191,
+      "step": 3226
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.90175446708995e-05,
+      "loss": 0.669,
+      "step": 3227
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.901693714459586e-05,
+      "loss": 0.9227,
+      "step": 3228
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9016329433639465e-05,
+      "loss": 1.1135,
+      "step": 3229
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9015721538034185e-05,
+      "loss": 0.7475,
+      "step": 3230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9015113457783895e-05,
+      "loss": 0.7241,
+      "step": 3231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.901450519289247e-05,
+      "loss": 1.2028,
+      "step": 3232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.901389674336376e-05,
+      "loss": 0.8122,
+      "step": 3233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9013288109201656e-05,
+      "loss": 1.0695,
+      "step": 3234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.901267929041002e-05,
+      "loss": 0.9754,
+      "step": 3235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9012070286992735e-05,
+      "loss": 1.2693,
+      "step": 3236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.901146109895367e-05,
+      "loss": 0.5327,
+      "step": 3237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9010851726296695e-05,
+      "loss": 1.1508,
+      "step": 3238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9010242169025704e-05,
+      "loss": 0.2984,
+      "step": 3239
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9009632427144566e-05,
+      "loss": 0.6157,
+      "step": 3240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9009022500657164e-05,
+      "loss": 0.3528,
+      "step": 3241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900841238956738e-05,
+      "loss": 0.4593,
+      "step": 3242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900780209387909e-05,
+      "loss": 1.2156,
+      "step": 3243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900719161359618e-05,
+      "loss": 0.3539,
+      "step": 3244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900658094872254e-05,
+      "loss": 1.2163,
+      "step": 3245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900597009926206e-05,
+      "loss": 1.167,
+      "step": 3246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900535906521861e-05,
+      "loss": 0.4452,
+      "step": 3247
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.90047478465961e-05,
+      "loss": 0.6498,
+      "step": 3248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.90041364433984e-05,
+      "loss": 0.3825,
+      "step": 3249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9003524855629414e-05,
+      "loss": 1.3789,
+      "step": 3250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900291308329303e-05,
+      "loss": 0.5492,
+      "step": 3251
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900230112639314e-05,
+      "loss": 0.3856,
+      "step": 3252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900168898493364e-05,
+      "loss": 1.3306,
+      "step": 3253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9001076658918416e-05,
+      "loss": 1.0986,
+      "step": 3254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9000464148351383e-05,
+      "loss": 0.7819,
+      "step": 3255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8999851453236424e-05,
+      "loss": 0.4933,
+      "step": 3256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899923857357744e-05,
+      "loss": 0.6041,
+      "step": 3257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8998625509378336e-05,
+      "loss": 0.3275,
+      "step": 3258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8998012260643015e-05,
+      "loss": 1.1732,
+      "step": 3259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899739882737537e-05,
+      "loss": 0.7959,
+      "step": 3260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899678520957931e-05,
+      "loss": 0.6991,
+      "step": 3261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899617140725874e-05,
+      "loss": 1.1714,
+      "step": 3262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899555742041757e-05,
+      "loss": 1.0148,
+      "step": 3263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.89949432490597e-05,
+      "loss": 0.7366,
+      "step": 3264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899432889318904e-05,
+      "loss": 0.5095,
+      "step": 3265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8993714352809506e-05,
+      "loss": 1.2302,
+      "step": 3266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8993099627925006e-05,
+      "loss": 0.5373,
+      "step": 3267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899248471853943e-05,
+      "loss": 1.2059,
+      "step": 3268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899186962465673e-05,
+      "loss": 0.5253,
+      "step": 3269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.89912543462808e-05,
+      "loss": 1.1642,
+      "step": 3270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899063888341555e-05,
+      "loss": 0.9109,
+      "step": 3271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8990023236064896e-05,
+      "loss": 0.5766,
+      "step": 3272
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.898940740423277e-05,
+      "loss": 1.2439,
+      "step": 3273
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.898879138792309e-05,
+      "loss": 0.1807,
+      "step": 3274
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.898817518713976e-05,
+      "loss": 1.3828,
+      "step": 3275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.898755880188672e-05,
+      "loss": 0.9087,
+      "step": 3276
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8986942232167866e-05,
+      "loss": 0.9561,
+      "step": 3277
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.898632547798715e-05,
+      "loss": 1.4656,
+      "step": 3278
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8985708539348495e-05,
+      "loss": 0.2311,
+      "step": 3279
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.89850914162558e-05,
+      "loss": 1.9623,
+      "step": 3280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8984474108713026e-05,
+      "loss": 1.042,
+      "step": 3281
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8983856616724075e-05,
+      "loss": 0.6724,
+      "step": 3282
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8983238940292886e-05,
+      "loss": 0.9655,
+      "step": 3283
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.898262107942339e-05,
+      "loss": 0.6492,
+      "step": 3284
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.898200303411953e-05,
+      "loss": 1.3109,
+      "step": 3285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.898138480438522e-05,
+      "loss": 0.4461,
+      "step": 3286
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8980766390224396e-05,
+      "loss": 0.761,
+      "step": 3287
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8980147791641014e-05,
+      "loss": 1.3946,
+      "step": 3288
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.897952900863898e-05,
+      "loss": 0.7648,
+      "step": 3289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8978910041222254e-05,
+      "loss": 1.0262,
+      "step": 3290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8978290889394776e-05,
+      "loss": 1.2347,
+      "step": 3291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.897767155316047e-05,
+      "loss": 0.3197,
+      "step": 3292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.897705203252329e-05,
+      "loss": 0.5322,
+      "step": 3293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.897643232748718e-05,
+      "loss": 1.2205,
+      "step": 3294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8975812438056063e-05,
+      "loss": 0.4588,
+      "step": 3295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.89751923642339e-05,
+      "loss": 0.9987,
+      "step": 3296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.897457210602465e-05,
+      "loss": 1.7143,
+      "step": 3297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8973951663432234e-05,
+      "loss": 1.0155,
+      "step": 3298
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.897333103646062e-05,
+      "loss": 0.4432,
+      "step": 3299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8972710225113743e-05,
+      "loss": 0.36,
+      "step": 3300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.897208922939556e-05,
+      "loss": 0.1236,
+      "step": 3301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8971468049310025e-05,
+      "loss": 1.41,
+      "step": 3302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8970846684861083e-05,
+      "loss": 0.8196,
+      "step": 3303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8970225136052704e-05,
+      "loss": 1.3211,
+      "step": 3304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896960340288883e-05,
+      "loss": 1.0442,
+      "step": 3305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896898148537342e-05,
+      "loss": 0.684,
+      "step": 3306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896835938351043e-05,
+      "loss": 0.7679,
+      "step": 3307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896773709730382e-05,
+      "loss": 0.5166,
+      "step": 3308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8967114626757555e-05,
+      "loss": 0.4887,
+      "step": 3309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896649197187559e-05,
+      "loss": 0.6267,
+      "step": 3310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896586913266188e-05,
+      "loss": 0.8373,
+      "step": 3311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896524610912041e-05,
+      "loss": 1.1324,
+      "step": 3312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896462290125513e-05,
+      "loss": 0.1647,
+      "step": 3313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8963999509070004e-05,
+      "loss": 1.1119,
+      "step": 3314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896337593256901e-05,
+      "loss": 0.9758,
+      "step": 3315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8962752171756116e-05,
+      "loss": 1.5248,
+      "step": 3316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896212822663527e-05,
+      "loss": 0.8213,
+      "step": 3317
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896150409721047e-05,
+      "loss": 0.8422,
+      "step": 3318
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896087978348567e-05,
+      "loss": 1.0109,
+      "step": 3319
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8960255285464844e-05,
+      "loss": 0.7699,
+      "step": 3320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895963060315198e-05,
+      "loss": 1.4057,
+      "step": 3321
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8959005736551034e-05,
+      "loss": 1.0115,
+      "step": 3322
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8958380685666e-05,
+      "loss": 1.2941,
+      "step": 3323
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895775545050084e-05,
+      "loss": 0.8889,
+      "step": 3324
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895713003105955e-05,
+      "loss": 0.9885,
+      "step": 3325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895650442734609e-05,
+      "loss": 1.2369,
+      "step": 3326
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895587863936446e-05,
+      "loss": 0.2925,
+      "step": 3327
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895525266711863e-05,
+      "loss": 1.4893,
+      "step": 3328
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8954626510612586e-05,
+      "loss": 1.3968,
+      "step": 3329
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895400016985032e-05,
+      "loss": 1.0884,
+      "step": 3330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895337364483581e-05,
+      "loss": 0.6071,
+      "step": 3331
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895274693557303e-05,
+      "loss": 0.4025,
+      "step": 3332
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8952120042065996e-05,
+      "loss": 0.5125,
+      "step": 3333
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895149296431869e-05,
+      "loss": 0.7736,
+      "step": 3334
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8950865702335084e-05,
+      "loss": 0.9958,
+      "step": 3335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895023825611918e-05,
+      "loss": 0.7308,
+      "step": 3336
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894961062567499e-05,
+      "loss": 0.4879,
+      "step": 3337
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894898281100647e-05,
+      "loss": 0.742,
+      "step": 3338
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894835481211766e-05,
+      "loss": 0.3114,
+      "step": 3339
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894772662901251e-05,
+      "loss": 1.3347,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894709826169505e-05,
+      "loss": 0.2736,
+      "step": 3341
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894646971016927e-05,
+      "loss": 1.0715,
+      "step": 3342
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894584097443917e-05,
+      "loss": 1.1559,
+      "step": 3343
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8945212054508744e-05,
+      "loss": 0.8747,
+      "step": 3344
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894458295038201e-05,
+      "loss": 1.0809,
+      "step": 3345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894395366206294e-05,
+      "loss": 0.6553,
+      "step": 3346
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894332418955558e-05,
+      "loss": 2.2535,
+      "step": 3347
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8942694532863905e-05,
+      "loss": 1.229,
+      "step": 3348
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8942064691991936e-05,
+      "loss": 0.8173,
+      "step": 3349
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894143466694368e-05,
+      "loss": 0.5435,
+      "step": 3350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894080445772314e-05,
+      "loss": 0.5655,
+      "step": 3351
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894017406433433e-05,
+      "loss": 0.4995,
+      "step": 3352
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893954348678127e-05,
+      "loss": 0.4179,
+      "step": 3353
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8938912725067953e-05,
+      "loss": 0.7089,
+      "step": 3354
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.89382817791984e-05,
+      "loss": 1.1017,
+      "step": 3355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893765064917664e-05,
+      "loss": 1.301,
+      "step": 3356
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893701933500668e-05,
+      "loss": 1.9343,
+      "step": 3357
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8936387836692537e-05,
+      "loss": 1.3105,
+      "step": 3358
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893575615423823e-05,
+      "loss": 0.4783,
+      "step": 3359
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893512428764777e-05,
+      "loss": 1.6913,
+      "step": 3360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893449223692519e-05,
+      "loss": 0.4916,
+      "step": 3361
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893386000207452e-05,
+      "loss": 0.7971,
+      "step": 3362
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8933227583099765e-05,
+      "loss": 0.366,
+      "step": 3363
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893259498000496e-05,
+      "loss": 0.8438,
+      "step": 3364
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893196219279411e-05,
+      "loss": 0.428,
+      "step": 3365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8931329221471274e-05,
+      "loss": 0.7396,
+      "step": 3366
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8930696066040466e-05,
+      "loss": 1.1552,
+      "step": 3367
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893006272650571e-05,
+      "loss": 1.8914,
+      "step": 3368
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8929429202871036e-05,
+      "loss": 1.8577,
+      "step": 3369
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.892879549514049e-05,
+      "loss": 1.3033,
+      "step": 3370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8928161603318085e-05,
+      "loss": 0.5055,
+      "step": 3371
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.892752752740787e-05,
+      "loss": 1.0713,
+      "step": 3372
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.892689326741388e-05,
+      "loss": 0.56,
+      "step": 3373
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.892625882334013e-05,
+      "loss": 0.9094,
+      "step": 3374
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.892562419519069e-05,
+      "loss": 1.6608,
+      "step": 3375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.892498938296957e-05,
+      "loss": 1.2266,
+      "step": 3376
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.892435438668082e-05,
+      "loss": 0.9834,
+      "step": 3377
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.892371920632848e-05,
+      "loss": 0.6629,
+      "step": 3378
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.89230838419166e-05,
+      "loss": 1.7338,
+      "step": 3379
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8922448293449226e-05,
+      "loss": 0.7707,
+      "step": 3380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8921812560930386e-05,
+      "loss": 0.8709,
+      "step": 3381
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.892117664436413e-05,
+      "loss": 1.7744,
+      "step": 3382
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.892054054375451e-05,
+      "loss": 1.0392,
+      "step": 3383
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.891990425910557e-05,
+      "loss": 1.0236,
+      "step": 3384
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8919267790421365e-05,
+      "loss": 0.8636,
+      "step": 3385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.891863113770595e-05,
+      "loss": 1.4785,
+      "step": 3386
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.891799430096335e-05,
+      "loss": 0.4754,
+      "step": 3387
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.891735728019765e-05,
+      "loss": 0.5487,
+      "step": 3388
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8916720075412874e-05,
+      "loss": 0.7881,
+      "step": 3389
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.891608268661311e-05,
+      "loss": 1.1368,
+      "step": 3390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.891544511380238e-05,
+      "loss": 1.1579,
+      "step": 3391
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.891480735698476e-05,
+      "loss": 0.4639,
+      "step": 3392
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.891416941616431e-05,
+      "loss": 0.9145,
+      "step": 3393
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.891353129134509e-05,
+      "loss": 1.0996,
+      "step": 3394
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.891289298253114e-05,
+      "loss": 0.6822,
+      "step": 3395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.891225448972655e-05,
+      "loss": 1.0082,
+      "step": 3396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.891161581293537e-05,
+      "loss": 1.2106,
+      "step": 3397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.891097695216166e-05,
+      "loss": 1.1716,
+      "step": 3398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.89103379074095e-05,
+      "loss": 1.0257,
+      "step": 3399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8909698678682944e-05,
+      "loss": 0.741,
+      "step": 3400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.890905926598606e-05,
+      "loss": 1.56,
+      "step": 3401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8908419669322926e-05,
+      "loss": 0.8407,
+      "step": 3402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8907779888697605e-05,
+      "loss": 0.6922,
+      "step": 3403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.890713992411417e-05,
+      "loss": 0.2846,
+      "step": 3404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8906499775576685e-05,
+      "loss": 0.8283,
+      "step": 3405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8905859443089246e-05,
+      "loss": 1.2255,
+      "step": 3406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.89052189266559e-05,
+      "loss": 0.4036,
+      "step": 3407
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.890457822628075e-05,
+      "loss": 1.1326,
+      "step": 3408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.890393734196785e-05,
+      "loss": 0.8437,
+      "step": 3409
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.89032962737213e-05,
+      "loss": 0.6786,
+      "step": 3410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8902655021545155e-05,
+      "loss": 0.9078,
+      "step": 3411
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.890201358544352e-05,
+      "loss": 0.9606,
+      "step": 3412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.890137196542045e-05,
+      "loss": 0.2947,
+      "step": 3413
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8900730161480054e-05,
+      "loss": 1.3145,
+      "step": 3414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.89000881736264e-05,
+      "loss": 0.616,
+      "step": 3415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.889944600186359e-05,
+      "loss": 1.4381,
+      "step": 3416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.88988036461957e-05,
+      "loss": 0.4314,
+      "step": 3417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.88981611066268e-05,
+      "loss": 1.0807,
+      "step": 3418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8897518383161004e-05,
+      "loss": 0.3365,
+      "step": 3419
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.88968754758024e-05,
+      "loss": 1.0352,
+      "step": 3420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.889623238455507e-05,
+      "loss": 0.2433,
+      "step": 3421
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8895589109423107e-05,
+      "loss": 0.8161,
+      "step": 3422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8894945650410614e-05,
+      "loss": 0.7082,
+      "step": 3423
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8894302007521675e-05,
+      "loss": 0.4755,
+      "step": 3424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.889365818076039e-05,
+      "loss": 0.4169,
+      "step": 3425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8893014170130855e-05,
+      "loss": 0.2656,
+      "step": 3426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.889236997563717e-05,
+      "loss": 0.2342,
+      "step": 3427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.889172559728343e-05,
+      "loss": 0.5268,
+      "step": 3428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.889108103507375e-05,
+      "loss": 0.4554,
+      "step": 3429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8890436289012215e-05,
+      "loss": 0.584,
+      "step": 3430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8889791359102934e-05,
+      "loss": 1.1995,
+      "step": 3431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.888914624535001e-05,
+      "loss": 1.3482,
+      "step": 3432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.888850094775755e-05,
+      "loss": 0.6695,
+      "step": 3433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.888785546632966e-05,
+      "loss": 1.3015,
+      "step": 3434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8887209801070445e-05,
+      "loss": 0.549,
+      "step": 3435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8886563951984014e-05,
+      "loss": 0.6433,
+      "step": 3436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.888591791907448e-05,
+      "loss": 1.1518,
+      "step": 3437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.888527170234597e-05,
+      "loss": 0.3103,
+      "step": 3438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.888462530180255e-05,
+      "loss": 0.6997,
+      "step": 3439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8883978717448386e-05,
+      "loss": 0.7393,
+      "step": 3440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.888333194928756e-05,
+      "loss": 0.3248,
+      "step": 3441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.88826849973242e-05,
+      "loss": 1.3735,
+      "step": 3442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8882037861562426e-05,
+      "loss": 0.7454,
+      "step": 3443
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8881390542006335e-05,
+      "loss": 0.3714,
+      "step": 3444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.888074303866007e-05,
+      "loss": 0.2779,
+      "step": 3445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8880095351527745e-05,
+      "loss": 1.2573,
+      "step": 3446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887944748061348e-05,
+      "loss": 0.2671,
+      "step": 3447
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887879942592139e-05,
+      "loss": 0.384,
+      "step": 3448
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8878151187455616e-05,
+      "loss": 1.1886,
+      "step": 3449
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887750276522027e-05,
+      "loss": 0.9315,
+      "step": 3450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887685415921948e-05,
+      "loss": 0.2024,
+      "step": 3451
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887620536945738e-05,
+      "loss": 0.4159,
+      "step": 3452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8875556395938086e-05,
+      "loss": 0.6584,
+      "step": 3453
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887490723866574e-05,
+      "loss": 0.2768,
+      "step": 3454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887425789764447e-05,
+      "loss": 0.1503,
+      "step": 3455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8873608372878405e-05,
+      "loss": 1.3673,
+      "step": 3456
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887295866437169e-05,
+      "loss": 0.7006,
+      "step": 3457
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887230877212844e-05,
+      "loss": 1.0991,
+      "step": 3458
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8871658696152804e-05,
+      "loss": 0.5982,
+      "step": 3459
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887100843644891e-05,
+      "loss": 0.5703,
+      "step": 3460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887035799302091e-05,
+      "loss": 1.1949,
+      "step": 3461
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.886970736587293e-05,
+      "loss": 0.3864,
+      "step": 3462
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.886905655500912e-05,
+      "loss": 0.9145,
+      "step": 3463
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8868405560433604e-05,
+      "loss": 1.3092,
+      "step": 3464
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.886775438215055e-05,
+      "loss": 1.4215,
+      "step": 3465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.886710302016408e-05,
+      "loss": 0.4753,
+      "step": 3466
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8866451474478355e-05,
+      "loss": 1.5962,
+      "step": 3467
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8865799745097504e-05,
+      "loss": 1.1006,
+      "step": 3468
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8865147832025694e-05,
+      "loss": 1.0095,
+      "step": 3469
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.886449573526706e-05,
+      "loss": 1.5328,
+      "step": 3470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8863843454825754e-05,
+      "loss": 0.705,
+      "step": 3471
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.886319099070594e-05,
+      "loss": 0.6299,
+      "step": 3472
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8862538342911746e-05,
+      "loss": 0.9109,
+      "step": 3473
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.886188551144734e-05,
+      "loss": 0.8463,
+      "step": 3474
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.886123249631688e-05,
+      "loss": 1.3161,
+      "step": 3475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.886057929752452e-05,
+      "loss": 0.2994,
+      "step": 3476
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8859925915074396e-05,
+      "loss": 1.3466,
+      "step": 3477
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8859272348970694e-05,
+      "loss": 1.375,
+      "step": 3478
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8858618599217556e-05,
+      "loss": 1.2225,
+      "step": 3479
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.885796466581915e-05,
+      "loss": 1.042,
+      "step": 3480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.885731054877964e-05,
+      "loss": 0.4315,
+      "step": 3481
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8856656248103175e-05,
+      "loss": 1.1288,
+      "step": 3482
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.885600176379393e-05,
+      "loss": 0.5521,
+      "step": 3483
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.885534709585607e-05,
+      "loss": 1.3983,
+      "step": 3484
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.885469224429376e-05,
+      "loss": 1.2684,
+      "step": 3485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.885403720911116e-05,
+      "loss": 0.4875,
+      "step": 3486
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.885338199031244e-05,
+      "loss": 1.3461,
+      "step": 3487
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.885272658790179e-05,
+      "loss": 0.4987,
+      "step": 3488
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8852071001883346e-05,
+      "loss": 1.1231,
+      "step": 3489
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.885141523226131e-05,
+      "loss": 0.7517,
+      "step": 3490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8850759279039836e-05,
+      "loss": 1.6189,
+      "step": 3491
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.88501031422231e-05,
+      "loss": 1.3794,
+      "step": 3492
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.88494468218153e-05,
+      "loss": 0.6777,
+      "step": 3493
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.884879031782058e-05,
+      "loss": 0.4401,
+      "step": 3494
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.884813363024314e-05,
+      "loss": 0.2282,
+      "step": 3495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.884747675908715e-05,
+      "loss": 1.2768,
+      "step": 3496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884681970435679e-05,
+      "loss": 0.972,
+      "step": 3497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884616246605625e-05,
+      "loss": 1.2212,
+      "step": 3498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.88455050441897e-05,
+      "loss": 1.0517,
+      "step": 3499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884484743876134e-05,
+      "loss": 0.3974,
+      "step": 3500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884418964977532e-05,
+      "loss": 1.1341,
+      "step": 3501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8843531677235874e-05,
+      "loss": 0.5975,
+      "step": 3502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8842873521147156e-05,
+      "loss": 0.3165,
+      "step": 3503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8842215181513365e-05,
+      "loss": 0.3344,
+      "step": 3504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884155665833868e-05,
+      "loss": 0.6578,
+      "step": 3505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884089795162731e-05,
+      "loss": 0.268,
+      "step": 3506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884023906138344e-05,
+      "loss": 1.1272,
+      "step": 3507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8839579987611245e-05,
+      "loss": 1.0217,
+      "step": 3508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.883892073031495e-05,
+      "loss": 0.9645,
+      "step": 3509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8838261289498726e-05,
+      "loss": 1.587,
+      "step": 3510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8837601665166784e-05,
+      "loss": 0.5394,
+      "step": 3511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.883694185732331e-05,
+      "loss": 0.7039,
+      "step": 3512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.883628186597251e-05,
+      "loss": 0.7427,
+      "step": 3513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8835621691118577e-05,
+      "loss": 0.5334,
+      "step": 3514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8834961332765724e-05,
+      "loss": 0.5265,
+      "step": 3515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8834300790918146e-05,
+      "loss": 1.021,
+      "step": 3516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.883364006558005e-05,
+      "loss": 1.9651,
+      "step": 3517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8832979156755624e-05,
+      "loss": 0.2336,
+      "step": 3518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8832318064449095e-05,
+      "loss": 0.7351,
+      "step": 3519
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.883165678866467e-05,
+      "loss": 1.213,
+      "step": 3520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.883099532940654e-05,
+      "loss": 1.4893,
+      "step": 3521
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8830333686678926e-05,
+      "loss": 0.7263,
+      "step": 3522
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882967186048604e-05,
+      "loss": 1.2321,
+      "step": 3523
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882900985083208e-05,
+      "loss": 0.2234,
+      "step": 3524
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8828347657721276e-05,
+      "loss": 0.9439,
+      "step": 3525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882768528115783e-05,
+      "loss": 1.1727,
+      "step": 3526
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882702272114596e-05,
+      "loss": 1.3027,
+      "step": 3527
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882635997768989e-05,
+      "loss": 1.285,
+      "step": 3528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882569705079383e-05,
+      "loss": 1.7336,
+      "step": 3529
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8825033940462e-05,
+      "loss": 0.7748,
+      "step": 3530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882437064669862e-05,
+      "loss": 0.8646,
+      "step": 3531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8823707169507904e-05,
+      "loss": 1.2713,
+      "step": 3532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882304350889409e-05,
+      "loss": 1.1937,
+      "step": 3533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8822379664861385e-05,
+      "loss": 1.4386,
+      "step": 3534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8821715637414025e-05,
+      "loss": 1.7115,
+      "step": 3535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882105142655622e-05,
+      "loss": 1.3327,
+      "step": 3536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8820387032292224e-05,
+      "loss": 0.2965,
+      "step": 3537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.881972245462624e-05,
+      "loss": 1.668,
+      "step": 3538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8819057693562507e-05,
+      "loss": 0.6563,
+      "step": 3539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8818392749105256e-05,
+      "loss": 1.3599,
+      "step": 3540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.881772762125872e-05,
+      "loss": 0.4142,
+      "step": 3541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8817062310027116e-05,
+      "loss": 1.4215,
+      "step": 3542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.88163968154147e-05,
+      "loss": 1.2715,
+      "step": 3543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.881573113742569e-05,
+      "loss": 1.2776,
+      "step": 3544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.881506527606433e-05,
+      "loss": 1.1619,
+      "step": 3545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.881439923133487e-05,
+      "loss": 0.3802,
+      "step": 3546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8813733003241527e-05,
+      "loss": 0.6063,
+      "step": 3547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.881306659178854e-05,
+      "loss": 0.5244,
+      "step": 3548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8812399996980164e-05,
+      "loss": 0.8083,
+      "step": 3549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8811733218820637e-05,
+      "loss": 0.4158,
+      "step": 3550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.88110662573142e-05,
+      "loss": 1.8306,
+      "step": 3551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.88103991124651e-05,
+      "loss": 1.1793,
+      "step": 3552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.880973178427757e-05,
+      "loss": 1.2929,
+      "step": 3553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.880906427275587e-05,
+      "loss": 0.8139,
+      "step": 3554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.880839657790425e-05,
+      "loss": 1.5976,
+      "step": 3555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.880772869972695e-05,
+      "loss": 0.8458,
+      "step": 3556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.880706063822823e-05,
+      "loss": 1.1844,
+      "step": 3557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.880639239341233e-05,
+      "loss": 0.9646,
+      "step": 3558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8805723965283505e-05,
+      "loss": 2.0395,
+      "step": 3559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8805055353846015e-05,
+      "loss": 1.2708,
+      "step": 3560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.880438655910411e-05,
+      "loss": 0.7165,
+      "step": 3561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.880371758106204e-05,
+      "loss": 1.1802,
+      "step": 3562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8803048419724074e-05,
+      "loss": 0.9779,
+      "step": 3563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8802379075094465e-05,
+      "loss": 1.7173,
+      "step": 3564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.880170954717747e-05,
+      "loss": 1.3611,
+      "step": 3565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8801039835977366e-05,
+      "loss": 1.0277,
+      "step": 3566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.880036994149838e-05,
+      "loss": 0.7042,
+      "step": 3567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.879969986374481e-05,
+      "loss": 0.3993,
+      "step": 3568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8799029602720907e-05,
+      "loss": 1.4045,
+      "step": 3569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.879835915843093e-05,
+      "loss": 0.6966,
+      "step": 3570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.879768853087915e-05,
+      "loss": 1.7418,
+      "step": 3571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.879701772006983e-05,
+      "loss": 0.6589,
+      "step": 3572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.879634672600725e-05,
+      "loss": 0.6995,
+      "step": 3573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.879567554869569e-05,
+      "loss": 0.8704,
+      "step": 3574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.879500418813937e-05,
+      "loss": 0.7963,
+      "step": 3575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.879433264434263e-05,
+      "loss": 2.1028,
+      "step": 3576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.87936609173097e-05,
+      "loss": 1.0323,
+      "step": 3577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.879298900704486e-05,
+      "loss": 0.7014,
+      "step": 3578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8792316913552396e-05,
+      "loss": 0.5964,
+      "step": 3579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.879164463683659e-05,
+      "loss": 0.5055,
+      "step": 3580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.879097217690169e-05,
+      "loss": 0.7304,
+      "step": 3581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8790299533752e-05,
+      "loss": 0.5561,
+      "step": 3582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.878962670739181e-05,
+      "loss": 0.6476,
+      "step": 3583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8788953697825384e-05,
+      "loss": 0.4813,
+      "step": 3584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8788280505056994e-05,
+      "loss": 0.6544,
+      "step": 3585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.878760712909096e-05,
+      "loss": 1.7263,
+      "step": 3586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.878693356993153e-05,
+      "loss": 0.9668,
+      "step": 3587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.878625982758301e-05,
+      "loss": 0.4634,
+      "step": 3588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.878558590204968e-05,
+      "loss": 0.4708,
+      "step": 3589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.878491179333584e-05,
+      "loss": 1.2585,
+      "step": 3590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.878423750144577e-05,
+      "loss": 0.6445,
+      "step": 3591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.878356302638375e-05,
+      "loss": 1.2615,
+      "step": 3592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8782888368154105e-05,
+      "loss": 1.2658,
+      "step": 3593
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8782213526761086e-05,
+      "loss": 0.2323,
+      "step": 3594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.878153850220903e-05,
+      "loss": 1.3082,
+      "step": 3595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87808632945022e-05,
+      "loss": 0.5207,
+      "step": 3596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.878018790364491e-05,
+      "loss": 0.7276,
+      "step": 3597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877951232964145e-05,
+      "loss": 0.8394,
+      "step": 3598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877883657249612e-05,
+      "loss": 1.0189,
+      "step": 3599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8778160632213225e-05,
+      "loss": 1.4657,
+      "step": 3600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8777484508797066e-05,
+      "loss": 1.6905,
+      "step": 3601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8776808202251937e-05,
+      "loss": 1.842,
+      "step": 3602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877613171258215e-05,
+      "loss": 0.4138,
+      "step": 3603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877545503979202e-05,
+      "loss": 1.1951,
+      "step": 3604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877477818388582e-05,
+      "loss": 0.9745,
+      "step": 3605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877410114486789e-05,
+      "loss": 1.4838,
+      "step": 3606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877342392274252e-05,
+      "loss": 0.9232,
+      "step": 3607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877274651751404e-05,
+      "loss": 0.8079,
+      "step": 3608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877206892918673e-05,
+      "loss": 0.5593,
+      "step": 3609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877139115776493e-05,
+      "loss": 1.0786,
+      "step": 3610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877071320325294e-05,
+      "loss": 0.646,
+      "step": 3611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877003506565507e-05,
+      "loss": 0.7549,
+      "step": 3612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876935674497564e-05,
+      "loss": 0.3868,
+      "step": 3613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8768678241218976e-05,
+      "loss": 1.3986,
+      "step": 3614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876799955438939e-05,
+      "loss": 0.2583,
+      "step": 3615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876732068449119e-05,
+      "loss": 1.1882,
+      "step": 3616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876664163152871e-05,
+      "loss": 1.1835,
+      "step": 3617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876596239550627e-05,
+      "loss": 0.4136,
+      "step": 3618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876528297642818e-05,
+      "loss": 0.4532,
+      "step": 3619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8764603374298775e-05,
+      "loss": 0.8457,
+      "step": 3620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876392358912238e-05,
+      "loss": 0.4777,
+      "step": 3621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876324362090332e-05,
+      "loss": 0.376,
+      "step": 3622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8762563469645906e-05,
+      "loss": 0.9026,
+      "step": 3623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8761883135354494e-05,
+      "loss": 1.2817,
+      "step": 3624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876120261803339e-05,
+      "loss": 0.4474,
+      "step": 3625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876052191768694e-05,
+      "loss": 0.7016,
+      "step": 3626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875984103431946e-05,
+      "loss": 1.6526,
+      "step": 3627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875915996793531e-05,
+      "loss": 1.0542,
+      "step": 3628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875847871853879e-05,
+      "loss": 1.511,
+      "step": 3629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8757797286134255e-05,
+      "loss": 0.8639,
+      "step": 3630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875711567072605e-05,
+      "loss": 0.4637,
+      "step": 3631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875643387231849e-05,
+      "loss": 0.8728,
+      "step": 3632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8755751890915925e-05,
+      "loss": 0.9051,
+      "step": 3633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8755069726522694e-05,
+      "loss": 0.7845,
+      "step": 3634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875438737914314e-05,
+      "loss": 0.5266,
+      "step": 3635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875370484878161e-05,
+      "loss": 0.5504,
+      "step": 3636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875302213544242e-05,
+      "loss": 0.8169,
+      "step": 3637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8752339239129954e-05,
+      "loss": 0.3925,
+      "step": 3638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8751656159848536e-05,
+      "loss": 1.655,
+      "step": 3639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875097289760251e-05,
+      "loss": 0.6487,
+      "step": 3640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875028945239624e-05,
+      "loss": 0.5784,
+      "step": 3641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.874960582423406e-05,
+      "loss": 1.1099,
+      "step": 3642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8748922013120326e-05,
+      "loss": 0.1282,
+      "step": 3643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8748238019059385e-05,
+      "loss": 1.3003,
+      "step": 3644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.874755384205559e-05,
+      "loss": 0.3525,
+      "step": 3645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.874686948211331e-05,
+      "loss": 0.6549,
+      "step": 3646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.874618493923689e-05,
+      "loss": 0.3201,
+      "step": 3647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.874550021343067e-05,
+      "loss": 1.429,
+      "step": 3648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.874481530469902e-05,
+      "loss": 0.6597,
+      "step": 3649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8744130213046324e-05,
+      "loss": 1.4156,
+      "step": 3650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87434449384769e-05,
+      "loss": 1.4723,
+      "step": 3651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.874275948099513e-05,
+      "loss": 1.3907,
+      "step": 3652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.874207384060538e-05,
+      "loss": 1.3793,
+      "step": 3653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8741388017312e-05,
+      "loss": 0.9297,
+      "step": 3654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8740702011119364e-05,
+      "loss": 0.5218,
+      "step": 3655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8740015822031826e-05,
+      "loss": 0.9069,
+      "step": 3656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8739329450053766e-05,
+      "loss": 0.2779,
+      "step": 3657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8738642895189544e-05,
+      "loss": 1.1787,
+      "step": 3658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.873795615744353e-05,
+      "loss": 0.2867,
+      "step": 3659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8737269236820105e-05,
+      "loss": 1.5357,
+      "step": 3660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.873658213332362e-05,
+      "loss": 0.61,
+      "step": 3661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.873589484695847e-05,
+      "loss": 0.3386,
+      "step": 3662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8735207377729e-05,
+      "loss": 0.3407,
+      "step": 3663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.873451972563961e-05,
+      "loss": 0.5919,
+      "step": 3664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.873383189069468e-05,
+      "loss": 1.395,
+      "step": 3665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8733143872898555e-05,
+      "loss": 1.8252,
+      "step": 3666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8732455672255645e-05,
+      "loss": 0.2498,
+      "step": 3667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8731767288770305e-05,
+      "loss": 0.2169,
+      "step": 3668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.873107872244695e-05,
+      "loss": 1.2684,
+      "step": 3669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.873038997328992e-05,
+      "loss": 0.4098,
+      "step": 3670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.872970104130362e-05,
+      "loss": 0.8841,
+      "step": 3671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.872901192649245e-05,
+      "loss": 1.6248,
+      "step": 3672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.872832262886076e-05,
+      "loss": 0.9103,
+      "step": 3673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.872763314841295e-05,
+      "loss": 0.6577,
+      "step": 3674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.872694348515343e-05,
+      "loss": 0.9259,
+      "step": 3675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8726253639086554e-05,
+      "loss": 0.995,
+      "step": 3676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8725563610216736e-05,
+      "loss": 0.3373,
+      "step": 3677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.872487339854835e-05,
+      "loss": 0.362,
+      "step": 3678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87241830040858e-05,
+      "loss": 0.805,
+      "step": 3679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.872349242683349e-05,
+      "loss": 1.0607,
+      "step": 3680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8722801666795784e-05,
+      "loss": 0.9683,
+      "step": 3681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8722110723977104e-05,
+      "loss": 1.0269,
+      "step": 3682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8721419598381826e-05,
+      "loss": 0.5952,
+      "step": 3683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.872072829001437e-05,
+      "loss": 1.232,
+      "step": 3684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.872003679887913e-05,
+      "loss": 1.4078,
+      "step": 3685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87193451249805e-05,
+      "loss": 0.5056,
+      "step": 3686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8718653268322876e-05,
+      "loss": 0.9677,
+      "step": 3687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.871796122891067e-05,
+      "loss": 0.3723,
+      "step": 3688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.871726900674828e-05,
+      "loss": 0.6786,
+      "step": 3689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8716576601840125e-05,
+      "loss": 0.6834,
+      "step": 3690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8715884014190595e-05,
+      "loss": 1.4157,
+      "step": 3691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87151912438041e-05,
+      "loss": 0.6111,
+      "step": 3692
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8714498290685054e-05,
+      "loss": 0.6755,
+      "step": 3693
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.871380515483787e-05,
+      "loss": 0.4524,
+      "step": 3694
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8713111836266944e-05,
+      "loss": 1.3105,
+      "step": 3695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8712418334976695e-05,
+      "loss": 1.1404,
+      "step": 3696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8711724650971554e-05,
+      "loss": 0.6821,
+      "step": 3697
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.871103078425591e-05,
+      "loss": 0.478,
+      "step": 3698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.871033673483419e-05,
+      "loss": 1.6746,
+      "step": 3699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870964250271081e-05,
+      "loss": 1.072,
+      "step": 3700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870894808789018e-05,
+      "loss": 1.0828,
+      "step": 3701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870825349037673e-05,
+      "loss": 1.1821,
+      "step": 3702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870755871017488e-05,
+      "loss": 1.4565,
+      "step": 3703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8706863747289035e-05,
+      "loss": 0.2149,
+      "step": 3704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8706168601723643e-05,
+      "loss": 0.6362,
+      "step": 3705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8705473273483115e-05,
+      "loss": 0.7691,
+      "step": 3706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870477776257186e-05,
+      "loss": 0.8962,
+      "step": 3707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870408206899434e-05,
+      "loss": 0.611,
+      "step": 3708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8703386192754944e-05,
+      "loss": 0.3463,
+      "step": 3709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870269013385812e-05,
+      "loss": 0.2274,
+      "step": 3710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8701993892308304e-05,
+      "loss": 0.5887,
+      "step": 3711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870129746810992e-05,
+      "loss": 1.3637,
+      "step": 3712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870060086126739e-05,
+      "loss": 0.4533,
+      "step": 3713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8699904071785156e-05,
+      "loss": 2.1436,
+      "step": 3714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8699207099667644e-05,
+      "loss": 0.63,
+      "step": 3715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8698509944919305e-05,
+      "loss": 0.8836,
+      "step": 3716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8697812607544565e-05,
+      "loss": 0.2451,
+      "step": 3717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8697115087547853e-05,
+      "loss": 1.109,
+      "step": 3718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8696417384933624e-05,
+      "loss": 1.0371,
+      "step": 3719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8695719499706315e-05,
+      "loss": 1.0683,
+      "step": 3720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8695021431870364e-05,
+      "loss": 1.0357,
+      "step": 3721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.86943231814302e-05,
+      "loss": 0.8074,
+      "step": 3722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.869362474839029e-05,
+      "loss": 0.4524,
+      "step": 3723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8692926132755074e-05,
+      "loss": 0.5748,
+      "step": 3724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.869222733452898e-05,
+      "loss": 1.3749,
+      "step": 3725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.869152835371647e-05,
+      "loss": 1.3399,
+      "step": 3726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.869082919032199e-05,
+      "loss": 0.8184,
+      "step": 3727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.869012984434998e-05,
+      "loss": 0.9761,
+      "step": 3728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.868943031580491e-05,
+      "loss": 1.1398,
+      "step": 3729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.868873060469121e-05,
+      "loss": 1.0444,
+      "step": 3730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.868803071101335e-05,
+      "loss": 0.5229,
+      "step": 3731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.868733063477577e-05,
+      "loss": 1.3627,
+      "step": 3732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.868663037598293e-05,
+      "loss": 0.982,
+      "step": 3733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8685929934639294e-05,
+      "loss": 0.7201,
+      "step": 3734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.868522931074931e-05,
+      "loss": 0.9566,
+      "step": 3735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8684528504317434e-05,
+      "loss": 0.2415,
+      "step": 3736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.868382751534813e-05,
+      "loss": 0.6577,
+      "step": 3737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8683126343845874e-05,
+      "loss": 0.3232,
+      "step": 3738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.86824249898151e-05,
+      "loss": 1.4851,
+      "step": 3739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8681723453260294e-05,
+      "loss": 0.4136,
+      "step": 3740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.86810217341859e-05,
+      "loss": 1.6428,
+      "step": 3741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8680319832596395e-05,
+      "loss": 1.0078,
+      "step": 3742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8679617748496256e-05,
+      "loss": 1.1266,
+      "step": 3743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.867891548188993e-05,
+      "loss": 0.6882,
+      "step": 3744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.86782130327819e-05,
+      "loss": 0.395,
+      "step": 3745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.867751040117664e-05,
+      "loss": 1.5079,
+      "step": 3746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8676807587078596e-05,
+      "loss": 0.2764,
+      "step": 3747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8676104590492265e-05,
+      "loss": 1.143,
+      "step": 3748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8675401411422124e-05,
+      "loss": 0.539,
+      "step": 3749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8674698049872626e-05,
+      "loss": 0.2572,
+      "step": 3750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.867399450584826e-05,
+      "loss": 0.4065,
+      "step": 3751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8673290779353496e-05,
+      "loss": 0.2989,
+      "step": 3752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.867258687039282e-05,
+      "loss": 1.1238,
+      "step": 3753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.867188277897072e-05,
+      "loss": 1.0109,
+      "step": 3754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.867117850509165e-05,
+      "loss": 0.9305,
+      "step": 3755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.867047404876011e-05,
+      "loss": 0.6781,
+      "step": 3756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.866976940998059e-05,
+      "loss": 0.93,
+      "step": 3757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.866906458875756e-05,
+      "loss": 0.5819,
+      "step": 3758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.86683595850955e-05,
+      "loss": 0.8689,
+      "step": 3759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.866765439899892e-05,
+      "loss": 0.5258,
+      "step": 3760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.866694903047228e-05,
+      "loss": 1.3705,
+      "step": 3761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.866624347952009e-05,
+      "loss": 0.4102,
+      "step": 3762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8665537746146834e-05,
+      "loss": 1.4762,
+      "step": 3763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8664831830356996e-05,
+      "loss": 1.1074,
+      "step": 3764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.866412573215507e-05,
+      "loss": 1.4841,
+      "step": 3765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8663419451545556e-05,
+      "loss": 1.0682,
+      "step": 3766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.866271298853294e-05,
+      "loss": 0.8948,
+      "step": 3767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.866200634312173e-05,
+      "loss": 0.8067,
+      "step": 3768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.866129951531641e-05,
+      "loss": 0.5117,
+      "step": 3769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.866059250512149e-05,
+      "loss": 1.3818,
+      "step": 3770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.865988531254145e-05,
+      "loss": 1.2533,
+      "step": 3771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.865917793758081e-05,
+      "loss": 1.3202,
+      "step": 3772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8658470380244066e-05,
+      "loss": 0.741,
+      "step": 3773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.865776264053572e-05,
+      "loss": 1.4049,
+      "step": 3774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.865705471846026e-05,
+      "loss": 1.0439,
+      "step": 3775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.865634661402222e-05,
+      "loss": 0.505,
+      "step": 3776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.865563832722609e-05,
+      "loss": 0.9405,
+      "step": 3777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.865492985807638e-05,
+      "loss": 1.1547,
+      "step": 3778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.865422120657759e-05,
+      "loss": 0.4077,
+      "step": 3779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.865351237273424e-05,
+      "loss": 0.8069,
+      "step": 3780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8652803356550845e-05,
+      "loss": 0.679,
+      "step": 3781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.86520941580319e-05,
+      "loss": 1.0285,
+      "step": 3782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.865138477718194e-05,
+      "loss": 0.3399,
+      "step": 3783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.865067521400545e-05,
+      "loss": 1.1662,
+      "step": 3784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8649965468506974e-05,
+      "loss": 0.3791,
+      "step": 3785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8649255540691005e-05,
+      "loss": 1.4079,
+      "step": 3786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.864854543056208e-05,
+      "loss": 0.2902,
+      "step": 3787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.864783513812472e-05,
+      "loss": 0.5728,
+      "step": 3788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.864712466338342e-05,
+      "loss": 0.6557,
+      "step": 3789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.864641400634272e-05,
+      "loss": 1.0501,
+      "step": 3790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.864570316700714e-05,
+      "loss": 1.1548,
+      "step": 3791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.864499214538121e-05,
+      "loss": 0.7386,
+      "step": 3792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.864428094146944e-05,
+      "loss": 0.4735,
+      "step": 3793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.864356955527637e-05,
+      "loss": 0.3725,
+      "step": 3794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8642857986806505e-05,
+      "loss": 0.4199,
+      "step": 3795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8642146236064394e-05,
+      "loss": 0.7338,
+      "step": 3796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.864143430305457e-05,
+      "loss": 0.7488,
+      "step": 3797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.864072218778154e-05,
+      "loss": 1.6355,
+      "step": 3798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.864000989024986e-05,
+      "loss": 0.5452,
+      "step": 3799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863929741046405e-05,
+      "loss": 0.9438,
+      "step": 3800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863858474842864e-05,
+      "loss": 1.1654,
+      "step": 3801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863787190414818e-05,
+      "loss": 1.0736,
+      "step": 3802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.86371588776272e-05,
+      "loss": 1.0099,
+      "step": 3803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863644566887023e-05,
+      "loss": 0.5217,
+      "step": 3804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863573227788181e-05,
+      "loss": 0.8162,
+      "step": 3805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863501870466648e-05,
+      "loss": 1.9419,
+      "step": 3806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863430494922879e-05,
+      "loss": 1.2912,
+      "step": 3807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863359101157328e-05,
+      "loss": 0.2574,
+      "step": 3808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863287689170448e-05,
+      "loss": 1.2719,
+      "step": 3809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863216258962695e-05,
+      "loss": 0.1854,
+      "step": 3810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863144810534524e-05,
+      "loss": 0.4131,
+      "step": 3811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8630733438863875e-05,
+      "loss": 0.7183,
+      "step": 3812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8630018590187406e-05,
+      "loss": 0.8501,
+      "step": 3813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.86293035593204e-05,
+      "loss": 1.4531,
+      "step": 3814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8628588346267404e-05,
+      "loss": 0.5166,
+      "step": 3815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.862787295103295e-05,
+      "loss": 0.6888,
+      "step": 3816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8627157373621614e-05,
+      "loss": 1.2378,
+      "step": 3817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.862644161403793e-05,
+      "loss": 1.5222,
+      "step": 3818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8625725672286464e-05,
+      "loss": 1.3888,
+      "step": 3819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.862500954837177e-05,
+      "loss": 1.0282,
+      "step": 3820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8624293242298404e-05,
+      "loss": 1.0886,
+      "step": 3821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.862357675407093e-05,
+      "loss": 0.4746,
+      "step": 3822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.86228600836939e-05,
+      "loss": 1.2721,
+      "step": 3823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.862214323117187e-05,
+      "loss": 0.9985,
+      "step": 3824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.862142619650942e-05,
+      "loss": 0.8326,
+      "step": 3825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.862070897971109e-05,
+      "loss": 1.0627,
+      "step": 3826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861999158078146e-05,
+      "loss": 1.221,
+      "step": 3827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.86192739997251e-05,
+      "loss": 0.2158,
+      "step": 3828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8618556236546556e-05,
+      "loss": 0.9124,
+      "step": 3829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861783829125042e-05,
+      "loss": 1.0485,
+      "step": 3830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861712016384124e-05,
+      "loss": 1.5361,
+      "step": 3831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861640185432358e-05,
+      "loss": 0.7173,
+      "step": 3832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8615683362702044e-05,
+      "loss": 0.796,
+      "step": 3833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861496468898118e-05,
+      "loss": 0.9636,
+      "step": 3834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861424583316556e-05,
+      "loss": 0.5986,
+      "step": 3835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8613526795259774e-05,
+      "loss": 0.9335,
+      "step": 3836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861280757526839e-05,
+      "loss": 1.5167,
+      "step": 3837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861208817319598e-05,
+      "loss": 1.4352,
+      "step": 3838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861136858904713e-05,
+      "loss": 1.7685,
+      "step": 3839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861064882282641e-05,
+      "loss": 0.4209,
+      "step": 3840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.86099288745384e-05,
+      "loss": 0.888,
+      "step": 3841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860920874418769e-05,
+      "loss": 1.3025,
+      "step": 3842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860848843177886e-05,
+      "loss": 0.9764,
+      "step": 3843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860776793731649e-05,
+      "loss": 0.9002,
+      "step": 3844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860704726080517e-05,
+      "loss": 1.3592,
+      "step": 3845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860632640224948e-05,
+      "loss": 1.0964,
+      "step": 3846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860560536165402e-05,
+      "loss": 1.5486,
+      "step": 3847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860488413902336e-05,
+      "loss": 0.7009,
+      "step": 3848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8604162734362115e-05,
+      "loss": 0.7518,
+      "step": 3849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860344114767485e-05,
+      "loss": 0.5252,
+      "step": 3850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860271937896616e-05,
+      "loss": 1.3633,
+      "step": 3851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860199742824065e-05,
+      "loss": 0.7052,
+      "step": 3852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860127529550291e-05,
+      "loss": 0.7411,
+      "step": 3853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860055298075754e-05,
+      "loss": 1.3901,
+      "step": 3854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8599830484009124e-05,
+      "loss": 0.1513,
+      "step": 3855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.859910780526227e-05,
+      "loss": 1.7545,
+      "step": 3856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.859838494452157e-05,
+      "loss": 0.5164,
+      "step": 3857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8597661901791625e-05,
+      "loss": 1.2791,
+      "step": 3858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8596938677077045e-05,
+      "loss": 1.433,
+      "step": 3859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8596215270382425e-05,
+      "loss": 1.1559,
+      "step": 3860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.859549168171237e-05,
+      "loss": 1.1875,
+      "step": 3861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8594767911071484e-05,
+      "loss": 0.3293,
+      "step": 3862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.859404395846437e-05,
+      "loss": 0.3274,
+      "step": 3863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.859331982389564e-05,
+      "loss": 1.0109,
+      "step": 3864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.85925955073699e-05,
+      "loss": 0.9844,
+      "step": 3865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8591871008891763e-05,
+      "loss": 1.2243,
+      "step": 3866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8591146328465826e-05,
+      "loss": 1.0069,
+      "step": 3867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8590421466096716e-05,
+      "loss": 1.384,
+      "step": 3868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.858969642178904e-05,
+      "loss": 0.6034,
+      "step": 3869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.858897119554742e-05,
+      "loss": 1.0258,
+      "step": 3870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.858824578737645e-05,
+      "loss": 0.8122,
+      "step": 3871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8587520197280766e-05,
+      "loss": 0.8688,
+      "step": 3872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.858679442526498e-05,
+      "loss": 1.4872,
+      "step": 3873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.858606847133371e-05,
+      "loss": 1.5541,
+      "step": 3874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8585342335491574e-05,
+      "loss": 0.9364,
+      "step": 3875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8584616017743184e-05,
+      "loss": 0.7491,
+      "step": 3876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.858388951809319e-05,
+      "loss": 1.5177,
+      "step": 3877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8583162836546186e-05,
+      "loss": 0.5462,
+      "step": 3878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.85824359731068e-05,
+      "loss": 1.1993,
+      "step": 3879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8581708927779674e-05,
+      "loss": 0.6877,
+      "step": 3880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8580981700569426e-05,
+      "loss": 0.88,
+      "step": 3881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8580254291480686e-05,
+      "loss": 0.354,
+      "step": 3882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.857952670051807e-05,
+      "loss": 0.192,
+      "step": 3883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.857879892768623e-05,
+      "loss": 1.3075,
+      "step": 3884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8578070972989774e-05,
+      "loss": 0.2202,
+      "step": 3885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8577342836433354e-05,
+      "loss": 0.8967,
+      "step": 3886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8576614518021596e-05,
+      "loss": 0.8661,
+      "step": 3887
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.857588601775913e-05,
+      "loss": 0.6478,
+      "step": 3888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.857515733565059e-05,
+      "loss": 0.9476,
+      "step": 3889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.857442847170063e-05,
+      "loss": 0.6176,
+      "step": 3890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.857369942591388e-05,
+      "loss": 0.4338,
+      "step": 3891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.857297019829497e-05,
+      "loss": 1.361,
+      "step": 3892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8572240788848544e-05,
+      "loss": 0.9187,
+      "step": 3893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8571511197579254e-05,
+      "loss": 0.4831,
+      "step": 3894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8570781424491734e-05,
+      "loss": 0.5339,
+      "step": 3895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.857005146959063e-05,
+      "loss": 0.9001,
+      "step": 3896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856932133288059e-05,
+      "loss": 0.7823,
+      "step": 3897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8568591014366244e-05,
+      "loss": 1.3509,
+      "step": 3898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8567860514052264e-05,
+      "loss": 1.0888,
+      "step": 3899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856712983194329e-05,
+      "loss": 0.7027,
+      "step": 3900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856639896804396e-05,
+      "loss": 1.2971,
+      "step": 3901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856566792235895e-05,
+      "loss": 1.3293,
+      "step": 3902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856493669489288e-05,
+      "loss": 0.6546,
+      "step": 3903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856420528565043e-05,
+      "loss": 0.2241,
+      "step": 3904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856347369463624e-05,
+      "loss": 0.4564,
+      "step": 3905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8562741921854965e-05,
+      "loss": 0.3811,
+      "step": 3906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856200996731127e-05,
+      "loss": 0.7942,
+      "step": 3907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.85612778310098e-05,
+      "loss": 1.345,
+      "step": 3908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8560545512955244e-05,
+      "loss": 0.572,
+      "step": 3909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.855981301315223e-05,
+      "loss": 1.493,
+      "step": 3910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.855908033160543e-05,
+      "loss": 0.2382,
+      "step": 3911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.85583474683195e-05,
+      "loss": 0.2887,
+      "step": 3912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8557614423299127e-05,
+      "loss": 0.3698,
+      "step": 3913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8556881196548943e-05,
+      "loss": 1.6532,
+      "step": 3914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.855614778807365e-05,
+      "loss": 0.2811,
+      "step": 3915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.855541419787788e-05,
+      "loss": 0.4652,
+      "step": 3916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.855468042596632e-05,
+      "loss": 0.5276,
+      "step": 3917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.855394647234364e-05,
+      "loss": 1.4757,
+      "step": 3918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.85532123370145e-05,
+      "loss": 0.6666,
+      "step": 3919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8552478019983586e-05,
+      "loss": 1.3196,
+      "step": 3920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8551743521255566e-05,
+      "loss": 0.2956,
+      "step": 3921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.85510088408351e-05,
+      "loss": 0.7341,
+      "step": 3922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8550273978726886e-05,
+      "loss": 1.1432,
+      "step": 3923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.85495389349356e-05,
+      "loss": 1.4311,
+      "step": 3924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8548803709465885e-05,
+      "loss": 1.2239,
+      "step": 3925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.854806830232245e-05,
+      "loss": 1.1793,
+      "step": 3926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8547332713509975e-05,
+      "loss": 1.2584,
+      "step": 3927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8546596943033136e-05,
+      "loss": 0.6804,
+      "step": 3928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.854586099089661e-05,
+      "loss": 0.9044,
+      "step": 3929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8545124857105086e-05,
+      "loss": 0.8026,
+      "step": 3930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.854438854166325e-05,
+      "loss": 1.1375,
+      "step": 3931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.854365204457577e-05,
+      "loss": 0.2905,
+      "step": 3932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.854291536584735e-05,
+      "loss": 0.3244,
+      "step": 3933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.854217850548269e-05,
+      "loss": 1.9312,
+      "step": 3934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.854144146348645e-05,
+      "loss": 1.3267,
+      "step": 3935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8540704239863337e-05,
+      "loss": 0.6767,
+      "step": 3936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.853996683461804e-05,
+      "loss": 0.4931,
+      "step": 3937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8539229247755256e-05,
+      "loss": 0.8532,
+      "step": 3938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.853849147927967e-05,
+      "loss": 0.5892,
+      "step": 3939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.853775352919598e-05,
+      "loss": 0.505,
+      "step": 3940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.853701539750889e-05,
+      "loss": 0.9093,
+      "step": 3941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.853627708422308e-05,
+      "loss": 0.8243,
+      "step": 3942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8535538589343264e-05,
+      "loss": 0.8791,
+      "step": 3943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.853479991287414e-05,
+      "loss": 0.7056,
+      "step": 3944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.853406105482039e-05,
+      "loss": 0.9335,
+      "step": 3945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8533322015186736e-05,
+      "loss": 1.2047,
+      "step": 3946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.853258279397789e-05,
+      "loss": 0.3136,
+      "step": 3947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8531843391198526e-05,
+      "loss": 0.4756,
+      "step": 3948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.853110380685337e-05,
+      "loss": 1.3142,
+      "step": 3949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8530364040947124e-05,
+      "loss": 0.7319,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.852962409348449e-05,
+      "loss": 0.7475,
+      "step": 3951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8528883964470184e-05,
+      "loss": 0.7048,
+      "step": 3952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.852814365390892e-05,
+      "loss": 0.5577,
+      "step": 3953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.852740316180539e-05,
+      "loss": 0.6738,
+      "step": 3954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.852666248816433e-05,
+      "loss": 0.7314,
+      "step": 3955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.852592163299044e-05,
+      "loss": 1.2109,
+      "step": 3956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8525180596288425e-05,
+      "loss": 0.9447,
+      "step": 3957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8524439378063024e-05,
+      "loss": 0.5493,
+      "step": 3958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.852369797831894e-05,
+      "loss": 0.4364,
+      "step": 3959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.852295639706089e-05,
+      "loss": 0.4578,
+      "step": 3960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8522214634293594e-05,
+      "loss": 1.6854,
+      "step": 3961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.852147269002178e-05,
+      "loss": 1.1755,
+      "step": 3962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.852073056425016e-05,
+      "loss": 0.9071,
+      "step": 3963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.851998825698347e-05,
+      "loss": 1.4036,
+      "step": 3964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8519245768226415e-05,
+      "loss": 0.3117,
+      "step": 3965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.851850309798373e-05,
+      "loss": 0.6643,
+      "step": 3966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8517760246260147e-05,
+      "loss": 0.4695,
+      "step": 3967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.851701721306037e-05,
+      "loss": 0.7685,
+      "step": 3968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.851627399838916e-05,
+      "loss": 1.2259,
+      "step": 3969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.851553060225123e-05,
+      "loss": 0.6485,
+      "step": 3970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.851478702465131e-05,
+      "loss": 0.9917,
+      "step": 3971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8514043265594126e-05,
+      "loss": 0.7956,
+      "step": 3972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8513299325084426e-05,
+      "loss": 1.1551,
+      "step": 3973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.851255520312693e-05,
+      "loss": 0.5984,
+      "step": 3974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.851181089972638e-05,
+      "loss": 1.1834,
+      "step": 3975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.851106641488751e-05,
+      "loss": 1.3141,
+      "step": 3976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.851032174861506e-05,
+      "loss": 0.5376,
+      "step": 3977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.850957690091377e-05,
+      "loss": 1.1534,
+      "step": 3978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.850883187178838e-05,
+      "loss": 1.2222,
+      "step": 3979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.850808666124363e-05,
+      "loss": 1.4715,
+      "step": 3980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8507341269284256e-05,
+      "loss": 0.8175,
+      "step": 3981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.850659569591502e-05,
+      "loss": 0.5986,
+      "step": 3982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8505849941140644e-05,
+      "loss": 0.8444,
+      "step": 3983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.850510400496588e-05,
+      "loss": 0.5559,
+      "step": 3984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.850435788739548e-05,
+      "loss": 1.3231,
+      "step": 3985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.850361158843419e-05,
+      "loss": 0.9525,
+      "step": 3986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.850286510808676e-05,
+      "loss": 0.3699,
+      "step": 3987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.850211844635794e-05,
+      "loss": 1.1132,
+      "step": 3988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8501371603252483e-05,
+      "loss": 0.9205,
+      "step": 3989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.850062457877514e-05,
+      "loss": 1.28,
+      "step": 3990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.849987737293065e-05,
+      "loss": 1.3916,
+      "step": 3991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.849912998572381e-05,
+      "loss": 0.682,
+      "step": 3992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.849838241715933e-05,
+      "loss": 0.2956,
+      "step": 3993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.849763466724199e-05,
+      "loss": 0.8347,
+      "step": 3994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8496886735976534e-05,
+      "loss": 0.2978,
+      "step": 3995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.849613862336774e-05,
+      "loss": 1.2135,
+      "step": 3996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.849539032942036e-05,
+      "loss": 0.7591,
+      "step": 3997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.849464185413916e-05,
+      "loss": 1.4763,
+      "step": 3998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8493893197528896e-05,
+      "loss": 1.5159,
+      "step": 3999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.849314435959433e-05,
+      "loss": 1.5775,
+      "step": 4000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8492395340340246e-05,
+      "loss": 0.8129,
+      "step": 4001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.849164613977138e-05,
+      "loss": 0.7379,
+      "step": 4002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.849089675789252e-05,
+      "loss": 0.6237,
+      "step": 4003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8490147194708446e-05,
+      "loss": 0.3609,
+      "step": 4004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8489397450223905e-05,
+      "loss": 0.735,
+      "step": 4005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.848864752444367e-05,
+      "loss": 1.3592,
+      "step": 4006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.848789741737253e-05,
+      "loss": 1.2962,
+      "step": 4007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.848714712901524e-05,
+      "loss": 1.0638,
+      "step": 4008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.848639665937659e-05,
+      "loss": 1.0956,
+      "step": 4009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8485646008461345e-05,
+      "loss": 0.7366,
+      "step": 4010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.848489517627429e-05,
+      "loss": 0.7213,
+      "step": 4011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.848414416282019e-05,
+      "loss": 1.3403,
+      "step": 4012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.848339296810383e-05,
+      "loss": 0.5295,
+      "step": 4013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.848264159213e-05,
+      "loss": 0.8344,
+      "step": 4014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8481890034903476e-05,
+      "loss": 0.8821,
+      "step": 4015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8481138296429035e-05,
+      "loss": 0.8297,
+      "step": 4016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.848038637671147e-05,
+      "loss": 0.6218,
+      "step": 4017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8479634275755555e-05,
+      "loss": 0.6468,
+      "step": 4018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8478881993566084e-05,
+      "loss": 0.9067,
+      "step": 4019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8478129530147836e-05,
+      "loss": 0.451,
+      "step": 4020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8477376885505615e-05,
+      "loss": 0.8239,
+      "step": 4021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.84766240596442e-05,
+      "loss": 0.3918,
+      "step": 4022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.847587105256837e-05,
+      "loss": 0.5075,
+      "step": 4023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8475117864282934e-05,
+      "loss": 0.6064,
+      "step": 4024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.847436449479269e-05,
+      "loss": 0.9601,
+      "step": 4025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.847361094410242e-05,
+      "loss": 0.9824,
+      "step": 4026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.847285721221691e-05,
+      "loss": 1.1474,
+      "step": 4027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.847210329914098e-05,
+      "loss": 1.4379,
+      "step": 4028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.847134920487942e-05,
+      "loss": 0.49,
+      "step": 4029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.847059492943702e-05,
+      "loss": 0.5068,
+      "step": 4030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846984047281857e-05,
+      "loss": 0.3126,
+      "step": 4031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.84690858350289e-05,
+      "loss": 1.543,
+      "step": 4032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8468331016072805e-05,
+      "loss": 1.5415,
+      "step": 4033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846757601595507e-05,
+      "loss": 1.0241,
+      "step": 4034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846682083468051e-05,
+      "loss": 0.3624,
+      "step": 4035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846606547225393e-05,
+      "loss": 1.074,
+      "step": 4036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846530992868015e-05,
+      "loss": 0.5837,
+      "step": 4037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846455420396396e-05,
+      "loss": 0.369,
+      "step": 4038
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846379829811017e-05,
+      "loss": 1.031,
+      "step": 4039
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.84630422111236e-05,
+      "loss": 2.1472,
+      "step": 4040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846228594300906e-05,
+      "loss": 1.2481,
+      "step": 4041
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846152949377136e-05,
+      "loss": 1.023,
+      "step": 4042
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846077286341531e-05,
+      "loss": 1.1317,
+      "step": 4043
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846001605194573e-05,
+      "loss": 1.1744,
+      "step": 4044
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8459259059367434e-05,
+      "loss": 1.3358,
+      "step": 4045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.845850188568524e-05,
+      "loss": 1.3313,
+      "step": 4046
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8457744530903976e-05,
+      "loss": 0.6484,
+      "step": 4047
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8456986995028436e-05,
+      "loss": 1.4232,
+      "step": 4048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8456229278063465e-05,
+      "loss": 0.2406,
+      "step": 4049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.845547138001387e-05,
+      "loss": 1.502,
+      "step": 4050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8454713300884484e-05,
+      "loss": 0.8618,
+      "step": 4051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.845395504068013e-05,
+      "loss": 1.1782,
+      "step": 4052
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.845319659940563e-05,
+      "loss": 0.4912,
+      "step": 4053
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8452437977065804e-05,
+      "loss": 1.4948,
+      "step": 4054
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8451679173665495e-05,
+      "loss": 0.4744,
+      "step": 4055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.845092018920952e-05,
+      "loss": 0.5421,
+      "step": 4056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.845016102370271e-05,
+      "loss": 0.5288,
+      "step": 4057
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.84494016771499e-05,
+      "loss": 1.4919,
+      "step": 4058
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8448642149555915e-05,
+      "loss": 1.6738,
+      "step": 4059
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.844788244092561e-05,
+      "loss": 0.6233,
+      "step": 4060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8447122551263786e-05,
+      "loss": 1.7061,
+      "step": 4061
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.84463624805753e-05,
+      "loss": 0.7954,
+      "step": 4062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.844560222886499e-05,
+      "loss": 1.2033,
+      "step": 4063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.844484179613769e-05,
+      "loss": 0.7024,
+      "step": 4064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.844408118239823e-05,
+      "loss": 1.4189,
+      "step": 4065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8443320387651464e-05,
+      "loss": 0.7944,
+      "step": 4066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8442559411902225e-05,
+      "loss": 0.4602,
+      "step": 4067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8441798255155355e-05,
+      "loss": 0.7906,
+      "step": 4068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8441036917415703e-05,
+      "loss": 0.3257,
+      "step": 4069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.844027539868812e-05,
+      "loss": 1.8791,
+      "step": 4070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8439513698977427e-05,
+      "loss": 1.6566,
+      "step": 4071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8438751818288504e-05,
+      "loss": 0.8907,
+      "step": 4072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.843798975662617e-05,
+      "loss": 0.8768,
+      "step": 4073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.84372275139953e-05,
+      "loss": 0.9634,
+      "step": 4074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.843646509040072e-05,
+      "loss": 1.3436,
+      "step": 4075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.84357024858473e-05,
+      "loss": 1.5085,
+      "step": 4076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.843493970033989e-05,
+      "loss": 0.3009,
+      "step": 4077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8434176733883336e-05,
+      "loss": 1.1047,
+      "step": 4078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.843341358648249e-05,
+      "loss": 0.9482,
+      "step": 4079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.843265025814224e-05,
+      "loss": 0.607,
+      "step": 4080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.84318867488674e-05,
+      "loss": 1.0146,
+      "step": 4081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.843112305866285e-05,
+      "loss": 1.09,
+      "step": 4082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8430359187533464e-05,
+      "loss": 1.3043,
+      "step": 4083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.842959513548407e-05,
+      "loss": 0.5756,
+      "step": 4084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.842883090251956e-05,
+      "loss": 1.2676,
+      "step": 4085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.842806648864478e-05,
+      "loss": 0.9291,
+      "step": 4086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.84273018938646e-05,
+      "loss": 0.8726,
+      "step": 4087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8426537118183886e-05,
+      "loss": 0.8327,
+      "step": 4088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8425772161607505e-05,
+      "loss": 0.4033,
+      "step": 4089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.842500702414032e-05,
+      "loss": 1.1266,
+      "step": 4090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.84242417057872e-05,
+      "loss": 0.4242,
+      "step": 4091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8423476206553024e-05,
+      "loss": 1.5827,
+      "step": 4092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.842271052644266e-05,
+      "loss": 0.5406,
+      "step": 4093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.842194466546097e-05,
+      "loss": 1.2537,
+      "step": 4094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8421178623612844e-05,
+      "loss": 0.4328,
+      "step": 4095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.842041240090315e-05,
+      "loss": 1.5114,
+      "step": 4096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8419645997336764e-05,
+      "loss": 1.9024,
+      "step": 4097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.841887941291855e-05,
+      "loss": 1.171,
+      "step": 4098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.841811264765342e-05,
+      "loss": 0.7626,
+      "step": 4099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.841734570154622e-05,
+      "loss": 1.5871,
+      "step": 4100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.841657857460185e-05,
+      "loss": 2.0151,
+      "step": 4101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.841581126682517e-05,
+      "loss": 0.9318,
+      "step": 4102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.841504377822108e-05,
+      "loss": 0.822,
+      "step": 4103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.841427610879447e-05,
+      "loss": 0.4919,
+      "step": 4104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.84135082585502e-05,
+      "loss": 0.4683,
+      "step": 4105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.841274022749319e-05,
+      "loss": 1.4384,
+      "step": 4106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8411972015628306e-05,
+      "loss": 0.5501,
+      "step": 4107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.841120362296044e-05,
+      "loss": 1.0473,
+      "step": 4108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8410435049494474e-05,
+      "loss": 1.3487,
+      "step": 4109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.840966629523532e-05,
+      "loss": 0.8396,
+      "step": 4110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8408897360187846e-05,
+      "loss": 1.5856,
+      "step": 4111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.840812824435696e-05,
+      "loss": 0.5829,
+      "step": 4112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8407358947747554e-05,
+      "loss": 0.9163,
+      "step": 4113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8406589470364525e-05,
+      "loss": 1.4817,
+      "step": 4114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8405819812212766e-05,
+      "loss": 0.2666,
+      "step": 4115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8405049973297175e-05,
+      "loss": 0.7987,
+      "step": 4116
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.840427995362265e-05,
+      "loss": 1.1555,
+      "step": 4117
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.840350975319409e-05,
+      "loss": 0.8843,
+      "step": 4118
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8402739372016405e-05,
+      "loss": 0.5758,
+      "step": 4119
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.840196881009449e-05,
+      "loss": 0.2847,
+      "step": 4120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8401198067433256e-05,
+      "loss": 1.089,
+      "step": 4121
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.840042714403759e-05,
+      "loss": 0.3474,
+      "step": 4122
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.839965603991242e-05,
+      "loss": 0.7069,
+      "step": 4123
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.839888475506263e-05,
+      "loss": 0.4063,
+      "step": 4124
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.839811328949315e-05,
+      "loss": 0.6072,
+      "step": 4125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8397341643208885e-05,
+      "loss": 0.7783,
+      "step": 4126
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8396569816214735e-05,
+      "loss": 1.3911,
+      "step": 4127
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8395797808515614e-05,
+      "loss": 0.3689,
+      "step": 4128
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.839502562011645e-05,
+      "loss": 0.8781,
+      "step": 4129
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.839425325102213e-05,
+      "loss": 1.3232,
+      "step": 4130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.839348070123759e-05,
+      "loss": 0.4425,
+      "step": 4131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8392707970767744e-05,
+      "loss": 1.1528,
+      "step": 4132
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.83919350596175e-05,
+      "loss": 0.422,
+      "step": 4133
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.839116196779179e-05,
+      "loss": 1.1722,
+      "step": 4134
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8390388695295516e-05,
+      "loss": 0.9734,
+      "step": 4135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.838961524213361e-05,
+      "loss": 1.3144,
+      "step": 4136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.838884160831099e-05,
+      "loss": 1.2016,
+      "step": 4137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.838806779383259e-05,
+      "loss": 1.2265,
+      "step": 4138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8387293798703324e-05,
+      "loss": 1.0876,
+      "step": 4139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.838651962292813e-05,
+      "loss": 1.4529,
+      "step": 4140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8385745266511906e-05,
+      "loss": 1.5215,
+      "step": 4141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8384970729459606e-05,
+      "loss": 1.3575,
+      "step": 4142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.838419601177615e-05,
+      "loss": 1.1861,
+      "step": 4143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8383421113466465e-05,
+      "loss": 0.5589,
+      "step": 4144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.838264603453549e-05,
+      "loss": 0.7368,
+      "step": 4145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.838187077498815e-05,
+      "loss": 1.2384,
+      "step": 4146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8381095334829375e-05,
+      "loss": 1.5568,
+      "step": 4147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8380319714064115e-05,
+      "loss": 0.4862,
+      "step": 4148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8379543912697296e-05,
+      "loss": 0.366,
+      "step": 4149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.837876793073386e-05,
+      "loss": 1.1435,
+      "step": 4150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.837799176817873e-05,
+      "loss": 1.5426,
+      "step": 4151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8377215425036866e-05,
+      "loss": 1.0456,
+      "step": 4152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.837643890131319e-05,
+      "loss": 0.5984,
+      "step": 4153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8375662197012654e-05,
+      "loss": 0.7974,
+      "step": 4154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.83748853121402e-05,
+      "loss": 0.4154,
+      "step": 4155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.837410824670077e-05,
+      "loss": 0.5432,
+      "step": 4156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.83733310006993e-05,
+      "loss": 0.4187,
+      "step": 4157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.837255357414076e-05,
+      "loss": 1.2893,
+      "step": 4158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.837177596703008e-05,
+      "loss": 0.7774,
+      "step": 4159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8370998179372205e-05,
+      "loss": 0.2779,
+      "step": 4160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.837022021117209e-05,
+      "loss": 1.4558,
+      "step": 4161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.83694420624347e-05,
+      "loss": 1.2215,
+      "step": 4162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.836866373316496e-05,
+      "loss": 0.8183,
+      "step": 4163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.836788522336784e-05,
+      "loss": 0.4592,
+      "step": 4164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8367106533048285e-05,
+      "loss": 0.7108,
+      "step": 4165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.836632766221126e-05,
+      "loss": 1.289,
+      "step": 4166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.836554861086172e-05,
+      "loss": 1.0196,
+      "step": 4167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.836476937900461e-05,
+      "loss": 1.4813,
+      "step": 4168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8363989966644905e-05,
+      "loss": 1.0259,
+      "step": 4169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.836321037378755e-05,
+      "loss": 0.3535,
+      "step": 4170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.836243060043753e-05,
+      "loss": 0.8456,
+      "step": 4171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8361650646599775e-05,
+      "loss": 0.5973,
+      "step": 4172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.836087051227927e-05,
+      "loss": 0.5287,
+      "step": 4173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.836009019748097e-05,
+      "loss": 0.3003,
+      "step": 4174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8359309702209856e-05,
+      "loss": 0.9623,
+      "step": 4175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.835852902647088e-05,
+      "loss": 0.9523,
+      "step": 4176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.835774817026901e-05,
+      "loss": 0.7222,
+      "step": 4177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.835696713360922e-05,
+      "loss": 0.6019,
+      "step": 4178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8356185916496474e-05,
+      "loss": 0.4807,
+      "step": 4179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8355404518935756e-05,
+      "loss": 1.2053,
+      "step": 4180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.835462294093203e-05,
+      "loss": 1.2986,
+      "step": 4181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.835384118249027e-05,
+      "loss": 1.1645,
+      "step": 4182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.835305924361544e-05,
+      "loss": 0.761,
+      "step": 4183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.835227712431253e-05,
+      "loss": 0.7344,
+      "step": 4184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.835149482458653e-05,
+      "loss": 1.5296,
+      "step": 4185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.835071234444239e-05,
+      "loss": 0.0704,
+      "step": 4186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8349929683885106e-05,
+      "loss": 0.3132,
+      "step": 4187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834914684291965e-05,
+      "loss": 1.417,
+      "step": 4188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834836382155101e-05,
+      "loss": 1.2959,
+      "step": 4189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834758061978416e-05,
+      "loss": 1.2256,
+      "step": 4190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.83467972376241e-05,
+      "loss": 0.488,
+      "step": 4191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834601367507581e-05,
+      "loss": 0.6859,
+      "step": 4192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834522993214426e-05,
+      "loss": 0.7648,
+      "step": 4193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834444600883446e-05,
+      "loss": 1.0991,
+      "step": 4194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834366190515139e-05,
+      "loss": 0.3456,
+      "step": 4195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834287762110003e-05,
+      "loss": 1.7089,
+      "step": 4196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8342093156685373e-05,
+      "loss": 1.4981,
+      "step": 4197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834130851191244e-05,
+      "loss": 0.6337,
+      "step": 4198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8340523686786183e-05,
+      "loss": 1.299,
+      "step": 4199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.833973868131162e-05,
+      "loss": 1.2893,
+      "step": 4200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.833895349549373e-05,
+      "loss": 0.7145,
+      "step": 4201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.833816812933753e-05,
+      "loss": 1.1853,
+      "step": 4202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.833738258284801e-05,
+      "loss": 0.6302,
+      "step": 4203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8336596856030174e-05,
+      "loss": 1.333,
+      "step": 4204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8335810948889e-05,
+      "loss": 1.8101,
+      "step": 4205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8335024861429514e-05,
+      "loss": 0.3582,
+      "step": 4206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.833423859365671e-05,
+      "loss": 0.6428,
+      "step": 4207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8333452145575584e-05,
+      "loss": 1.2129,
+      "step": 4208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.833266551719115e-05,
+      "loss": 1.4624,
+      "step": 4209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8331878708508416e-05,
+      "loss": 0.8217,
+      "step": 4210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8331091719532374e-05,
+      "loss": 1.4818,
+      "step": 4211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8330304550268046e-05,
+      "loss": 0.4478,
+      "step": 4212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8329517200720446e-05,
+      "loss": 1.5228,
+      "step": 4213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8328729670894555e-05,
+      "loss": 0.7577,
+      "step": 4214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832794196079542e-05,
+      "loss": 0.7379,
+      "step": 4215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832715407042804e-05,
+      "loss": 0.8559,
+      "step": 4216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832636599979743e-05,
+      "loss": 0.2582,
+      "step": 4217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832557774890859e-05,
+      "loss": 1.4149,
+      "step": 4218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832478931776656e-05,
+      "loss": 1.1833,
+      "step": 4219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832400070637634e-05,
+      "loss": 0.9916,
+      "step": 4220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832321191474296e-05,
+      "loss": 0.6812,
+      "step": 4221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8322422942871435e-05,
+      "loss": 0.5962,
+      "step": 4222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832163379076678e-05,
+      "loss": 0.5274,
+      "step": 4223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832084445843402e-05,
+      "loss": 0.0901,
+      "step": 4224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832005494587818e-05,
+      "loss": 1.2744,
+      "step": 4225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.831926525310429e-05,
+      "loss": 0.9169,
+      "step": 4226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8318475380117364e-05,
+      "loss": 0.427,
+      "step": 4227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8317685326922433e-05,
+      "loss": 0.7651,
+      "step": 4228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.831689509352453e-05,
+      "loss": 0.6317,
+      "step": 4229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8316104679928675e-05,
+      "loss": 0.3594,
+      "step": 4230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.83153140861399e-05,
+      "loss": 0.2327,
+      "step": 4231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.831452331216324e-05,
+      "loss": 0.8213,
+      "step": 4232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.831373235800373e-05,
+      "loss": 1.099,
+      "step": 4233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8312941223666396e-05,
+      "loss": 0.9793,
+      "step": 4234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.831214990915626e-05,
+      "loss": 0.7167,
+      "step": 4235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.831135841447839e-05,
+      "loss": 1.0419,
+      "step": 4236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.83105667396378e-05,
+      "loss": 0.8918,
+      "step": 4237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8309774884639534e-05,
+      "loss": 0.8127,
+      "step": 4238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.830898284948863e-05,
+      "loss": 1.3018,
+      "step": 4239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8308190634190127e-05,
+      "loss": 0.5224,
+      "step": 4240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8307398238749075e-05,
+      "loss": 1.1143,
+      "step": 4241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8306605663170496e-05,
+      "loss": 0.5994,
+      "step": 4242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8305812907459464e-05,
+      "loss": 0.4508,
+      "step": 4243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8305019971620994e-05,
+      "loss": 0.837,
+      "step": 4244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.830422685566015e-05,
+      "loss": 0.5042,
+      "step": 4245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.830343355958197e-05,
+      "loss": 1.2279,
+      "step": 4246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.830264008339151e-05,
+      "loss": 1.0839,
+      "step": 4247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.830184642709381e-05,
+      "loss": 0.6157,
+      "step": 4248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.830105259069393e-05,
+      "loss": 1.2016,
+      "step": 4249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.830025857419691e-05,
+      "loss": 0.962,
+      "step": 4250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.829946437760782e-05,
+      "loss": 0.6216,
+      "step": 4251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.82986700009317e-05,
+      "loss": 0.5361,
+      "step": 4252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.829787544417361e-05,
+      "loss": 1.1081,
+      "step": 4253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8297080707338607e-05,
+      "loss": 0.259,
+      "step": 4254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.829628579043175e-05,
+      "loss": 1.4929,
+      "step": 4255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.829549069345809e-05,
+      "loss": 0.5148,
+      "step": 4256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.829469541642269e-05,
+      "loss": 0.7714,
+      "step": 4257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8293899959330615e-05,
+      "loss": 0.4428,
+      "step": 4258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.829310432218693e-05,
+      "loss": 0.5385,
+      "step": 4259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.829230850499668e-05,
+      "loss": 0.7465,
+      "step": 4260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.829151250776494e-05,
+      "loss": 1.1664,
+      "step": 4261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.829071633049679e-05,
+      "loss": 0.6402,
+      "step": 4262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8289919973197276e-05,
+      "loss": 1.358,
+      "step": 4263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.828912343587148e-05,
+      "loss": 1.1341,
+      "step": 4264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8288326718524456e-05,
+      "loss": 1.4698,
+      "step": 4265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.828752982116129e-05,
+      "loss": 0.641,
+      "step": 4266
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.828673274378704e-05,
+      "loss": 1.2515,
+      "step": 4267
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.828593548640679e-05,
+      "loss": 0.595,
+      "step": 4268
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8285138049025595e-05,
+      "loss": 0.3685,
+      "step": 4269
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8284340431648555e-05,
+      "loss": 0.449,
+      "step": 4270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.828354263428072e-05,
+      "loss": 1.3685,
+      "step": 4271
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8282744656927194e-05,
+      "loss": 0.9106,
+      "step": 4272
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.828194649959302e-05,
+      "loss": 0.9129,
+      "step": 4273
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.828114816228332e-05,
+      "loss": 1.3988,
+      "step": 4274
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.828034964500314e-05,
+      "loss": 0.9929,
+      "step": 4275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.827955094775757e-05,
+      "loss": 1.911,
+      "step": 4276
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.827875207055171e-05,
+      "loss": 1.1147,
+      "step": 4277
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.827795301339061e-05,
+      "loss": 0.6811,
+      "step": 4278
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.827715377627939e-05,
+      "loss": 0.7264,
+      "step": 4279
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.827635435922312e-05,
+      "loss": 0.4307,
+      "step": 4280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.827555476222687e-05,
+      "loss": 0.5567,
+      "step": 4281
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.827475498529577e-05,
+      "loss": 1.2304,
+      "step": 4282
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8273955028434864e-05,
+      "loss": 0.9962,
+      "step": 4283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.827315489164928e-05,
+      "loss": 0.2936,
+      "step": 4284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.827235457494409e-05,
+      "loss": 0.796,
+      "step": 4285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8271554078324385e-05,
+      "loss": 0.3606,
+      "step": 4286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.827075340179527e-05,
+      "loss": 1.3008,
+      "step": 4287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8269952545361836e-05,
+      "loss": 0.9632,
+      "step": 4288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.826915150902917e-05,
+      "loss": 1.151,
+      "step": 4289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.826835029280239e-05,
+      "loss": 1.3181,
+      "step": 4290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.826754889668658e-05,
+      "loss": 0.5716,
+      "step": 4291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8266747320686834e-05,
+      "loss": 1.0764,
+      "step": 4292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.826594556480827e-05,
+      "loss": 0.5152,
+      "step": 4293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.826514362905598e-05,
+      "loss": 0.5841,
+      "step": 4294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.826434151343506e-05,
+      "loss": 0.9992,
+      "step": 4295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.826353921795064e-05,
+      "loss": 1.4805,
+      "step": 4296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.82627367426078e-05,
+      "loss": 1.2048,
+      "step": 4297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8261934087411644e-05,
+      "loss": 1.0214,
+      "step": 4298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.82611312523673e-05,
+      "loss": 0.4267,
+      "step": 4299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.826032823747987e-05,
+      "loss": 1.2248,
+      "step": 4300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.825952504275446e-05,
+      "loss": 0.9772,
+      "step": 4301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.825872166819618e-05,
+      "loss": 0.8259,
+      "step": 4302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.825791811381015e-05,
+      "loss": 1.1414,
+      "step": 4303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8257114379601474e-05,
+      "loss": 0.6575,
+      "step": 4304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8256310465575285e-05,
+      "loss": 0.9216,
+      "step": 4305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8255506371736666e-05,
+      "loss": 0.5151,
+      "step": 4306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.825470209809077e-05,
+      "loss": 0.544,
+      "step": 4307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8253897644642685e-05,
+      "loss": 1.3926,
+      "step": 4308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8253093011397554e-05,
+      "loss": 0.5925,
+      "step": 4309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.825228819836048e-05,
+      "loss": 0.3784,
+      "step": 4310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.82514832055366e-05,
+      "loss": 0.4083,
+      "step": 4311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.825067803293102e-05,
+      "loss": 1.8083,
+      "step": 4312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8249872680548875e-05,
+      "loss": 0.6315,
+      "step": 4313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.824906714839529e-05,
+      "loss": 0.5723,
+      "step": 4314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.824826143647538e-05,
+      "loss": 0.7026,
+      "step": 4315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.824745554479429e-05,
+      "loss": 1.2999,
+      "step": 4316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.824664947335713e-05,
+      "loss": 0.8583,
+      "step": 4317
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8245843222169046e-05,
+      "loss": 1.0799,
+      "step": 4318
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8245036791235154e-05,
+      "loss": 0.8851,
+      "step": 4319
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.824423018056059e-05,
+      "loss": 1.2272,
+      "step": 4320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.82434233901505e-05,
+      "loss": 0.4859,
+      "step": 4321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.824261642001e-05,
+      "loss": 1.4541,
+      "step": 4322
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8241809270144245e-05,
+      "loss": 1.0099,
+      "step": 4323
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.824100194055836e-05,
+      "loss": 0.579,
+      "step": 4324
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.824019443125746e-05,
+      "loss": 1.0271,
+      "step": 4325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8239386742246723e-05,
+      "loss": 0.633,
+      "step": 4326
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.823857887353127e-05,
+      "loss": 1.353,
+      "step": 4327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8237770825116245e-05,
+      "loss": 1.0469,
+      "step": 4328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.823696259700678e-05,
+      "loss": 0.8279,
+      "step": 4329
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8236154189208035e-05,
+      "loss": 0.9286,
+      "step": 4330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.823534560172515e-05,
+      "loss": 0.475,
+      "step": 4331
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.823453683456325e-05,
+      "loss": 0.1999,
+      "step": 4332
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.823372788772751e-05,
+      "loss": 1.0367,
+      "step": 4333
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8232918761223074e-05,
+      "loss": 0.3271,
+      "step": 4334
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.823210945505507e-05,
+      "loss": 0.7751,
+      "step": 4335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8231299969228675e-05,
+      "loss": 0.4809,
+      "step": 4336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8230490303749014e-05,
+      "loss": 0.4573,
+      "step": 4337
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8229680458621254e-05,
+      "loss": 0.5104,
+      "step": 4338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.822887043385055e-05,
+      "loss": 0.2644,
+      "step": 4339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.822806022944205e-05,
+      "loss": 0.3403,
+      "step": 4340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8227249845400916e-05,
+      "loss": 0.9038,
+      "step": 4341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.822643928173231e-05,
+      "loss": 0.6959,
+      "step": 4342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8225628538441365e-05,
+      "loss": 1.0042,
+      "step": 4343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.822481761553327e-05,
+      "loss": 1.5711,
+      "step": 4344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.822400651301317e-05,
+      "loss": 0.5293,
+      "step": 4345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.822319523088623e-05,
+      "loss": 0.9098,
+      "step": 4346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.822238376915761e-05,
+      "loss": 0.8033,
+      "step": 4347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.822157212783247e-05,
+      "loss": 0.9025,
+      "step": 4348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8220760306916e-05,
+      "loss": 1.2007,
+      "step": 4349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8219948306413335e-05,
+      "loss": 0.3532,
+      "step": 4350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.821913612632965e-05,
+      "loss": 1.5123,
+      "step": 4351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.821832376667013e-05,
+      "loss": 1.2188,
+      "step": 4352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.821751122743992e-05,
+      "loss": 0.3207,
+      "step": 4353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.82166985086442e-05,
+      "loss": 1.0521,
+      "step": 4354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8215885610288165e-05,
+      "loss": 0.8636,
+      "step": 4355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.821507253237695e-05,
+      "loss": 1.6131,
+      "step": 4356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.821425927491575e-05,
+      "loss": 0.5584,
+      "step": 4357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.821344583790974e-05,
+      "loss": 1.2205,
+      "step": 4358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.82126322213641e-05,
+      "loss": 0.4372,
+      "step": 4359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.821181842528399e-05,
+      "loss": 1.4569,
+      "step": 4360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8211004449674606e-05,
+      "loss": 0.3378,
+      "step": 4361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8210190294541125e-05,
+      "loss": 0.7785,
+      "step": 4362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8209375959888714e-05,
+      "loss": 0.7586,
+      "step": 4363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.820856144572257e-05,
+      "loss": 0.7355,
+      "step": 4364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.820774675204788e-05,
+      "loss": 0.2721,
+      "step": 4365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.820693187886981e-05,
+      "loss": 1.0751,
+      "step": 4366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.820611682619356e-05,
+      "loss": 0.7535,
+      "step": 4367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8205301594024314e-05,
+      "loss": 1.2515,
+      "step": 4368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.820448618236726e-05,
+      "loss": 0.4917,
+      "step": 4369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.820367059122758e-05,
+      "loss": 0.5659,
+      "step": 4370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.820285482061047e-05,
+      "loss": 0.6074,
+      "step": 4371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8202038870521123e-05,
+      "loss": 0.4681,
+      "step": 4372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.820122274096473e-05,
+      "loss": 1.4503,
+      "step": 4373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.820040643194649e-05,
+      "loss": 1.2674,
+      "step": 4374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.819958994347158e-05,
+      "loss": 1.0212,
+      "step": 4375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.819877327554521e-05,
+      "loss": 0.9742,
+      "step": 4376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.819795642817258e-05,
+      "loss": 0.3151,
+      "step": 4377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.819713940135888e-05,
+      "loss": 1.1802,
+      "step": 4378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8196322195109314e-05,
+      "loss": 1.1028,
+      "step": 4379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8195504809429075e-05,
+      "loss": 1.5855,
+      "step": 4380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.819468724432337e-05,
+      "loss": 0.4004,
+      "step": 4381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8193869499797395e-05,
+      "loss": 1.813,
+      "step": 4382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.819305157585637e-05,
+      "loss": 0.2971,
+      "step": 4383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.819223347250549e-05,
+      "loss": 1.514,
+      "step": 4384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.819141518974996e-05,
+      "loss": 0.4015,
+      "step": 4385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.819059672759498e-05,
+      "loss": 0.9506,
+      "step": 4386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.818977808604578e-05,
+      "loss": 1.0014,
+      "step": 4387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.818895926510754e-05,
+      "loss": 0.4924,
+      "step": 4388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8188140264785503e-05,
+      "loss": 0.4966,
+      "step": 4389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8187321085084854e-05,
+      "loss": 1.3572,
+      "step": 4390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8186501726010816e-05,
+      "loss": 0.4825,
+      "step": 4391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8185682187568614e-05,
+      "loss": 1.0268,
+      "step": 4392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.818486246976345e-05,
+      "loss": 1.0225,
+      "step": 4393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.818404257260054e-05,
+      "loss": 0.8759,
+      "step": 4394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8183222496085095e-05,
+      "loss": 0.4399,
+      "step": 4395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.818240224022236e-05,
+      "loss": 1.0998,
+      "step": 4396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.818158180501753e-05,
+      "loss": 0.4764,
+      "step": 4397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.818076119047584e-05,
+      "loss": 0.5198,
+      "step": 4398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.81799403966025e-05,
+      "loss": 0.4705,
+      "step": 4399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817911942340274e-05,
+      "loss": 1.0517,
+      "step": 4400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817829827088178e-05,
+      "loss": 0.5805,
+      "step": 4401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817747693904487e-05,
+      "loss": 0.81,
+      "step": 4402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817665542789719e-05,
+      "loss": 1.2291,
+      "step": 4403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817583373744401e-05,
+      "loss": 0.7772,
+      "step": 4404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817501186769053e-05,
+      "loss": 1.1513,
+      "step": 4405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817418981864201e-05,
+      "loss": 1.1218,
+      "step": 4406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817336759030365e-05,
+      "loss": 1.6565,
+      "step": 4407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.81725451826807e-05,
+      "loss": 0.6271,
+      "step": 4408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8171722595778395e-05,
+      "loss": 0.772,
+      "step": 4409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817089982960195e-05,
+      "loss": 1.0942,
+      "step": 4410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8170076884156634e-05,
+      "loss": 0.8823,
+      "step": 4411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816925375944765e-05,
+      "loss": 1.2732,
+      "step": 4412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8168430455480264e-05,
+      "loss": 0.3544,
+      "step": 4413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816760697225969e-05,
+      "loss": 1.0212,
+      "step": 4414
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8166783309791194e-05,
+      "loss": 0.4155,
+      "step": 4415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8165959468079995e-05,
+      "loss": 0.4673,
+      "step": 4416
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816513544713135e-05,
+      "loss": 1.4242,
+      "step": 4417
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816431124695049e-05,
+      "loss": 0.4091,
+      "step": 4418
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816348686754267e-05,
+      "loss": 0.5582,
+      "step": 4419
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816266230891315e-05,
+      "loss": 1.4113,
+      "step": 4420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816183757106713e-05,
+      "loss": 0.8423,
+      "step": 4421
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816101265400991e-05,
+      "loss": 0.9216,
+      "step": 4422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816018755774671e-05,
+      "loss": 0.4651,
+      "step": 4423
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8159362282282797e-05,
+      "loss": 0.4871,
+      "step": 4424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815853682762341e-05,
+      "loss": 0.4178,
+      "step": 4425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815771119377381e-05,
+      "loss": 0.4096,
+      "step": 4426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815688538073924e-05,
+      "loss": 0.7131,
+      "step": 4427
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815605938852497e-05,
+      "loss": 0.5843,
+      "step": 4428
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815523321713625e-05,
+      "loss": 0.512,
+      "step": 4429
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8154406866578324e-05,
+      "loss": 0.5134,
+      "step": 4430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8153580336856476e-05,
+      "loss": 1.0096,
+      "step": 4431
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815275362797595e-05,
+      "loss": 1.1955,
+      "step": 4432
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815192673994201e-05,
+      "loss": 0.8036,
+      "step": 4433
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815109967275992e-05,
+      "loss": 0.8903,
+      "step": 4434
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815027242643494e-05,
+      "loss": 1.2878,
+      "step": 4435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814944500097233e-05,
+      "loss": 0.4881,
+      "step": 4436
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814861739637736e-05,
+      "loss": 0.5356,
+      "step": 4437
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.81477896126553e-05,
+      "loss": 1.3708,
+      "step": 4438
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814696164981142e-05,
+      "loss": 1.2849,
+      "step": 4439
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814613350785098e-05,
+      "loss": 1.5471,
+      "step": 4440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8145305186779256e-05,
+      "loss": 0.8752,
+      "step": 4441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814447668660151e-05,
+      "loss": 0.5189,
+      "step": 4442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814364800732303e-05,
+      "loss": 0.5611,
+      "step": 4443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814281914894907e-05,
+      "loss": 0.9727,
+      "step": 4444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814199011148492e-05,
+      "loss": 1.5502,
+      "step": 4445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8141160894935854e-05,
+      "loss": 1.2189,
+      "step": 4446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8140331499307147e-05,
+      "loss": 0.9115,
+      "step": 4447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8139501924604066e-05,
+      "loss": 0.8241,
+      "step": 4448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8138672170831905e-05,
+      "loss": 1.3197,
+      "step": 4449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.813784223799594e-05,
+      "loss": 0.562,
+      "step": 4450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.813701212610145e-05,
+      "loss": 1.3518,
+      "step": 4451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.813618183515371e-05,
+      "loss": 0.3428,
+      "step": 4452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.813535136515803e-05,
+      "loss": 0.5,
+      "step": 4453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.813452071611967e-05,
+      "loss": 1.05,
+      "step": 4454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8133689888043916e-05,
+      "loss": 0.8166,
+      "step": 4455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.813285888093607e-05,
+      "loss": 1.0936,
+      "step": 4456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.813202769480141e-05,
+      "loss": 1.4965,
+      "step": 4457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8131196329645225e-05,
+      "loss": 0.6444,
+      "step": 4458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.813036478547281e-05,
+      "loss": 0.2491,
+      "step": 4459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.812953306228946e-05,
+      "loss": 0.9075,
+      "step": 4460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8128701160100456e-05,
+      "loss": 0.5219,
+      "step": 4461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8127869078911105e-05,
+      "loss": 1.431,
+      "step": 4462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.812703681872669e-05,
+      "loss": 1.1114,
+      "step": 4463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.812620437955251e-05,
+      "loss": 0.3306,
+      "step": 4464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.812537176139387e-05,
+      "loss": 0.2944,
+      "step": 4465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.812453896425606e-05,
+      "loss": 0.8628,
+      "step": 4466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.812370598814439e-05,
+      "loss": 1.2377,
+      "step": 4467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8122872833064156e-05,
+      "loss": 1.2914,
+      "step": 4468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.812203949902064e-05,
+      "loss": 0.1842,
+      "step": 4469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.812120598601918e-05,
+      "loss": 1.3441,
+      "step": 4470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.812037229406505e-05,
+      "loss": 1.5614,
+      "step": 4471
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8119538423163575e-05,
+      "loss": 0.5738,
+      "step": 4472
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.811870437332006e-05,
+      "loss": 0.6658,
+      "step": 4473
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.81178701445398e-05,
+      "loss": 0.5116,
+      "step": 4474
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8117035736828115e-05,
+      "loss": 0.5589,
+      "step": 4475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.811620115019031e-05,
+      "loss": 0.3154,
+      "step": 4476
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8115366384631694e-05,
+      "loss": 1.1496,
+      "step": 4477
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.811453144015759e-05,
+      "loss": 0.7989,
+      "step": 4478
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8113696316773295e-05,
+      "loss": 0.265,
+      "step": 4479
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.811286101448413e-05,
+      "loss": 1.1255,
+      "step": 4480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8112025533295424e-05,
+      "loss": 1.2684,
+      "step": 4481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.811118987321247e-05,
+      "loss": 1.2147,
+      "step": 4482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.81103540342406e-05,
+      "loss": 1.0347,
+      "step": 4483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810951801638514e-05,
+      "loss": 0.3934,
+      "step": 4484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810868181965139e-05,
+      "loss": 0.9942,
+      "step": 4485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8107845444044686e-05,
+      "loss": 0.5255,
+      "step": 4486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810700888957034e-05,
+      "loss": 0.928,
+      "step": 4487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810617215623369e-05,
+      "loss": 1.4162,
+      "step": 4488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810533524404005e-05,
+      "loss": 1.0642,
+      "step": 4489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8104498152994744e-05,
+      "loss": 1.1991,
+      "step": 4490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810366088310311e-05,
+      "loss": 0.6571,
+      "step": 4491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810282343437046e-05,
+      "loss": 0.9152,
+      "step": 4492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810198580680214e-05,
+      "loss": 0.6199,
+      "step": 4493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8101148000403475e-05,
+      "loss": 1.3667,
+      "step": 4494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810031001517979e-05,
+      "loss": 0.741,
+      "step": 4495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8099471851136424e-05,
+      "loss": 1.4875,
+      "step": 4496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809863350827871e-05,
+      "loss": 0.9647,
+      "step": 4497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809779498661198e-05,
+      "loss": 0.9572,
+      "step": 4498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809695628614157e-05,
+      "loss": 0.3792,
+      "step": 4499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809611740687282e-05,
+      "loss": 0.8118,
+      "step": 4500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8095278348811074e-05,
+      "loss": 1.1146,
+      "step": 4501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809443911196166e-05,
+      "loss": 0.6798,
+      "step": 4502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8093599696329915e-05,
+      "loss": 1.2027,
+      "step": 4503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809276010192119e-05,
+      "loss": 0.5569,
+      "step": 4504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809192032874083e-05,
+      "loss": 1.3029,
+      "step": 4505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8091080376794186e-05,
+      "loss": 1.1712,
+      "step": 4506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809024024608658e-05,
+      "loss": 0.7223,
+      "step": 4507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8089399936623375e-05,
+      "loss": 1.0885,
+      "step": 4508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808855944840992e-05,
+      "loss": 0.4174,
+      "step": 4509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808771878145155e-05,
+      "loss": 1.2145,
+      "step": 4510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808687793575363e-05,
+      "loss": 0.9159,
+      "step": 4511
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808603691132149e-05,
+      "loss": 1.6274,
+      "step": 4512
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808519570816051e-05,
+      "loss": 0.6929,
+      "step": 4513
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8084354326276024e-05,
+      "loss": 1.3744,
+      "step": 4514
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808351276567339e-05,
+      "loss": 0.6713,
+      "step": 4515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8082671026357964e-05,
+      "loss": 1.009,
+      "step": 4516
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8081829108335103e-05,
+      "loss": 0.8544,
+      "step": 4517
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8080987011610165e-05,
+      "loss": 0.9086,
+      "step": 4518
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8080144736188515e-05,
+      "loss": 0.4151,
+      "step": 4519
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.80793022820755e-05,
+      "loss": 0.4483,
+      "step": 4520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807845964927648e-05,
+      "loss": 1.0366,
+      "step": 4521
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807761683779684e-05,
+      "loss": 1.2418,
+      "step": 4522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8076773847641915e-05,
+      "loss": 0.8193,
+      "step": 4523
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807593067881709e-05,
+      "loss": 0.3889,
+      "step": 4524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807508733132771e-05,
+      "loss": 0.6067,
+      "step": 4525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8074243805179164e-05,
+      "loss": 0.705,
+      "step": 4526
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807340010037681e-05,
+      "loss": 1.2983,
+      "step": 4527
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8072556216926015e-05,
+      "loss": 1.2973,
+      "step": 4528
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8071712154832156e-05,
+      "loss": 0.4752,
+      "step": 4529
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.80708679141006e-05,
+      "loss": 1.4291,
+      "step": 4530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807002349473672e-05,
+      "loss": 0.4292,
+      "step": 4531
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8069178896745876e-05,
+      "loss": 1.3923,
+      "step": 4532
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.806833412013347e-05,
+      "loss": 0.6371,
+      "step": 4533
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8067489164904856e-05,
+      "loss": 1.1667,
+      "step": 4534
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8066644031065426e-05,
+      "loss": 1.4652,
+      "step": 4535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.806579871862055e-05,
+      "loss": 1.4126,
+      "step": 4536
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.806495322757559e-05,
+      "loss": 0.3559,
+      "step": 4537
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.806410755793596e-05,
+      "loss": 1.3343,
+      "step": 4538
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.806326170970703e-05,
+      "loss": 1.3992,
+      "step": 4539
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.806241568289417e-05,
+      "loss": 0.9757,
+      "step": 4540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.806156947750277e-05,
+      "loss": 0.382,
+      "step": 4541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.806072309353822e-05,
+      "loss": 0.9225,
+      "step": 4542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.805987653100591e-05,
+      "loss": 0.9262,
+      "step": 4543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8059029789911214e-05,
+      "loss": 0.4299,
+      "step": 4544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.805818287025952e-05,
+      "loss": 0.4727,
+      "step": 4545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.805733577205623e-05,
+      "loss": 1.1527,
+      "step": 4546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.805648849530673e-05,
+      "loss": 1.2572,
+      "step": 4547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.805564104001641e-05,
+      "loss": 1.0621,
+      "step": 4548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8054793406190665e-05,
+      "loss": 0.4635,
+      "step": 4549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8053945593834884e-05,
+      "loss": 0.3694,
+      "step": 4550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.805309760295447e-05,
+      "loss": 0.7575,
+      "step": 4551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8052249433554804e-05,
+      "loss": 0.9029,
+      "step": 4552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8051401085641306e-05,
+      "loss": 0.5052,
+      "step": 4553
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8050552559219356e-05,
+      "loss": 1.3405,
+      "step": 4554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.804970385429437e-05,
+      "loss": 0.3251,
+      "step": 4555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.804885497087173e-05,
+      "loss": 0.3268,
+      "step": 4556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.804800590895685e-05,
+      "loss": 0.8816,
+      "step": 4557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8047156668555125e-05,
+      "loss": 1.0109,
+      "step": 4558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.804630724967197e-05,
+      "loss": 1.5707,
+      "step": 4559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.804545765231279e-05,
+      "loss": 0.4072,
+      "step": 4560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.804460787648298e-05,
+      "loss": 0.6141,
+      "step": 4561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8043757922187955e-05,
+      "loss": 1.9025,
+      "step": 4562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.804290778943312e-05,
+      "loss": 0.4705,
+      "step": 4563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.804205747822389e-05,
+      "loss": 0.2603,
+      "step": 4564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.804120698856569e-05,
+      "loss": 1.4475,
+      "step": 4565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.804035632046389e-05,
+      "loss": 0.9056,
+      "step": 4566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.803950547392394e-05,
+      "loss": 0.4449,
+      "step": 4567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.803865444895125e-05,
+      "loss": 1.3844,
+      "step": 4568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8037803245551216e-05,
+      "loss": 1.3092,
+      "step": 4569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.803695186372928e-05,
+      "loss": 0.5906,
+      "step": 4570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.803610030349085e-05,
+      "loss": 1.4666,
+      "step": 4571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.803524856484133e-05,
+      "loss": 0.3632,
+      "step": 4572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.803439664778617e-05,
+      "loss": 0.8167,
+      "step": 4573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.803354455233076e-05,
+      "loss": 0.5114,
+      "step": 4574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.803269227848055e-05,
+      "loss": 0.5058,
+      "step": 4575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8031839826240933e-05,
+      "loss": 0.5727,
+      "step": 4576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8030987195617365e-05,
+      "loss": 0.9648,
+      "step": 4577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.803013438661525e-05,
+      "loss": 1.0088,
+      "step": 4578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.802928139924003e-05,
+      "loss": 1.0925,
+      "step": 4579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8028428233497126e-05,
+      "loss": 0.4521,
+      "step": 4580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802757488939197e-05,
+      "loss": 1.5301,
+      "step": 4581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8026721366929986e-05,
+      "loss": 0.7829,
+      "step": 4582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802586766611661e-05,
+      "loss": 0.5005,
+      "step": 4583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802501378695727e-05,
+      "loss": 0.3935,
+      "step": 4584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.80241597294574e-05,
+      "loss": 0.9835,
+      "step": 4585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802330549362245e-05,
+      "loss": 0.4869,
+      "step": 4586
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802245107945783e-05,
+      "loss": 2.0095,
+      "step": 4587
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8021596486969e-05,
+      "loss": 1.2369,
+      "step": 4588
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802074171616139e-05,
+      "loss": 0.4486,
+      "step": 4589
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8019886767040435e-05,
+      "loss": 0.3502,
+      "step": 4590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8019031639611586e-05,
+      "loss": 0.6838,
+      "step": 4591
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8018176333880276e-05,
+      "loss": 0.3897,
+      "step": 4592
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801732084985195e-05,
+      "loss": 0.4246,
+      "step": 4593
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801646518753206e-05,
+      "loss": 0.8474,
+      "step": 4594
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801560934692604e-05,
+      "loss": 1.7773,
+      "step": 4595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801475332803933e-05,
+      "loss": 1.4932,
+      "step": 4596
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8013897130877394e-05,
+      "loss": 0.8464,
+      "step": 4597
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801304075544567e-05,
+      "loss": 2.1701,
+      "step": 4598
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8012184201749624e-05,
+      "loss": 1.2559,
+      "step": 4599
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801132746979468e-05,
+      "loss": 0.4527,
+      "step": 4600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801047055958631e-05,
+      "loss": 0.7374,
+      "step": 4601
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8009613471129954e-05,
+      "loss": 1.5675,
+      "step": 4602
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800875620443108e-05,
+      "loss": 1.2448,
+      "step": 4603
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800789875949513e-05,
+      "loss": 0.8076,
+      "step": 4604
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800704113632758e-05,
+      "loss": 1.5001,
+      "step": 4605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800618333493387e-05,
+      "loss": 1.7026,
+      "step": 4606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800532535531946e-05,
+      "loss": 0.7612,
+      "step": 4607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800446719748981e-05,
+      "loss": 0.5772,
+      "step": 4608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800360886145039e-05,
+      "loss": 0.735,
+      "step": 4609
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8002750347206655e-05,
+      "loss": 0.6182,
+      "step": 4610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800189165476407e-05,
+      "loss": 0.7969,
+      "step": 4611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8001032784128096e-05,
+      "loss": 0.3371,
+      "step": 4612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8000173735304206e-05,
+      "loss": 1.4578,
+      "step": 4613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.799931450829786e-05,
+      "loss": 0.868,
+      "step": 4614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.799845510311453e-05,
+      "loss": 0.192,
+      "step": 4615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.799759551975968e-05,
+      "loss": 0.7148,
+      "step": 4616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7996735758238784e-05,
+      "loss": 0.6043,
+      "step": 4617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.799587581855731e-05,
+      "loss": 1.2046,
+      "step": 4618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.799501570072074e-05,
+      "loss": 0.387,
+      "step": 4619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.799415540473453e-05,
+      "loss": 1.2669,
+      "step": 4620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.799329493060417e-05,
+      "loss": 0.7032,
+      "step": 4621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.799243427833513e-05,
+      "loss": 0.4525,
+      "step": 4622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7991573447932884e-05,
+      "loss": 1.7875,
+      "step": 4623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.799071243940291e-05,
+      "loss": 1.5165,
+      "step": 4624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.79898512527507e-05,
+      "loss": 0.9732,
+      "step": 4625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.798898988798171e-05,
+      "loss": 0.3329,
+      "step": 4626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.798812834510145e-05,
+      "loss": 0.8181,
+      "step": 4627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.798726662411539e-05,
+      "loss": 0.7309,
+      "step": 4628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.798640472502901e-05,
+      "loss": 1.7384,
+      "step": 4629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7985542647847794e-05,
+      "loss": 0.7101,
+      "step": 4630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7984680392577226e-05,
+      "loss": 0.3681,
+      "step": 4631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.79838179592228e-05,
+      "loss": 0.668,
+      "step": 4632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.798295534779001e-05,
+      "loss": 0.3078,
+      "step": 4633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.798209255828433e-05,
+      "loss": 0.4378,
+      "step": 4634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.798122959071126e-05,
+      "loss": 1.048,
+      "step": 4635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.798036644507628e-05,
+      "loss": 0.5578,
+      "step": 4636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.797950312138491e-05,
+      "loss": 0.6624,
+      "step": 4637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.797863961964261e-05,
+      "loss": 0.4924,
+      "step": 4638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.79777759398549e-05,
+      "loss": 1.247,
+      "step": 4639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7976912082027264e-05,
+      "loss": 0.6043,
+      "step": 4640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7976048046165206e-05,
+      "loss": 0.7782,
+      "step": 4641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.797518383227422e-05,
+      "loss": 1.2017,
+      "step": 4642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.79743194403598e-05,
+      "loss": 1.2905,
+      "step": 4643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.797345487042746e-05,
+      "loss": 0.9793,
+      "step": 4644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.79725901224827e-05,
+      "loss": 1.4879,
+      "step": 4645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7971725196531005e-05,
+      "loss": 0.6704,
+      "step": 4646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7970860092577904e-05,
+      "loss": 1.93,
+      "step": 4647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.796999481062888e-05,
+      "loss": 1.2962,
+      "step": 4648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.796912935068946e-05,
+      "loss": 0.7179,
+      "step": 4649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.796826371276513e-05,
+      "loss": 0.3409,
+      "step": 4650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.796739789686141e-05,
+      "loss": 0.2462,
+      "step": 4651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7966531902983817e-05,
+      "loss": 1.2003,
+      "step": 4652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.796566573113785e-05,
+      "loss": 0.4994,
+      "step": 4653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7964799381329024e-05,
+      "loss": 1.1204,
+      "step": 4654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.796393285356285e-05,
+      "loss": 1.291,
+      "step": 4655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7963066147844845e-05,
+      "loss": 1.4831,
+      "step": 4656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7962199264180526e-05,
+      "loss": 1.1702,
+      "step": 4657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7961332202575415e-05,
+      "loss": 1.0728,
+      "step": 4658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.796046496303502e-05,
+      "loss": 0.1574,
+      "step": 4659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795959754556486e-05,
+      "loss": 1.0623,
+      "step": 4660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795872995017046e-05,
+      "loss": 1.4677,
+      "step": 4661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795786217685734e-05,
+      "loss": 0.6109,
+      "step": 4662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7956994225631015e-05,
+      "loss": 0.8004,
+      "step": 4663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795612609649702e-05,
+      "loss": 0.5664,
+      "step": 4664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795525778946087e-05,
+      "loss": 1.0942,
+      "step": 4665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7954389304528096e-05,
+      "loss": 0.3384,
+      "step": 4666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795352064170422e-05,
+      "loss": 0.3932,
+      "step": 4667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795265180099478e-05,
+      "loss": 1.216,
+      "step": 4668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795178278240529e-05,
+      "loss": 1.339,
+      "step": 4669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7950913585941286e-05,
+      "loss": 0.4119,
+      "step": 4670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795004421160831e-05,
+      "loss": 1.0588,
+      "step": 4671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.794917465941188e-05,
+      "loss": 1.8467,
+      "step": 4672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7948304929357536e-05,
+      "loss": 0.9383,
+      "step": 4673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.794743502145081e-05,
+      "loss": 1.1007,
+      "step": 4674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.794656493569724e-05,
+      "loss": 1.4818,
+      "step": 4675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.794569467210236e-05,
+      "loss": 0.6596,
+      "step": 4676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.794482423067172e-05,
+      "loss": 0.2362,
+      "step": 4677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7943953611410835e-05,
+      "loss": 0.3762,
+      "step": 4678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.794308281432527e-05,
+      "loss": 1.2673,
+      "step": 4679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.794221183942055e-05,
+      "loss": 1.2109,
+      "step": 4680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.794134068670223e-05,
+      "loss": 0.3356,
+      "step": 4681
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7940469356175844e-05,
+      "loss": 0.7109,
+      "step": 4682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.793959784784694e-05,
+      "loss": 0.4231,
+      "step": 4683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7938726161721066e-05,
+      "loss": 1.0656,
+      "step": 4684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.793785429780376e-05,
+      "loss": 0.8498,
+      "step": 4685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.793698225610058e-05,
+      "loss": 1.2623,
+      "step": 4686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.793611003661708e-05,
+      "loss": 1.2107,
+      "step": 4687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.79352376393588e-05,
+      "loss": 0.5182,
+      "step": 4688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.793436506433129e-05,
+      "loss": 0.6568,
+      "step": 4689
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.793349231154011e-05,
+      "loss": 0.3186,
+      "step": 4690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.793261938099081e-05,
+      "loss": 1.1711,
+      "step": 4691
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.793174627268895e-05,
+      "loss": 0.6561,
+      "step": 4692
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7930872986640086e-05,
+      "loss": 1.081,
+      "step": 4693
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792999952284976e-05,
+      "loss": 1.2555,
+      "step": 4694
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792912588132355e-05,
+      "loss": 0.4434,
+      "step": 4695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792825206206701e-05,
+      "loss": 0.3451,
+      "step": 4696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.79273780650857e-05,
+      "loss": 1.4397,
+      "step": 4697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792650389038517e-05,
+      "loss": 1.2494,
+      "step": 4698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792562953797099e-05,
+      "loss": 0.6159,
+      "step": 4699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792475500784874e-05,
+      "loss": 0.976,
+      "step": 4700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7923880300023964e-05,
+      "loss": 0.455,
+      "step": 4701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792300541450224e-05,
+      "loss": 0.6373,
+      "step": 4702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7922130351289135e-05,
+      "loss": 0.6214,
+      "step": 4703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792125511039021e-05,
+      "loss": 0.4587,
+      "step": 4704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792037969181104e-05,
+      "loss": 1.5292,
+      "step": 4705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.79195040955572e-05,
+      "loss": 0.1956,
+      "step": 4706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791862832163425e-05,
+      "loss": 0.7634,
+      "step": 4707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7917752370047775e-05,
+      "loss": 1.0414,
+      "step": 4708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791687624080334e-05,
+      "loss": 1.4056,
+      "step": 4709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7915999933906526e-05,
+      "loss": 0.9966,
+      "step": 4710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791512344936291e-05,
+      "loss": 0.2772,
+      "step": 4711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791424678717807e-05,
+      "loss": 1.0258,
+      "step": 4712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791336994735757e-05,
+      "loss": 0.5563,
+      "step": 4713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791249292990702e-05,
+      "loss": 0.7782,
+      "step": 4714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791161573483197e-05,
+      "loss": 0.8417,
+      "step": 4715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7910738362138025e-05,
+      "loss": 1.5645,
+      "step": 4716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7909860811830756e-05,
+      "loss": 1.0947,
+      "step": 4717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790898308391575e-05,
+      "loss": 1.3276,
+      "step": 4718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790810517839859e-05,
+      "loss": 0.6551,
+      "step": 4719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790722709528487e-05,
+      "loss": 0.4704,
+      "step": 4720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790634883458017e-05,
+      "loss": 1.4095,
+      "step": 4721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790547039629008e-05,
+      "loss": 1.1531,
+      "step": 4722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790459178042019e-05,
+      "loss": 0.4542,
+      "step": 4723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7903712986976104e-05,
+      "loss": 0.8381,
+      "step": 4724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.79028340159634e-05,
+      "loss": 0.3418,
+      "step": 4725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7901954867387674e-05,
+      "loss": 1.1713,
+      "step": 4726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790107554125452e-05,
+      "loss": 1.1185,
+      "step": 4727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790019603756953e-05,
+      "loss": 0.4483,
+      "step": 4728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789931635633831e-05,
+      "loss": 1.1353,
+      "step": 4729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789843649756646e-05,
+      "loss": 1.0254,
+      "step": 4730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789755646125957e-05,
+      "loss": 1.5113,
+      "step": 4731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789667624742325e-05,
+      "loss": 1.5753,
+      "step": 4732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789579585606308e-05,
+      "loss": 1.4674,
+      "step": 4733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789491528718469e-05,
+      "loss": 0.8589,
+      "step": 4734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789403454079366e-05,
+      "loss": 0.5104,
+      "step": 4735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7893153616895615e-05,
+      "loss": 0.6218,
+      "step": 4736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789227251549615e-05,
+      "loss": 0.6215,
+      "step": 4737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789139123660087e-05,
+      "loss": 0.3307,
+      "step": 4738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7890509780215386e-05,
+      "loss": 0.6158,
+      "step": 4739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788962814634531e-05,
+      "loss": 0.6867,
+      "step": 4740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788874633499625e-05,
+      "loss": 0.2887,
+      "step": 4741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788786434617382e-05,
+      "loss": 0.9147,
+      "step": 4742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788698217988363e-05,
+      "loss": 1.1697,
+      "step": 4743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788609983613129e-05,
+      "loss": 0.5777,
+      "step": 4744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788521731492242e-05,
+      "loss": 1.278,
+      "step": 4745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788433461626264e-05,
+      "loss": 1.1223,
+      "step": 4746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7883451740157564e-05,
+      "loss": 1.0627,
+      "step": 4747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7882568686612794e-05,
+      "loss": 1.325,
+      "step": 4748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7881685455633975e-05,
+      "loss": 0.4178,
+      "step": 4749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7880802047226724e-05,
+      "loss": 2.2962,
+      "step": 4750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.787991846139664e-05,
+      "loss": 0.2396,
+      "step": 4751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.787903469814937e-05,
+      "loss": 1.0295,
+      "step": 4752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.787815075749053e-05,
+      "loss": 0.4328,
+      "step": 4753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.787726663942573e-05,
+      "loss": 1.3856,
+      "step": 4754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.787638234396062e-05,
+      "loss": 0.3812,
+      "step": 4755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.787549787110082e-05,
+      "loss": 0.6686,
+      "step": 4756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7874613220851954e-05,
+      "loss": 0.4168,
+      "step": 4757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.787372839321965e-05,
+      "loss": 1.2049,
+      "step": 4758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.787284338820955e-05,
+      "loss": 0.3735,
+      "step": 4759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7871958205827274e-05,
+      "loss": 0.3737,
+      "step": 4760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7871072846078455e-05,
+      "loss": 1.1924,
+      "step": 4761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.787018730896873e-05,
+      "loss": 1.323,
+      "step": 4762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.786930159450375e-05,
+      "loss": 0.3736,
+      "step": 4763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.786841570268912e-05,
+      "loss": 1.8889,
+      "step": 4764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.78675296335305e-05,
+      "loss": 1.4234,
+      "step": 4765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.786664338703353e-05,
+      "loss": 0.2356,
+      "step": 4766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.786575696320383e-05,
+      "loss": 1.1219,
+      "step": 4767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7864870362047056e-05,
+      "loss": 1.1053,
+      "step": 4768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.786398358356886e-05,
+      "loss": 1.1597,
+      "step": 4769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.786309662777486e-05,
+      "loss": 1.5782,
+      "step": 4770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.786220949467071e-05,
+      "loss": 1.3343,
+      "step": 4771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7861322184262065e-05,
+      "loss": 0.2612,
+      "step": 4772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.786043469655456e-05,
+      "loss": 0.9108,
+      "step": 4773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.785954703155386e-05,
+      "loss": 0.3211,
+      "step": 4774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7858659189265585e-05,
+      "loss": 0.3762,
+      "step": 4775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7857771169695415e-05,
+      "loss": 0.6296,
+      "step": 4776
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7856882972848974e-05,
+      "loss": 0.6665,
+      "step": 4777
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.785599459873194e-05,
+      "loss": 1.0123,
+      "step": 4778
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.785510604734994e-05,
+      "loss": 1.2571,
+      "step": 4779
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.785421731870865e-05,
+      "loss": 1.1353,
+      "step": 4780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.785332841281371e-05,
+      "loss": 0.8392,
+      "step": 4781
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.785243932967079e-05,
+      "loss": 0.4735,
+      "step": 4782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.785155006928554e-05,
+      "loss": 1.0473,
+      "step": 4783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.785066063166363e-05,
+      "loss": 0.8019,
+      "step": 4784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7849771016810694e-05,
+      "loss": 1.5673,
+      "step": 4785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.784888122473242e-05,
+      "loss": 0.2352,
+      "step": 4786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.784799125543446e-05,
+      "loss": 0.3802,
+      "step": 4787
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7847101108922474e-05,
+      "loss": 0.3795,
+      "step": 4788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.784621078520213e-05,
+      "loss": 0.2803,
+      "step": 4789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.784532028427909e-05,
+      "loss": 1.1455,
+      "step": 4790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7844429606159034e-05,
+      "loss": 0.8081,
+      "step": 4791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.784353875084762e-05,
+      "loss": 0.5294,
+      "step": 4792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.784264771835051e-05,
+      "loss": 0.5222,
+      "step": 4793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.784175650867339e-05,
+      "loss": 0.5129,
+      "step": 4794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.784086512182191e-05,
+      "loss": 0.7854,
+      "step": 4795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.783997355780177e-05,
+      "loss": 0.617,
+      "step": 4796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7839081816618625e-05,
+      "loss": 0.5389,
+      "step": 4797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.783818989827816e-05,
+      "loss": 1.7378,
+      "step": 4798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7837297802786034e-05,
+      "loss": 0.073,
+      "step": 4799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7836405530147935e-05,
+      "loss": 1.4077,
+      "step": 4800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.783551308036954e-05,
+      "loss": 1.4334,
+      "step": 4801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.783462045345653e-05,
+      "loss": 1.1322,
+      "step": 4802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.78337276494146e-05,
+      "loss": 0.5742,
+      "step": 4803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.78328346682494e-05,
+      "loss": 1.0242,
+      "step": 4804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7831941509966625e-05,
+      "loss": 0.3774,
+      "step": 4805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7831048174571966e-05,
+      "loss": 0.4758,
+      "step": 4806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.783015466207111e-05,
+      "loss": 1.3982,
+      "step": 4807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7829260972469734e-05,
+      "loss": 1.1374,
+      "step": 4808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.782836710577353e-05,
+      "loss": 0.5847,
+      "step": 4809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7827473061988176e-05,
+      "loss": 0.4615,
+      "step": 4810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.782657884111937e-05,
+      "loss": 1.2651,
+      "step": 4811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7825684443172816e-05,
+      "loss": 0.9409,
+      "step": 4812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.782478986815418e-05,
+      "loss": 0.6181,
+      "step": 4813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.782389511606917e-05,
+      "loss": 1.0193,
+      "step": 4814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.782300018692347e-05,
+      "loss": 0.9539,
+      "step": 4815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7822105080722786e-05,
+      "loss": 0.6991,
+      "step": 4816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7821209797472805e-05,
+      "loss": 1.1151,
+      "step": 4817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7820314337179236e-05,
+      "loss": 0.4767,
+      "step": 4818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7819418699847766e-05,
+      "loss": 0.6986,
+      "step": 4819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.78185228854841e-05,
+      "loss": 0.3872,
+      "step": 4820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.781762689409394e-05,
+      "loss": 0.9576,
+      "step": 4821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7816730725682976e-05,
+      "loss": 0.9833,
+      "step": 4822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7815834380256925e-05,
+      "loss": 0.8369,
+      "step": 4823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7814937857821486e-05,
+      "loss": 0.6203,
+      "step": 4824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7814041158382365e-05,
+      "loss": 1.4902,
+      "step": 4825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.781314428194527e-05,
+      "loss": 1.2907,
+      "step": 4826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.781224722851589e-05,
+      "loss": 0.358,
+      "step": 4827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7811349998099966e-05,
+      "loss": 0.4896,
+      "step": 4828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7810452590703186e-05,
+      "loss": 0.6444,
+      "step": 4829
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780955500633126e-05,
+      "loss": 0.4796,
+      "step": 4830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780865724498991e-05,
+      "loss": 0.5271,
+      "step": 4831
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780775930668485e-05,
+      "loss": 0.3365,
+      "step": 4832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780686119142178e-05,
+      "loss": 0.7048,
+      "step": 4833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780596289920642e-05,
+      "loss": 0.7213,
+      "step": 4834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.78050644300445e-05,
+      "loss": 0.7906,
+      "step": 4835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780416578394172e-05,
+      "loss": 0.5022,
+      "step": 4836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7803266960903804e-05,
+      "loss": 1.2211,
+      "step": 4837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780236796093649e-05,
+      "loss": 1.2595,
+      "step": 4838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780146878404546e-05,
+      "loss": 1.7927,
+      "step": 4839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780056943023647e-05,
+      "loss": 1.7147,
+      "step": 4840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7799669899515234e-05,
+      "loss": 0.6825,
+      "step": 4841
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.779877019188747e-05,
+      "loss": 0.5487,
+      "step": 4842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.779787030735891e-05,
+      "loss": 0.4929,
+      "step": 4843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7796970245935285e-05,
+      "loss": 0.6925,
+      "step": 4844
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.779607000762231e-05,
+      "loss": 0.594,
+      "step": 4845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.779516959242572e-05,
+      "loss": 1.2907,
+      "step": 4846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.779426900035124e-05,
+      "loss": 0.5538,
+      "step": 4847
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.779336823140461e-05,
+      "loss": 0.8529,
+      "step": 4848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.779246728559156e-05,
+      "loss": 1.025,
+      "step": 4849
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.779156616291782e-05,
+      "loss": 0.5485,
+      "step": 4850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.779066486338912e-05,
+      "loss": 0.4452,
+      "step": 4851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.77897633870112e-05,
+      "loss": 0.437,
+      "step": 4852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.778886173378981e-05,
+      "loss": 0.7462,
+      "step": 4853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.778795990373067e-05,
+      "loss": 0.7456,
+      "step": 4854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.778705789683952e-05,
+      "loss": 0.5226,
+      "step": 4855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.77861557131221e-05,
+      "loss": 1.5325,
+      "step": 4856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.778525335258416e-05,
+      "loss": 0.2201,
+      "step": 4857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7784350815231446e-05,
+      "loss": 0.8737,
+      "step": 4858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.778344810106968e-05,
+      "loss": 1.7831,
+      "step": 4859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7782545210104626e-05,
+      "loss": 1.7092,
+      "step": 4860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.778164214234203e-05,
+      "loss": 0.8803,
+      "step": 4861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.778073889778762e-05,
+      "loss": 1.358,
+      "step": 4862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7779835476447164e-05,
+      "loss": 1.0897,
+      "step": 4863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7778931878326405e-05,
+      "loss": 0.8862,
+      "step": 4864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.777802810343108e-05,
+      "loss": 1.0778,
+      "step": 4865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.777712415176696e-05,
+      "loss": 0.6084,
+      "step": 4866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.777622002333979e-05,
+      "loss": 1.3334,
+      "step": 4867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7775315718155316e-05,
+      "loss": 0.2925,
+      "step": 4868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.77744112362193e-05,
+      "loss": 1.1014,
+      "step": 4869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7773506577537504e-05,
+      "loss": 1.0409,
+      "step": 4870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.777260174211567e-05,
+      "loss": 0.9795,
+      "step": 4871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.777169672995957e-05,
+      "loss": 0.9712,
+      "step": 4872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7770791541074944e-05,
+      "loss": 0.4286,
+      "step": 4873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7769886175467583e-05,
+      "loss": 1.3075,
+      "step": 4874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.776898063314322e-05,
+      "loss": 0.1199,
+      "step": 4875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.776807491410763e-05,
+      "loss": 0.5432,
+      "step": 4876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.776716901836657e-05,
+      "loss": 0.9994,
+      "step": 4877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.776626294592581e-05,
+      "loss": 1.1785,
+      "step": 4878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.776535669679113e-05,
+      "loss": 1.4669,
+      "step": 4879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7764450270968275e-05,
+      "loss": 0.5472,
+      "step": 4880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.776354366846301e-05,
+      "loss": 1.3927,
+      "step": 4881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.776263688928113e-05,
+      "loss": 1.1615,
+      "step": 4882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.776172993342838e-05,
+      "loss": 0.7844,
+      "step": 4883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7760822800910544e-05,
+      "loss": 0.906,
+      "step": 4884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.77599154917334e-05,
+      "loss": 1.3282,
+      "step": 4885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7759008005902704e-05,
+      "loss": 1.5549,
+      "step": 4886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.775810034342424e-05,
+      "loss": 0.39,
+      "step": 4887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7757192504303794e-05,
+      "loss": 0.9538,
+      "step": 4888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.775628448854713e-05,
+      "loss": 1.4803,
+      "step": 4889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7755376296160036e-05,
+      "loss": 0.5494,
+      "step": 4890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.775446792714828e-05,
+      "loss": 0.8385,
+      "step": 4891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.775355938151764e-05,
+      "loss": 1.0005,
+      "step": 4892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.775265065927392e-05,
+      "loss": 0.5242,
+      "step": 4893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.775174176042289e-05,
+      "loss": 1.2586,
+      "step": 4894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.775083268497032e-05,
+      "loss": 1.0432,
+      "step": 4895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.774992343292201e-05,
+      "loss": 1.6142,
+      "step": 4896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.774901400428375e-05,
+      "loss": 1.428,
+      "step": 4897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7748104399061326e-05,
+      "loss": 1.4019,
+      "step": 4898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.774719461726051e-05,
+      "loss": 0.3962,
+      "step": 4899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.774628465888711e-05,
+      "loss": 1.2255,
+      "step": 4900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7745374523946914e-05,
+      "loss": 0.5231,
+      "step": 4901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.77444642124457e-05,
+      "loss": 1.1656,
+      "step": 4902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.774355372438927e-05,
+      "loss": 0.4493,
+      "step": 4903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.774264305978342e-05,
+      "loss": 0.0934,
+      "step": 4904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7741732218633946e-05,
+      "loss": 0.8359,
+      "step": 4905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.774082120094664e-05,
+      "loss": 1.3524,
+      "step": 4906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.773991000672731e-05,
+      "loss": 0.2861,
+      "step": 4907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.773899863598173e-05,
+      "loss": 0.326,
+      "step": 4908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.773808708871573e-05,
+      "loss": 0.7796,
+      "step": 4909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7737175364935084e-05,
+      "loss": 0.6997,
+      "step": 4910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.773626346464561e-05,
+      "loss": 0.7389,
+      "step": 4911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.773535138785311e-05,
+      "loss": 0.5787,
+      "step": 4912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7734439134563375e-05,
+      "loss": 1.0017,
+      "step": 4913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.773352670478223e-05,
+      "loss": 0.9165,
+      "step": 4914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.773261409851547e-05,
+      "loss": 0.8888,
+      "step": 4915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7731701315768907e-05,
+      "loss": 0.7062,
+      "step": 4916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7730788356548335e-05,
+      "loss": 1.1247,
+      "step": 4917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.772987522085959e-05,
+      "loss": 1.3558,
+      "step": 4918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7728961908708454e-05,
+      "loss": 0.8621,
+      "step": 4919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7728048420100754e-05,
+      "loss": 1.495,
+      "step": 4920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.772713475504231e-05,
+      "loss": 0.759,
+      "step": 4921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.772622091353893e-05,
+      "loss": 1.662,
+      "step": 4922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.772530689559642e-05,
+      "loss": 0.9571,
+      "step": 4923
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.772439270122061e-05,
+      "loss": 0.3256,
+      "step": 4924
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.772347833041731e-05,
+      "loss": 1.0355,
+      "step": 4925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.772256378319234e-05,
+      "loss": 0.4115,
+      "step": 4926
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.772164905955151e-05,
+      "loss": 0.6201,
+      "step": 4927
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7720734159500665e-05,
+      "loss": 1.086,
+      "step": 4928
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.771981908304561e-05,
+      "loss": 0.5675,
+      "step": 4929
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7718903830192166e-05,
+      "loss": 0.2062,
+      "step": 4930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7717988400946175e-05,
+      "loss": 1.3657,
+      "step": 4931
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.771707279531344e-05,
+      "loss": 0.5173,
+      "step": 4932
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7716157013299804e-05,
+      "loss": 1.3414,
+      "step": 4933
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7715241054911084e-05,
+      "loss": 1.0449,
+      "step": 4934
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.771432492015312e-05,
+      "loss": 0.1589,
+      "step": 4935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.771340860903172e-05,
+      "loss": 0.3882,
+      "step": 4936
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.771249212155275e-05,
+      "loss": 0.6262,
+      "step": 4937
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7711575457722005e-05,
+      "loss": 0.493,
+      "step": 4938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.771065861754534e-05,
+      "loss": 1.0549,
+      "step": 4939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.770974160102859e-05,
+      "loss": 1.2237,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.770882440817759e-05,
+      "loss": 0.5894,
+      "step": 4941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.770790703899817e-05,
+      "loss": 1.0943,
+      "step": 4942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7706989493496165e-05,
+      "loss": 1.4908,
+      "step": 4943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.770607177167742e-05,
+      "loss": 0.8523,
+      "step": 4944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.770515387354778e-05,
+      "loss": 1.2132,
+      "step": 4945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.770423579911308e-05,
+      "loss": 2.0764,
+      "step": 4946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.770331754837915e-05,
+      "loss": 0.5671,
+      "step": 4947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.770239912135186e-05,
+      "loss": 0.4404,
+      "step": 4948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.770148051803704e-05,
+      "loss": 1.4097,
+      "step": 4949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7700561738440526e-05,
+      "loss": 0.844,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.769964278256818e-05,
+      "loss": 0.7752,
+      "step": 4951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7698723650425846e-05,
+      "loss": 2.336,
+      "step": 4952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7697804342019374e-05,
+      "loss": 1.4555,
+      "step": 4953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.76968848573546e-05,
+      "loss": 1.033,
+      "step": 4954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.76959651964374e-05,
+      "loss": 0.3658,
+      "step": 4955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.769504535927361e-05,
+      "loss": 0.6996,
+      "step": 4956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.769412534586909e-05,
+      "loss": 1.4427,
+      "step": 4957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.769320515622968e-05,
+      "loss": 1.0644,
+      "step": 4958
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.769228479036126e-05,
+      "loss": 0.3576,
+      "step": 4959
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.769136424826967e-05,
+      "loss": 1.1142,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7690443529960777e-05,
+      "loss": 2.0422,
+      "step": 4961
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.768952263544043e-05,
+      "loss": 1.3762,
+      "step": 4962
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.768860156471449e-05,
+      "loss": 1.0491,
+      "step": 4963
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.768768031778883e-05,
+      "loss": 1.1335,
+      "step": 4964
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7686758894669304e-05,
+      "loss": 0.943,
+      "step": 4965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7685837295361775e-05,
+      "loss": 0.595,
+      "step": 4966
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.768491551987211e-05,
+      "loss": 0.3978,
+      "step": 4967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7683993568206174e-05,
+      "loss": 1.1256,
+      "step": 4968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7683071440369845e-05,
+      "loss": 0.7414,
+      "step": 4969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7682149136368966e-05,
+      "loss": 1.775,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7681226656209425e-05,
+      "loss": 0.5167,
+      "step": 4971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.768030399989709e-05,
+      "loss": 1.3664,
+      "step": 4972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.767938116743783e-05,
+      "loss": 0.9651,
+      "step": 4973
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.767845815883752e-05,
+      "loss": 0.8368,
+      "step": 4974
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.767753497410203e-05,
+      "loss": 0.4746,
+      "step": 4975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.767661161323723e-05,
+      "loss": 0.4058,
+      "step": 4976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.767568807624901e-05,
+      "loss": 0.4593,
+      "step": 4977
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.767476436314324e-05,
+      "loss": 0.6497,
+      "step": 4978
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.76738404739258e-05,
+      "loss": 1.4118,
+      "step": 4979
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.767291640860256e-05,
+      "loss": 1.4202,
+      "step": 4980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.767199216717941e-05,
+      "loss": 0.737,
+      "step": 4981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.767106774966223e-05,
+      "loss": 0.4569,
+      "step": 4982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.76701431560569e-05,
+      "loss": 0.4252,
+      "step": 4983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7669218386369304e-05,
+      "loss": 0.6445,
+      "step": 4984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.766829344060533e-05,
+      "loss": 1.3421,
+      "step": 4985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7667368318770863e-05,
+      "loss": 0.2338,
+      "step": 4986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7666443020871785e-05,
+      "loss": 1.4329,
+      "step": 4987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7665517546913994e-05,
+      "loss": 0.7924,
+      "step": 4988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7664591896903366e-05,
+      "loss": 1.0294,
+      "step": 4989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.766366607084581e-05,
+      "loss": 1.0354,
+      "step": 4990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7662740068747195e-05,
+      "loss": 0.7682,
+      "step": 4991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7661813890613436e-05,
+      "loss": 1.3985,
+      "step": 4992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7660887536450414e-05,
+      "loss": 0.909,
+      "step": 4993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765996100626402e-05,
+      "loss": 1.8707,
+      "step": 4994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7659034300060166e-05,
+      "loss": 0.9548,
+      "step": 4995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765810741784473e-05,
+      "loss": 0.6611,
+      "step": 4996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7657180359623626e-05,
+      "loss": 1.6092,
+      "step": 4997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765625312540275e-05,
+      "loss": 0.3906,
+      "step": 4998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765532571518799e-05,
+      "loss": 0.4726,
+      "step": 4999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765439812898527e-05,
+      "loss": 0.4486,
+      "step": 5000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765347036680047e-05,
+      "loss": 0.9736,
+      "step": 5001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765254242863952e-05,
+      "loss": 1.1078,
+      "step": 5002
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7651614314508294e-05,
+      "loss": 1.3432,
+      "step": 5003
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7650686024412725e-05,
+      "loss": 1.0822,
+      "step": 5004
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.76497575583587e-05,
+      "loss": 0.5394,
+      "step": 5005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.764882891635214e-05,
+      "loss": 1.0446,
+      "step": 5006
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.764790009839895e-05,
+      "loss": 0.8231,
+      "step": 5007
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.764697110450504e-05,
+      "loss": 1.4414,
+      "step": 5008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.764604193467632e-05,
+      "loss": 0.9528,
+      "step": 5009
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.764511258891872e-05,
+      "loss": 0.52,
+      "step": 5010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.764418306723813e-05,
+      "loss": 0.312,
+      "step": 5011
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.764325336964048e-05,
+      "loss": 0.9049,
+      "step": 5012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.764232349613168e-05,
+      "loss": 1.3152,
+      "step": 5013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.764139344671764e-05,
+      "loss": 0.6079,
+      "step": 5014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.76404632214043e-05,
+      "loss": 1.4229,
+      "step": 5015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.763953282019756e-05,
+      "loss": 0.8083,
+      "step": 5016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.763860224310335e-05,
+      "loss": 0.1521,
+      "step": 5017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7637671490127584e-05,
+      "loss": 2.2171,
+      "step": 5018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7636740561276204e-05,
+      "loss": 0.6658,
+      "step": 5019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.763580945655511e-05,
+      "loss": 1.0195,
+      "step": 5020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.763487817597024e-05,
+      "loss": 1.1228,
+      "step": 5021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.763394671952752e-05,
+      "loss": 0.1929,
+      "step": 5022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.763301508723287e-05,
+      "loss": 1.1595,
+      "step": 5023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7632083279092233e-05,
+      "loss": 1.5415,
+      "step": 5024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.763115129511152e-05,
+      "loss": 0.4868,
+      "step": 5025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7630219135296676e-05,
+      "loss": 0.2463,
+      "step": 5026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7629286799653626e-05,
+      "loss": 0.8982,
+      "step": 5027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7628354288188316e-05,
+      "loss": 0.5718,
+      "step": 5028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7627421600906654e-05,
+      "loss": 1.3579,
+      "step": 5029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.76264887378146e-05,
+      "loss": 1.0292,
+      "step": 5030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7625555698918086e-05,
+      "loss": 0.7215,
+      "step": 5031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7624622484223044e-05,
+      "loss": 0.8515,
+      "step": 5032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.76236890937354e-05,
+      "loss": 0.4018,
+      "step": 5033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.762275552746112e-05,
+      "loss": 0.7852,
+      "step": 5034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7621821785406125e-05,
+      "loss": 0.8766,
+      "step": 5035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.762088786757637e-05,
+      "loss": 0.1946,
+      "step": 5036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7619953773977784e-05,
+      "loss": 0.2827,
+      "step": 5037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7619019504616325e-05,
+      "loss": 0.459,
+      "step": 5038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7618085059497934e-05,
+      "loss": 1.6701,
+      "step": 5039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7617150438628556e-05,
+      "loss": 1.0389,
+      "step": 5040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.761621564201414e-05,
+      "loss": 0.9136,
+      "step": 5041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7615280669660625e-05,
+      "loss": 0.5971,
+      "step": 5042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.761434552157398e-05,
+      "loss": 1.6564,
+      "step": 5043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.761341019776014e-05,
+      "loss": 1.4292,
+      "step": 5044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.761247469822506e-05,
+      "loss": 0.7629,
+      "step": 5045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.76115390229747e-05,
+      "loss": 1.6347,
+      "step": 5046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7610603172015e-05,
+      "loss": 1.0636,
+      "step": 5047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.760966714535194e-05,
+      "loss": 0.8174,
+      "step": 5048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.760873094299145e-05,
+      "loss": 0.186,
+      "step": 5049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7607794564939505e-05,
+      "loss": 0.7599,
+      "step": 5050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7606858011202056e-05,
+      "loss": 0.9816,
+      "step": 5051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.760592128178507e-05,
+      "loss": 1.5171,
+      "step": 5052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.760498437669449e-05,
+      "loss": 0.0942,
+      "step": 5053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.76040472959363e-05,
+      "loss": 1.6352,
+      "step": 5054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.760311003951645e-05,
+      "loss": 0.9512,
+      "step": 5055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7602172607440915e-05,
+      "loss": 0.4512,
+      "step": 5056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.760123499971564e-05,
+      "loss": 1.3386,
+      "step": 5057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.760029721634662e-05,
+      "loss": 0.4706,
+      "step": 5058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.75993592573398e-05,
+      "loss": 1.1595,
+      "step": 5059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7598421122701164e-05,
+      "loss": 0.5836,
+      "step": 5060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7597482812436675e-05,
+      "loss": 0.4987,
+      "step": 5061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.75965443265523e-05,
+      "loss": 1.0494,
+      "step": 5062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.759560566505401e-05,
+      "loss": 1.2527,
+      "step": 5063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.759466682794779e-05,
+      "loss": 0.4006,
+      "step": 5064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.759372781523961e-05,
+      "loss": 1.2279,
+      "step": 5065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7592788626935436e-05,
+      "loss": 0.4716,
+      "step": 5066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.759184926304126e-05,
+      "loss": 1.0103,
+      "step": 5067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.759090972356305e-05,
+      "loss": 1.2849,
+      "step": 5068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.758997000850678e-05,
+      "loss": 1.1956,
+      "step": 5069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.758903011787844e-05,
+      "loss": 0.4327,
+      "step": 5070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.758809005168401e-05,
+      "loss": 0.6985,
+      "step": 5071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.758714980992947e-05,
+      "loss": 0.1752,
+      "step": 5072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7586209392620804e-05,
+      "loss": 0.796,
+      "step": 5073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.758526879976399e-05,
+      "loss": 1.1676,
+      "step": 5074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.758432803136503e-05,
+      "loss": 1.5201,
+      "step": 5075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.758338708742989e-05,
+      "loss": 0.8356,
+      "step": 5076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.758244596796457e-05,
+      "loss": 0.4342,
+      "step": 5077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.758150467297506e-05,
+      "loss": 1.5442,
+      "step": 5078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7580563202467344e-05,
+      "loss": 0.5475,
+      "step": 5079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.757962155644741e-05,
+      "loss": 1.0952,
+      "step": 5080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.757867973492127e-05,
+      "loss": 1.5005,
+      "step": 5081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.75777377378949e-05,
+      "loss": 0.5011,
+      "step": 5082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7576795565374286e-05,
+      "loss": 1.3398,
+      "step": 5083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.757585321736545e-05,
+      "loss": 1.4767,
+      "step": 5084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7574910693874365e-05,
+      "loss": 0.9357,
+      "step": 5085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.757396799490705e-05,
+      "loss": 0.4791,
+      "step": 5086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.757302512046948e-05,
+      "loss": 0.3293,
+      "step": 5087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.757208207056767e-05,
+      "loss": 0.7206,
+      "step": 5088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.757113884520762e-05,
+      "loss": 0.3994,
+      "step": 5089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7570195444395334e-05,
+      "loss": 1.277,
+      "step": 5090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.756925186813681e-05,
+      "loss": 1.4609,
+      "step": 5091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.756830811643806e-05,
+      "loss": 0.7356,
+      "step": 5092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7567364189305076e-05,
+      "loss": 0.2459,
+      "step": 5093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.756642008674387e-05,
+      "loss": 0.9969,
+      "step": 5094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.756547580876047e-05,
+      "loss": 0.7941,
+      "step": 5095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.756453135536086e-05,
+      "loss": 0.7048,
+      "step": 5096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7563586726551056e-05,
+      "loss": 0.3802,
+      "step": 5097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.756264192233708e-05,
+      "loss": 1.1052,
+      "step": 5098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.756169694272492e-05,
+      "loss": 0.4648,
+      "step": 5099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7560751787720616e-05,
+      "loss": 0.7399,
+      "step": 5100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.755980645733017e-05,
+      "loss": 1.2383,
+      "step": 5101
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.755886095155961e-05,
+      "loss": 0.7643,
+      "step": 5102
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.755791527041493e-05,
+      "loss": 0.6845,
+      "step": 5103
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.755696941390217e-05,
+      "loss": 0.1658,
+      "step": 5104
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7556023382027335e-05,
+      "loss": 0.969,
+      "step": 5105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.755507717479645e-05,
+      "loss": 1.596,
+      "step": 5106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.755413079221553e-05,
+      "loss": 1.645,
+      "step": 5107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.755318423429061e-05,
+      "loss": 1.5731,
+      "step": 5108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.755223750102771e-05,
+      "loss": 1.4538,
+      "step": 5109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7551290592432846e-05,
+      "loss": 0.7246,
+      "step": 5110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.755034350851205e-05,
+      "loss": 1.4946,
+      "step": 5111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.754939624927135e-05,
+      "loss": 1.3124,
+      "step": 5112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.754844881471678e-05,
+      "loss": 0.9159,
+      "step": 5113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7547501204854356e-05,
+      "loss": 0.4296,
+      "step": 5114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.754655341969012e-05,
+      "loss": 1.4805,
+      "step": 5115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.754560545923009e-05,
+      "loss": 1.5005,
+      "step": 5116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7544657323480304e-05,
+      "loss": 0.847,
+      "step": 5117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7543709012446804e-05,
+      "loss": 0.4244,
+      "step": 5118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7542760526135604e-05,
+      "loss": 0.7499,
+      "step": 5119
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.754181186455277e-05,
+      "loss": 0.907,
+      "step": 5120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.754086302770432e-05,
+      "loss": 0.5179,
+      "step": 5121
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.753991401559629e-05,
+      "loss": 0.1907,
+      "step": 5122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7538964828234725e-05,
+      "loss": 1.2311,
+      "step": 5123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7538015465625664e-05,
+      "loss": 0.8646,
+      "step": 5124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.753706592777515e-05,
+      "loss": 0.4851,
+      "step": 5125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7536116214689225e-05,
+      "loss": 1.3596,
+      "step": 5126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.753516632637393e-05,
+      "loss": 0.2206,
+      "step": 5127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.753421626283531e-05,
+      "loss": 1.467,
+      "step": 5128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.753326602407942e-05,
+      "loss": 0.3642,
+      "step": 5129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7532315610112295e-05,
+      "loss": 0.6297,
+      "step": 5130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.753136502093998e-05,
+      "loss": 1.2405,
+      "step": 5131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7530414256568535e-05,
+      "loss": 0.8759,
+      "step": 5132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.752946331700401e-05,
+      "loss": 0.9191,
+      "step": 5133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.752851220225244e-05,
+      "loss": 1.4759,
+      "step": 5134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7527560912319914e-05,
+      "loss": 0.4893,
+      "step": 5135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.752660944721245e-05,
+      "loss": 0.1707,
+      "step": 5136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7525657806936115e-05,
+      "loss": 0.3747,
+      "step": 5137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7524705991496966e-05,
+      "loss": 1.3061,
+      "step": 5138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.752375400090106e-05,
+      "loss": 1.1225,
+      "step": 5139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7522801835154456e-05,
+      "loss": 0.2298,
+      "step": 5140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.75218494942632e-05,
+      "loss": 1.4647,
+      "step": 5141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.752089697823338e-05,
+      "loss": 0.7321,
+      "step": 5142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.751994428707103e-05,
+      "loss": 1.3454,
+      "step": 5143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.751899142078222e-05,
+      "loss": 1.2707,
+      "step": 5144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.751803837937304e-05,
+      "loss": 1.0334,
+      "step": 5145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.751708516284951e-05,
+      "loss": 1.6906,
+      "step": 5146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.751613177121772e-05,
+      "loss": 0.7854,
+      "step": 5147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.751517820448374e-05,
+      "loss": 0.6072,
+      "step": 5148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7514224462653636e-05,
+      "loss": 0.3824,
+      "step": 5149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.751327054573347e-05,
+      "loss": 1.183,
+      "step": 5150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.751231645372932e-05,
+      "loss": 0.5085,
+      "step": 5151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.751136218664725e-05,
+      "loss": 1.1385,
+      "step": 5152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.751040774449334e-05,
+      "loss": 0.614,
+      "step": 5153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.750945312727366e-05,
+      "loss": 1.0926,
+      "step": 5154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.750849833499429e-05,
+      "loss": 1.3532,
+      "step": 5155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7507543367661294e-05,
+      "loss": 1.2455,
+      "step": 5156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.750658822528076e-05,
+      "loss": 1.0926,
+      "step": 5157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.750563290785876e-05,
+      "loss": 0.5375,
+      "step": 5158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.750467741540138e-05,
+      "loss": 1.3504,
+      "step": 5159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.750372174791469e-05,
+      "loss": 0.5499,
+      "step": 5160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7502765905404785e-05,
+      "loss": 0.5532,
+      "step": 5161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.750180988787773e-05,
+      "loss": 1.0962,
+      "step": 5162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.750085369533963e-05,
+      "loss": 0.6333,
+      "step": 5163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.749989732779655e-05,
+      "loss": 0.5717,
+      "step": 5164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.749894078525459e-05,
+      "loss": 0.4978,
+      "step": 5165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.749798406771984e-05,
+      "loss": 0.3178,
+      "step": 5166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7497027175198366e-05,
+      "loss": 0.2397,
+      "step": 5167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.749607010769628e-05,
+      "loss": 1.4312,
+      "step": 5168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7495112865219663e-05,
+      "loss": 0.4739,
+      "step": 5169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.74941554477746e-05,
+      "loss": 1.1106,
+      "step": 5170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.74931978553672e-05,
+      "loss": 0.6613,
+      "step": 5171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.749224008800354e-05,
+      "loss": 0.8123,
+      "step": 5172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7491282145689735e-05,
+      "loss": 0.3222,
+      "step": 5173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.749032402843186e-05,
+      "loss": 0.4851,
+      "step": 5174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.748936573623602e-05,
+      "loss": 1.0068,
+      "step": 5175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.748840726910832e-05,
+      "loss": 0.8183,
+      "step": 5176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7487448627054844e-05,
+      "loss": 1.3195,
+      "step": 5177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.748648981008171e-05,
+      "loss": 0.8792,
+      "step": 5178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7485530818195005e-05,
+      "loss": 0.9176,
+      "step": 5179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.748457165140084e-05,
+      "loss": 0.6012,
+      "step": 5180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.748361230970532e-05,
+      "loss": 0.5491,
+      "step": 5181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7482652793114536e-05,
+      "loss": 1.9012,
+      "step": 5182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.748169310163462e-05,
+      "loss": 0.8786,
+      "step": 5183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.748073323527165e-05,
+      "loss": 0.3814,
+      "step": 5184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.747977319403175e-05,
+      "loss": 2.0822,
+      "step": 5185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7478812977921034e-05,
+      "loss": 1.0314,
+      "step": 5186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.747785258694561e-05,
+      "loss": 1.517,
+      "step": 5187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7476892021111575e-05,
+      "loss": 0.5671,
+      "step": 5188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.747593128042506e-05,
+      "loss": 0.5629,
+      "step": 5189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.747497036489216e-05,
+      "loss": 0.4282,
+      "step": 5190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7474009274519004e-05,
+      "loss": 0.6205,
+      "step": 5191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7473048009311706e-05,
+      "loss": 0.3221,
+      "step": 5192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7472086569276387e-05,
+      "loss": 1.0929,
+      "step": 5193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.747112495441915e-05,
+      "loss": 0.6129,
+      "step": 5194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7470163164746136e-05,
+      "loss": 0.5697,
+      "step": 5195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.746920120026344e-05,
+      "loss": 0.8895,
+      "step": 5196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.746823906097722e-05,
+      "loss": 0.1982,
+      "step": 5197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.746727674689355e-05,
+      "loss": 0.5806,
+      "step": 5198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7466314258018596e-05,
+      "loss": 1.4763,
+      "step": 5199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7465351594358465e-05,
+      "loss": 1.3875,
+      "step": 5200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.746438875591928e-05,
+      "loss": 0.8067,
+      "step": 5201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.746342574270718e-05,
+      "loss": 1.1169,
+      "step": 5202
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.746246255472828e-05,
+      "loss": 0.5057,
+      "step": 5203
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7461499191988714e-05,
+      "loss": 1.471,
+      "step": 5204
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.746053565449461e-05,
+      "loss": 1.3069,
+      "step": 5205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.745957194225211e-05,
+      "loss": 0.9134,
+      "step": 5206
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.745860805526734e-05,
+      "loss": 0.765,
+      "step": 5207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.745764399354644e-05,
+      "loss": 0.4046,
+      "step": 5208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7456679757095535e-05,
+      "loss": 1.146,
+      "step": 5209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.745571534592076e-05,
+      "loss": 1.4679,
+      "step": 5210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.745475076002826e-05,
+      "loss": 1.3294,
+      "step": 5211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.745378599942417e-05,
+      "loss": 1.0397,
+      "step": 5212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.745282106411463e-05,
+      "loss": 0.5218,
+      "step": 5213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.745185595410579e-05,
+      "loss": 0.7614,
+      "step": 5214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7450890669403764e-05,
+      "loss": 1.1768,
+      "step": 5215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.744992521001473e-05,
+      "loss": 0.4899,
+      "step": 5216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.74489595759448e-05,
+      "loss": 0.3363,
+      "step": 5217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.744799376720014e-05,
+      "loss": 0.4786,
+      "step": 5218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7447027783786896e-05,
+      "loss": 0.5619,
+      "step": 5219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7446061625711206e-05,
+      "loss": 1.4779,
+      "step": 5220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.744509529297921e-05,
+      "loss": 0.6889,
+      "step": 5221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7444128785597076e-05,
+      "loss": 1.3041,
+      "step": 5222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.744316210357095e-05,
+      "loss": 0.2338,
+      "step": 5223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.744219524690697e-05,
+      "loss": 0.4597,
+      "step": 5224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7441228215611306e-05,
+      "loss": 0.3761,
+      "step": 5225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7440261009690104e-05,
+      "loss": 0.2165,
+      "step": 5226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7439293629149515e-05,
+      "loss": 1.196,
+      "step": 5227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.743832607399571e-05,
+      "loss": 1.1989,
+      "step": 5228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.743735834423483e-05,
+      "loss": 0.7689,
+      "step": 5229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7436390439873036e-05,
+      "loss": 1.7019,
+      "step": 5230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.74354223609165e-05,
+      "loss": 0.8981,
+      "step": 5231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.743445410737136e-05,
+      "loss": 1.6555,
+      "step": 5232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.743348567924381e-05,
+      "loss": 1.2861,
+      "step": 5233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.743251707653997e-05,
+      "loss": 0.2468,
+      "step": 5234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.743154829926604e-05,
+      "loss": 1.1048,
+      "step": 5235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.743057934742817e-05,
+      "loss": 1.1077,
+      "step": 5236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7429610221032526e-05,
+      "loss": 0.3837,
+      "step": 5237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.742864092008529e-05,
+      "loss": 1.4546,
+      "step": 5238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7427671444592605e-05,
+      "loss": 0.7057,
+      "step": 5239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.742670179456066e-05,
+      "loss": 0.6651,
+      "step": 5240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.742573196999561e-05,
+      "loss": 0.9309,
+      "step": 5241
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7424761970903634e-05,
+      "loss": 1.2617,
+      "step": 5242
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.742379179729091e-05,
+      "loss": 0.8255,
+      "step": 5243
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.742282144916361e-05,
+      "loss": 0.7807,
+      "step": 5244
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.742185092652791e-05,
+      "loss": 0.6661,
+      "step": 5245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.742088022938997e-05,
+      "loss": 1.1492,
+      "step": 5246
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.741990935775599e-05,
+      "loss": 1.3604,
+      "step": 5247
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.741893831163213e-05,
+      "loss": 0.6377,
+      "step": 5248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.741796709102458e-05,
+      "loss": 0.7951,
+      "step": 5249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7416995695939524e-05,
+      "loss": 0.898,
+      "step": 5250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7416024126383136e-05,
+      "loss": 0.3118,
+      "step": 5251
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7415052382361596e-05,
+      "loss": 1.6568,
+      "step": 5252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.741408046388109e-05,
+      "loss": 0.5093,
+      "step": 5253
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7413108370947816e-05,
+      "loss": 1.5717,
+      "step": 5254
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.741213610356793e-05,
+      "loss": 0.4469,
+      "step": 5255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.741116366174766e-05,
+      "loss": 1.7029,
+      "step": 5256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7410191045493154e-05,
+      "loss": 0.5213,
+      "step": 5257
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7409218254810626e-05,
+      "loss": 0.9326,
+      "step": 5258
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.740824528970626e-05,
+      "loss": 0.3739,
+      "step": 5259
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.740727215018625e-05,
+      "loss": 0.512,
+      "step": 5260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7406298836256785e-05,
+      "loss": 0.2063,
+      "step": 5261
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7405325347924054e-05,
+      "loss": 1.3163,
+      "step": 5262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.740435168519427e-05,
+      "loss": 0.8674,
+      "step": 5263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.74033778480736e-05,
+      "loss": 0.952,
+      "step": 5264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7402403836568275e-05,
+      "loss": 0.3169,
+      "step": 5265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.740142965068447e-05,
+      "loss": 0.8278,
+      "step": 5266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.740045529042839e-05,
+      "loss": 1.3784,
+      "step": 5267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7399480755806235e-05,
+      "loss": 0.6364,
+      "step": 5268
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.739850604682421e-05,
+      "loss": 0.8567,
+      "step": 5269
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.739753116348852e-05,
+      "loss": 1.3387,
+      "step": 5270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.739655610580536e-05,
+      "loss": 1.0708,
+      "step": 5271
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.739558087378093e-05,
+      "loss": 1.5922,
+      "step": 5272
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.739460546742146e-05,
+      "loss": 0.7124,
+      "step": 5273
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7393629886733136e-05,
+      "loss": 0.4345,
+      "step": 5274
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.739265413172217e-05,
+      "loss": 0.6896,
+      "step": 5275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.739167820239478e-05,
+      "loss": 1.0459,
+      "step": 5276
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.739070209875717e-05,
+      "loss": 0.4152,
+      "step": 5277
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.738972582081555e-05,
+      "loss": 0.6307,
+      "step": 5278
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.738874936857612e-05,
+      "loss": 1.1596,
+      "step": 5279
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.738777274204513e-05,
+      "loss": 0.7807,
+      "step": 5280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.738679594122876e-05,
+      "loss": 0.9352,
+      "step": 5281
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7385818966133245e-05,
+      "loss": 0.4838,
+      "step": 5282
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.738484181676479e-05,
+      "loss": 0.3394,
+      "step": 5283
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7383864493129625e-05,
+      "loss": 0.479,
+      "step": 5284
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.738288699523396e-05,
+      "loss": 1.2525,
+      "step": 5285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.738190932308402e-05,
+      "loss": 0.6336,
+      "step": 5286
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7380931476686036e-05,
+      "loss": 1.4065,
+      "step": 5287
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7379953456046206e-05,
+      "loss": 0.5024,
+      "step": 5288
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.737897526117077e-05,
+      "loss": 0.5721,
+      "step": 5289
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7377996892065956e-05,
+      "loss": 1.0342,
+      "step": 5290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7377018348737985e-05,
+      "loss": 0.517,
+      "step": 5291
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.737603963119308e-05,
+      "loss": 1.0513,
+      "step": 5292
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.737506073943748e-05,
+      "loss": 1.7179,
+      "step": 5293
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7374081673477406e-05,
+      "loss": 0.4172,
+      "step": 5294
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7373102433319084e-05,
+      "loss": 1.4269,
+      "step": 5295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7372123018968756e-05,
+      "loss": 0.2776,
+      "step": 5296
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.737114343043264e-05,
+      "loss": 1.389,
+      "step": 5297
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.737016366771699e-05,
+      "loss": 1.2982,
+      "step": 5298
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.736918373082803e-05,
+      "loss": 1.1387,
+      "step": 5299
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7368203619771994e-05,
+      "loss": 1.0778,
+      "step": 5300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.736722333455512e-05,
+      "loss": 1.3609,
+      "step": 5301
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7366242875183655e-05,
+      "loss": 0.4157,
+      "step": 5302
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7365262241663827e-05,
+      "loss": 0.5608,
+      "step": 5303
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.736428143400188e-05,
+      "loss": 0.2312,
+      "step": 5304
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.736330045220406e-05,
+      "loss": 0.4858,
+      "step": 5305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.736231929627659e-05,
+      "loss": 0.9325,
+      "step": 5306
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.736133796622575e-05,
+      "loss": 0.3019,
+      "step": 5307
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7360356462057754e-05,
+      "loss": 0.6096,
+      "step": 5308
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.735937478377886e-05,
+      "loss": 1.2555,
+      "step": 5309
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.73583929313953e-05,
+      "loss": 1.3346,
+      "step": 5310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.735741090491335e-05,
+      "loss": 0.2781,
+      "step": 5311
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.735642870433924e-05,
+      "loss": 1.4316,
+      "step": 5312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7355446329679225e-05,
+      "loss": 1.5764,
+      "step": 5313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7354463780939556e-05,
+      "loss": 0.5442,
+      "step": 5314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7353481058126484e-05,
+      "loss": 0.7837,
+      "step": 5315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7352498161246265e-05,
+      "loss": 1.3522,
+      "step": 5316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.735151509030516e-05,
+      "loss": 0.762,
+      "step": 5317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.735053184530941e-05,
+      "loss": 1.3382,
+      "step": 5318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.734954842626528e-05,
+      "loss": 1.181,
+      "step": 5319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.734856483317903e-05,
+      "loss": 1.2884,
+      "step": 5320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7347581066056916e-05,
+      "loss": 0.9174,
+      "step": 5321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.73465971249052e-05,
+      "loss": 0.3203,
+      "step": 5322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.734561300973014e-05,
+      "loss": 0.7196,
+      "step": 5323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.734462872053802e-05,
+      "loss": 1.2237,
+      "step": 5324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7343644257335065e-05,
+      "loss": 1.1445,
+      "step": 5325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.734265962012757e-05,
+      "loss": 0.3809,
+      "step": 5326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7341674808921786e-05,
+      "loss": 0.6845,
+      "step": 5327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7340689823723984e-05,
+      "loss": 1.0162,
+      "step": 5328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.733970466454043e-05,
+      "loss": 0.5561,
+      "step": 5329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.733871933137741e-05,
+      "loss": 0.8231,
+      "step": 5330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.733773382424117e-05,
+      "loss": 0.493,
+      "step": 5331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.733674814313799e-05,
+      "loss": 1.2845,
+      "step": 5332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7335762288074154e-05,
+      "loss": 1.1984,
+      "step": 5333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.733477625905592e-05,
+      "loss": 1.2675,
+      "step": 5334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.733379005608956e-05,
+      "loss": 0.3411,
+      "step": 5335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7332803679181364e-05,
+      "loss": 0.4623,
+      "step": 5336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.73318171283376e-05,
+      "loss": 0.5768,
+      "step": 5337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7330830403564564e-05,
+      "loss": 0.9943,
+      "step": 5338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.732984350486851e-05,
+      "loss": 0.8107,
+      "step": 5339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.732885643225573e-05,
+      "loss": 0.5727,
+      "step": 5340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7327869185732505e-05,
+      "loss": 1.5054,
+      "step": 5341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.732688176530511e-05,
+      "loss": 0.8214,
+      "step": 5342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7325894170979844e-05,
+      "loss": 0.9082,
+      "step": 5343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.732490640276298e-05,
+      "loss": 0.9414,
+      "step": 5344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7323918460660804e-05,
+      "loss": 0.7423,
+      "step": 5345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.73229303446796e-05,
+      "loss": 0.7322,
+      "step": 5346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.732194205482568e-05,
+      "loss": 0.9761,
+      "step": 5347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7320953591105304e-05,
+      "loss": 0.9507,
+      "step": 5348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.731996495352478e-05,
+      "loss": 1.4479,
+      "step": 5349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.731897614209038e-05,
+      "loss": 0.3216,
+      "step": 5350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7317987156808404e-05,
+      "loss": 1.0809,
+      "step": 5351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.731699799768516e-05,
+      "loss": 1.2037,
+      "step": 5352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7316008664726934e-05,
+      "loss": 1.1168,
+      "step": 5353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.731501915794002e-05,
+      "loss": 0.5549,
+      "step": 5354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.731402947733071e-05,
+      "loss": 0.5056,
+      "step": 5355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.731303962290532e-05,
+      "loss": 0.3824,
+      "step": 5356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.731204959467012e-05,
+      "loss": 0.6601,
+      "step": 5357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.731105939263143e-05,
+      "loss": 0.5515,
+      "step": 5358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.731006901679555e-05,
+      "loss": 0.4723,
+      "step": 5359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.730907846716878e-05,
+      "loss": 0.8134,
+      "step": 5360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7308087743757424e-05,
+      "loss": 1.2668,
+      "step": 5361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.730709684656779e-05,
+      "loss": 0.9245,
+      "step": 5362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.730610577560617e-05,
+      "loss": 1.0989,
+      "step": 5363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7305114530878895e-05,
+      "loss": 0.3165,
+      "step": 5364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.730412311239225e-05,
+      "loss": 0.6695,
+      "step": 5365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7303131520152553e-05,
+      "loss": 0.4067,
+      "step": 5366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.730213975416611e-05,
+      "loss": 1.3367,
+      "step": 5367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.730114781443924e-05,
+      "loss": 0.635,
+      "step": 5368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.730015570097825e-05,
+      "loss": 1.1641,
+      "step": 5369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729916341378946e-05,
+      "loss": 1.8184,
+      "step": 5370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729817095287917e-05,
+      "loss": 1.0689,
+      "step": 5371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729717831825371e-05,
+      "loss": 0.616,
+      "step": 5372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729618550991939e-05,
+      "loss": 0.7754,
+      "step": 5373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729519252788253e-05,
+      "loss": 0.5778,
+      "step": 5374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729419937214945e-05,
+      "loss": 1.302,
+      "step": 5375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729320604272647e-05,
+      "loss": 0.7887,
+      "step": 5376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.72922125396199e-05,
+      "loss": 1.0606,
+      "step": 5377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729121886283608e-05,
+      "loss": 0.3292,
+      "step": 5378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729022501238133e-05,
+      "loss": 0.5953,
+      "step": 5379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.728923098826196e-05,
+      "loss": 0.3727,
+      "step": 5380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7288236790484306e-05,
+      "loss": 0.3981,
+      "step": 5381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7287242419054695e-05,
+      "loss": 1.004,
+      "step": 5382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.728624787397946e-05,
+      "loss": 0.6403,
+      "step": 5383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7285253155264924e-05,
+      "loss": 1.2202,
+      "step": 5384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.728425826291741e-05,
+      "loss": 1.0034,
+      "step": 5385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.728326319694325e-05,
+      "loss": 0.4511,
+      "step": 5386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.72822679573488e-05,
+      "loss": 0.7494,
+      "step": 5387
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.728127254414036e-05,
+      "loss": 0.5985,
+      "step": 5388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.728027695732429e-05,
+      "loss": 0.7684,
+      "step": 5389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.727928119690691e-05,
+      "loss": 1.2555,
+      "step": 5390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.727828526289456e-05,
+      "loss": 0.8953,
+      "step": 5391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.727728915529358e-05,
+      "loss": 1.377,
+      "step": 5392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.727629287411032e-05,
+      "loss": 0.4294,
+      "step": 5393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.727529641935109e-05,
+      "loss": 0.5936,
+      "step": 5394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.727429979102227e-05,
+      "loss": 0.6275,
+      "step": 5395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.727330298913017e-05,
+      "loss": 0.3373,
+      "step": 5396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.727230601368114e-05,
+      "loss": 0.9276,
+      "step": 5397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.727130886468154e-05,
+      "loss": 0.5001,
+      "step": 5398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7270311542137695e-05,
+      "loss": 1.402,
+      "step": 5399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7269314046055976e-05,
+      "loss": 0.4343,
+      "step": 5400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7268316376442705e-05,
+      "loss": 0.5003,
+      "step": 5401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7267318533304245e-05,
+      "loss": 1.0835,
+      "step": 5402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.726632051664694e-05,
+      "loss": 0.2552,
+      "step": 5403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.726532232647715e-05,
+      "loss": 0.7406,
+      "step": 5404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.726432396280121e-05,
+      "loss": 0.779,
+      "step": 5405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.72633254256255e-05,
+      "loss": 0.9489,
+      "step": 5406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.726232671495635e-05,
+      "loss": 0.6779,
+      "step": 5407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7261327830800124e-05,
+      "loss": 1.0155,
+      "step": 5408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.726032877316317e-05,
+      "loss": 1.3712,
+      "step": 5409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7259329542051865e-05,
+      "loss": 0.2887,
+      "step": 5410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7258330137472555e-05,
+      "loss": 1.6252,
+      "step": 5411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7257330559431604e-05,
+      "loss": 1.4843,
+      "step": 5412
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.725633080793536e-05,
+      "loss": 1.6219,
+      "step": 5413
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.725533088299021e-05,
+      "loss": 0.5592,
+      "step": 5414
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.725433078460249e-05,
+      "loss": 1.1503,
+      "step": 5415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.725333051277859e-05,
+      "loss": 0.5382,
+      "step": 5416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.725233006752485e-05,
+      "loss": 0.2903,
+      "step": 5417
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7251329448847656e-05,
+      "loss": 1.0825,
+      "step": 5418
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.725032865675337e-05,
+      "loss": 0.9619,
+      "step": 5419
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.724932769124835e-05,
+      "loss": 1.7268,
+      "step": 5420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.724832655233898e-05,
+      "loss": 0.5806,
+      "step": 5421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7247325240031626e-05,
+      "loss": 0.3407,
+      "step": 5422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.724632375433266e-05,
+      "loss": 1.0794,
+      "step": 5423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7245322095248454e-05,
+      "loss": 0.7496,
+      "step": 5424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.724432026278538e-05,
+      "loss": 0.7125,
+      "step": 5425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.724331825694981e-05,
+      "loss": 0.3637,
+      "step": 5426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.724231607774815e-05,
+      "loss": 0.3754,
+      "step": 5427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.724131372518673e-05,
+      "loss": 0.4595,
+      "step": 5428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7240311199271966e-05,
+      "loss": 0.4615,
+      "step": 5429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.723930850001022e-05,
+      "loss": 1.1547,
+      "step": 5430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.723830562740787e-05,
+      "loss": 1.1979,
+      "step": 5431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7237302581471314e-05,
+      "loss": 0.4228,
+      "step": 5432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7236299362206925e-05,
+      "loss": 1.0828,
+      "step": 5433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.723529596962108e-05,
+      "loss": 0.9788,
+      "step": 5434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.723429240372018e-05,
+      "loss": 1.3334,
+      "step": 5435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.723328866451061e-05,
+      "loss": 1.1388,
+      "step": 5436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.723228475199874e-05,
+      "loss": 0.7258,
+      "step": 5437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.723128066619097e-05,
+      "loss": 0.753,
+      "step": 5438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.723027640709369e-05,
+      "loss": 0.8152,
+      "step": 5439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.72292719747133e-05,
+      "loss": 0.6176,
+      "step": 5440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.722826736905618e-05,
+      "loss": 0.8456,
+      "step": 5441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.722726259012872e-05,
+      "loss": 1.2093,
+      "step": 5442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7226257637937315e-05,
+      "loss": 1.096,
+      "step": 5443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7225252512488365e-05,
+      "loss": 1.1909,
+      "step": 5444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7224247213788265e-05,
+      "loss": 0.7148,
+      "step": 5445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.722324174184341e-05,
+      "loss": 0.3738,
+      "step": 5446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.722223609666021e-05,
+      "loss": 0.2815,
+      "step": 5447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.722123027824505e-05,
+      "loss": 0.5374,
+      "step": 5448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.722022428660433e-05,
+      "loss": 1.2861,
+      "step": 5449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7219218121744464e-05,
+      "loss": 0.7391,
+      "step": 5450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.721821178367185e-05,
+      "loss": 0.5965,
+      "step": 5451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7217205272392886e-05,
+      "loss": 0.9767,
+      "step": 5452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.721619858791398e-05,
+      "loss": 0.3625,
+      "step": 5453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.721519173024155e-05,
+      "loss": 1.1092,
+      "step": 5454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7214184699381994e-05,
+      "loss": 0.7242,
+      "step": 5455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7213177495341696e-05,
+      "loss": 0.5163,
+      "step": 5456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.721217011812711e-05,
+      "loss": 1.4254,
+      "step": 5457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7211162567744624e-05,
+      "loss": 1.1584,
+      "step": 5458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.721015484420065e-05,
+      "loss": 0.5658,
+      "step": 5459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.72091469475016e-05,
+      "loss": 1.8828,
+      "step": 5460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.720813887765389e-05,
+      "loss": 1.308,
+      "step": 5461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.720713063466393e-05,
+      "loss": 1.6957,
+      "step": 5462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.720612221853815e-05,
+      "loss": 1.4021,
+      "step": 5463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.720511362928294e-05,
+      "loss": 1.067,
+      "step": 5464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7204104866904755e-05,
+      "loss": 1.041,
+      "step": 5465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7203095931409986e-05,
+      "loss": 0.9776,
+      "step": 5466
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.720208682280507e-05,
+      "loss": 0.8097,
+      "step": 5467
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.720107754109642e-05,
+      "loss": 1.0081,
+      "step": 5468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.720006808629046e-05,
+      "loss": 0.6186,
+      "step": 5469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.719905845839361e-05,
+      "loss": 1.5912,
+      "step": 5470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7198048657412304e-05,
+      "loss": 0.8953,
+      "step": 5471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7197038683352956e-05,
+      "loss": 0.7486,
+      "step": 5472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7196028536222e-05,
+      "loss": 0.6646,
+      "step": 5473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.719501821602587e-05,
+      "loss": 0.6672,
+      "step": 5474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.719400772277099e-05,
+      "loss": 0.3792,
+      "step": 5475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.71929970564638e-05,
+      "loss": 0.4935,
+      "step": 5476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.71919862171107e-05,
+      "loss": 1.3422,
+      "step": 5477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.719097520471816e-05,
+      "loss": 0.3579,
+      "step": 5478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.718996401929259e-05,
+      "loss": 1.4375,
+      "step": 5479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7188952660840434e-05,
+      "loss": 0.7536,
+      "step": 5480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.718794112936813e-05,
+      "loss": 1.4927,
+      "step": 5481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.71869294248821e-05,
+      "loss": 1.1227,
+      "step": 5482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7185917547388805e-05,
+      "loss": 0.9996,
+      "step": 5483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.718490549689467e-05,
+      "loss": 0.2974,
+      "step": 5484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7183893273406136e-05,
+      "loss": 0.6944,
+      "step": 5485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.718288087692965e-05,
+      "loss": 0.8951,
+      "step": 5486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7181868307471647e-05,
+      "loss": 0.4342,
+      "step": 5487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.718085556503858e-05,
+      "loss": 0.4309,
+      "step": 5488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7179842649636885e-05,
+      "loss": 0.3188,
+      "step": 5489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7178829561273014e-05,
+      "loss": 0.9428,
+      "step": 5490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7177816299953405e-05,
+      "loss": 1.5583,
+      "step": 5491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7176802865684514e-05,
+      "loss": 0.5776,
+      "step": 5492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.717578925847279e-05,
+      "loss": 1.5792,
+      "step": 5493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.717477547832468e-05,
+      "loss": 0.5854,
+      "step": 5494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.717376152524664e-05,
+      "loss": 0.9549,
+      "step": 5495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7172747399245115e-05,
+      "loss": 0.9187,
+      "step": 5496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7171733100326564e-05,
+      "loss": 0.2944,
+      "step": 5497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7170718628497445e-05,
+      "loss": 0.8494,
+      "step": 5498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.71697039837642e-05,
+      "loss": 0.4518,
+      "step": 5499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.71686891661333e-05,
+      "loss": 1.6186,
+      "step": 5500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7167674175611194e-05,
+      "loss": 0.84,
+      "step": 5501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7166659012204346e-05,
+      "loss": 1.5556,
+      "step": 5502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.716564367591922e-05,
+      "loss": 1.3827,
+      "step": 5503
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.716462816676227e-05,
+      "loss": 0.8374,
+      "step": 5504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.716361248473996e-05,
+      "loss": 0.6958,
+      "step": 5505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.716259662985875e-05,
+      "loss": 1.5111,
+      "step": 5506
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.716158060212511e-05,
+      "loss": 0.6608,
+      "step": 5507
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.716056440154551e-05,
+      "loss": 1.6615,
+      "step": 5508
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.71595480281264e-05,
+      "loss": 1.1886,
+      "step": 5509
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.715853148187427e-05,
+      "loss": 0.4311,
+      "step": 5510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.715751476279556e-05,
+      "loss": 0.4094,
+      "step": 5511
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.715649787089677e-05,
+      "loss": 0.2764,
+      "step": 5512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.715548080618436e-05,
+      "loss": 1.0156,
+      "step": 5513
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7154463568664793e-05,
+      "loss": 1.3476,
+      "step": 5514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.715344615834455e-05,
+      "loss": 0.4534,
+      "step": 5515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.715242857523011e-05,
+      "loss": 0.889,
+      "step": 5516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.715141081932795e-05,
+      "loss": 1.2035,
+      "step": 5517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.715039289064453e-05,
+      "loss": 0.2107,
+      "step": 5518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7149374789186344e-05,
+      "loss": 1.5279,
+      "step": 5519
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.714835651495986e-05,
+      "loss": 1.0305,
+      "step": 5520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7147338067971566e-05,
+      "loss": 0.9832,
+      "step": 5521
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.714631944822794e-05,
+      "loss": 1.2064,
+      "step": 5522
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7145300655735465e-05,
+      "loss": 1.0173,
+      "step": 5523
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7144281690500624e-05,
+      "loss": 0.7399,
+      "step": 5524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.714326255252989e-05,
+      "loss": 1.2819,
+      "step": 5525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.714224324182977e-05,
+      "loss": 0.9599,
+      "step": 5526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.714122375840674e-05,
+      "loss": 0.9305,
+      "step": 5527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7140204102267285e-05,
+      "loss": 0.6007,
+      "step": 5528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.713918427341789e-05,
+      "loss": 0.408,
+      "step": 5529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.713816427186505e-05,
+      "loss": 0.1217,
+      "step": 5530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.713714409761527e-05,
+      "loss": 1.4237,
+      "step": 5531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.713612375067502e-05,
+      "loss": 0.354,
+      "step": 5532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.71351032310508e-05,
+      "loss": 1.1894,
+      "step": 5533
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.713408253874911e-05,
+      "loss": 0.3575,
+      "step": 5534
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7133061673776434e-05,
+      "loss": 1.1832,
+      "step": 5535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.713204063613928e-05,
+      "loss": 0.5375,
+      "step": 5536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.713101942584414e-05,
+      "loss": 0.9021,
+      "step": 5537
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.712999804289752e-05,
+      "loss": 0.7339,
+      "step": 5538
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7128976487305904e-05,
+      "loss": 0.3484,
+      "step": 5539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7127954759075806e-05,
+      "loss": 1.1318,
+      "step": 5540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.712693285821372e-05,
+      "loss": 2.6407,
+      "step": 5541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.712591078472616e-05,
+      "loss": 1.2409,
+      "step": 5542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.712488853861961e-05,
+      "loss": 0.9054,
+      "step": 5543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.712386611990061e-05,
+      "loss": 1.3265,
+      "step": 5544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.712284352857563e-05,
+      "loss": 0.7057,
+      "step": 5545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.712182076465119e-05,
+      "loss": 0.3383,
+      "step": 5546
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.71207978281338e-05,
+      "loss": 1.2238,
+      "step": 5547
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7119774719029976e-05,
+      "loss": 1.4968,
+      "step": 5548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.711875143734622e-05,
+      "loss": 0.9457,
+      "step": 5549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.711772798308905e-05,
+      "loss": 0.2897,
+      "step": 5550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.711670435626497e-05,
+      "loss": 1.3448,
+      "step": 5551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.71156805568805e-05,
+      "loss": 0.397,
+      "step": 5552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.711465658494216e-05,
+      "loss": 0.4717,
+      "step": 5553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.711363244045645e-05,
+      "loss": 0.5015,
+      "step": 5554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7112608123429907e-05,
+      "loss": 0.3747,
+      "step": 5555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.711158363386904e-05,
+      "loss": 0.5755,
+      "step": 5556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.711055897178037e-05,
+      "loss": 0.549,
+      "step": 5557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7109534137170405e-05,
+      "loss": 0.8004,
+      "step": 5558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.710850913004568e-05,
+      "loss": 0.2033,
+      "step": 5559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7107483950412724e-05,
+      "loss": 1.0782,
+      "step": 5560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7106458598278044e-05,
+      "loss": 0.4822,
+      "step": 5561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7105433073648174e-05,
+      "loss": 0.4518,
+      "step": 5562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.710440737652965e-05,
+      "loss": 0.9231,
+      "step": 5563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.710338150692897e-05,
+      "loss": 1.2473,
+      "step": 5564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.71023554648527e-05,
+      "loss": 1.1316,
+      "step": 5565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7101329250307344e-05,
+      "loss": 0.5806,
+      "step": 5566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.710030286329944e-05,
+      "loss": 1.0795,
+      "step": 5567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7099276303835514e-05,
+      "loss": 1.132,
+      "step": 5568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.70982495719221e-05,
+      "loss": 0.5017,
+      "step": 5569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.709722266756574e-05,
+      "loss": 0.7667,
+      "step": 5570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.709619559077296e-05,
+      "loss": 0.6383,
+      "step": 5571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7095168341550296e-05,
+      "loss": 0.9983,
+      "step": 5572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.709414091990429e-05,
+      "loss": 0.7349,
+      "step": 5573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7093113325841485e-05,
+      "loss": 0.6384,
+      "step": 5574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.709208555936841e-05,
+      "loss": 0.4994,
+      "step": 5575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7091057620491616e-05,
+      "loss": 1.2089,
+      "step": 5576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.709002950921762e-05,
+      "loss": 0.3206,
+      "step": 5577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.708900122555299e-05,
+      "loss": 0.2144,
+      "step": 5578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.708797276950426e-05,
+      "loss": 1.3518,
+      "step": 5579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.708694414107799e-05,
+      "loss": 0.4965,
+      "step": 5580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.70859153402807e-05,
+      "loss": 1.5926,
+      "step": 5581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.708488636711895e-05,
+      "loss": 1.7128,
+      "step": 5582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.708385722159928e-05,
+      "loss": 1.0239,
+      "step": 5583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.708282790372826e-05,
+      "loss": 1.2725,
+      "step": 5584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7081798413512405e-05,
+      "loss": 0.3942,
+      "step": 5585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7080768750958306e-05,
+      "loss": 1.1826,
+      "step": 5586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.707973891607249e-05,
+      "loss": 0.7228,
+      "step": 5587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7078708908861516e-05,
+      "loss": 0.6506,
+      "step": 5588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.707767872933195e-05,
+      "loss": 0.7497,
+      "step": 5589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.707664837749033e-05,
+      "loss": 0.5559,
+      "step": 5590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.707561785334322e-05,
+      "loss": 0.4259,
+      "step": 5591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7074587156897175e-05,
+      "loss": 0.7337,
+      "step": 5592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7073556288158753e-05,
+      "loss": 0.8813,
+      "step": 5593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.707252524713452e-05,
+      "loss": 0.9626,
+      "step": 5594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7071494033831046e-05,
+      "loss": 0.6137,
+      "step": 5595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.707046264825488e-05,
+      "loss": 0.5052,
+      "step": 5596
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7069431090412586e-05,
+      "loss": 1.6634,
+      "step": 5597
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.706839936031073e-05,
+      "loss": 0.3044,
+      "step": 5598
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7067367457955876e-05,
+      "loss": 1.2811,
+      "step": 5599
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7066335383354585e-05,
+      "loss": 0.2438,
+      "step": 5600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.706530313651344e-05,
+      "loss": 0.6855,
+      "step": 5601
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.706427071743901e-05,
+      "loss": 0.4839,
+      "step": 5602
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7063238126137846e-05,
+      "loss": 0.8254,
+      "step": 5603
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.706220536261653e-05,
+      "loss": 1.2558,
+      "step": 5604
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.706117242688164e-05,
+      "loss": 0.6644,
+      "step": 5605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.706013931893974e-05,
+      "loss": 0.8314,
+      "step": 5606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.705910603879741e-05,
+      "loss": 1.2764,
+      "step": 5607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.705807258646122e-05,
+      "loss": 0.3476,
+      "step": 5608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.705703896193775e-05,
+      "loss": 2.5087,
+      "step": 5609
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.705600516523358e-05,
+      "loss": 0.7595,
+      "step": 5610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.705497119635528e-05,
+      "loss": 1.338,
+      "step": 5611
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.705393705530944e-05,
+      "loss": 0.8056,
+      "step": 5612
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.705290274210264e-05,
+      "loss": 1.7067,
+      "step": 5613
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7051868256741454e-05,
+      "loss": 1.5051,
+      "step": 5614
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.705083359923246e-05,
+      "loss": 0.8481,
+      "step": 5615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7049798769582265e-05,
+      "loss": 0.2993,
+      "step": 5616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7048763767797434e-05,
+      "loss": 0.1732,
+      "step": 5617
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.704772859388456e-05,
+      "loss": 1.4769,
+      "step": 5618
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.704669324785024e-05,
+      "loss": 1.1502,
+      "step": 5619
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7045657729701046e-05,
+      "loss": 0.2654,
+      "step": 5620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.704462203944357e-05,
+      "loss": 1.1663,
+      "step": 5621
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.704358617708442e-05,
+      "loss": 0.4602,
+      "step": 5622
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.704255014263017e-05,
+      "loss": 0.4528,
+      "step": 5623
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.704151393608741e-05,
+      "loss": 0.4903,
+      "step": 5624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7040477557462755e-05,
+      "loss": 0.4473,
+      "step": 5625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.703944100676278e-05,
+      "loss": 0.6829,
+      "step": 5626
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.703840428399409e-05,
+      "loss": 1.4403,
+      "step": 5627
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.703736738916328e-05,
+      "loss": 1.0598,
+      "step": 5628
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.703633032227695e-05,
+      "loss": 1.0886,
+      "step": 5629
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7035293083341706e-05,
+      "loss": 0.2775,
+      "step": 5630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7034255672364136e-05,
+      "loss": 1.3862,
+      "step": 5631
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7033218089350847e-05,
+      "loss": 1.0593,
+      "step": 5632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.703218033430844e-05,
+      "loss": 0.8662,
+      "step": 5633
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.703114240724352e-05,
+      "loss": 1.0916,
+      "step": 5634
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.70301043081627e-05,
+      "loss": 1.39,
+      "step": 5635
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.702906603707257e-05,
+      "loss": 1.6559,
+      "step": 5636
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.702802759397976e-05,
+      "loss": 0.3985,
+      "step": 5637
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.702698897889086e-05,
+      "loss": 1.2727,
+      "step": 5638
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.702595019181248e-05,
+      "loss": 0.444,
+      "step": 5639
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7024911232751224e-05,
+      "loss": 0.9676,
+      "step": 5640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.702387210171373e-05,
+      "loss": 0.879,
+      "step": 5641
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.702283279870658e-05,
+      "loss": 1.03,
+      "step": 5642
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.702179332373641e-05,
+      "loss": 0.3608,
+      "step": 5643
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7020753676809824e-05,
+      "loss": 1.1861,
+      "step": 5644
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.701971385793345e-05,
+      "loss": 0.6451,
+      "step": 5645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7018673867113886e-05,
+      "loss": 1.088,
+      "step": 5646
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.701763370435777e-05,
+      "loss": 1.0112,
+      "step": 5647
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.70165933696717e-05,
+      "loss": 2.0302,
+      "step": 5648
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7015552863062306e-05,
+      "loss": 1.1629,
+      "step": 5649
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.701451218453622e-05,
+      "loss": 0.3219,
+      "step": 5650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.701347133410006e-05,
+      "loss": 0.7919,
+      "step": 5651
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.701243031176043e-05,
+      "loss": 0.6249,
+      "step": 5652
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.701138911752398e-05,
+      "loss": 1.0739,
+      "step": 5653
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.701034775139732e-05,
+      "loss": 1.1371,
+      "step": 5654
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.700930621338708e-05,
+      "loss": 1.4479,
+      "step": 5655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.700826450349989e-05,
+      "loss": 0.8592,
+      "step": 5656
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.700722262174239e-05,
+      "loss": 1.2611,
+      "step": 5657
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.700618056812119e-05,
+      "loss": 0.9291,
+      "step": 5658
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.700513834264293e-05,
+      "loss": 0.9458,
+      "step": 5659
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.700409594531425e-05,
+      "loss": 0.4446,
+      "step": 5660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.700305337614177e-05,
+      "loss": 1.1317,
+      "step": 5661
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.700201063513213e-05,
+      "loss": 0.7725,
+      "step": 5662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.700096772229197e-05,
+      "loss": 1.7083,
+      "step": 5663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6999924637627926e-05,
+      "loss": 0.8609,
+      "step": 5664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.699888138114663e-05,
+      "loss": 1.7691,
+      "step": 5665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6997837952854724e-05,
+      "loss": 1.5053,
+      "step": 5666
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.699679435275885e-05,
+      "loss": 1.0364,
+      "step": 5667
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.699575058086564e-05,
+      "loss": 1.882,
+      "step": 5668
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.699470663718174e-05,
+      "loss": 0.3045,
+      "step": 5669
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.699366252171381e-05,
+      "loss": 1.7116,
+      "step": 5670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6992618234468474e-05,
+      "loss": 0.9153,
+      "step": 5671
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6991573775452386e-05,
+      "loss": 1.7081,
+      "step": 5672
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6990529144672186e-05,
+      "loss": 0.5214,
+      "step": 5673
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.698948434213453e-05,
+      "loss": 0.627,
+      "step": 5674
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.698843936784606e-05,
+      "loss": 0.4578,
+      "step": 5675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6987394221813425e-05,
+      "loss": 0.2324,
+      "step": 5676
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6986348904043284e-05,
+      "loss": 1.9524,
+      "step": 5677
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6985303414542295e-05,
+      "loss": 1.2751,
+      "step": 5678
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.698425775331709e-05,
+      "loss": 0.3034,
+      "step": 5679
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6983211920374335e-05,
+      "loss": 1.7728,
+      "step": 5680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.698216591572068e-05,
+      "loss": 0.644,
+      "step": 5681
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6981119739362796e-05,
+      "loss": 0.4912,
+      "step": 5682
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.698007339130733e-05,
+      "loss": 0.7877,
+      "step": 5683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.697902687156093e-05,
+      "loss": 1.071,
+      "step": 5684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.697798018013027e-05,
+      "loss": 0.6309,
+      "step": 5685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6976933317022006e-05,
+      "loss": 1.1134,
+      "step": 5686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.69758862822428e-05,
+      "loss": 1.2,
+      "step": 5687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.697483907579932e-05,
+      "loss": 0.8697,
+      "step": 5688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.697379169769823e-05,
+      "loss": 0.6722,
+      "step": 5689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.697274414794619e-05,
+      "loss": 0.739,
+      "step": 5690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6971696426549866e-05,
+      "loss": 1.3874,
+      "step": 5691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.697064853351592e-05,
+      "loss": 0.4477,
+      "step": 5692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.696960046885102e-05,
+      "loss": 0.703,
+      "step": 5693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.696855223256186e-05,
+      "loss": 1.4703,
+      "step": 5694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.696750382465508e-05,
+      "loss": 0.3912,
+      "step": 5695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6966455245137374e-05,
+      "loss": 0.8088,
+      "step": 5696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6965406494015403e-05,
+      "loss": 0.5255,
+      "step": 5697
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.696435757129584e-05,
+      "loss": 0.6061,
+      "step": 5698
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.696330847698537e-05,
+      "loss": 1.155,
+      "step": 5699
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.696225921109066e-05,
+      "loss": 0.804,
+      "step": 5700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.696120977361838e-05,
+      "loss": 1.4233,
+      "step": 5701
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.696016016457523e-05,
+      "loss": 1.39,
+      "step": 5702
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.695911038396787e-05,
+      "loss": 0.9486,
+      "step": 5703
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6958060431802994e-05,
+      "loss": 0.3415,
+      "step": 5704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.695701030808727e-05,
+      "loss": 0.5608,
+      "step": 5705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.695596001282739e-05,
+      "loss": 0.4752,
+      "step": 5706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.695490954603003e-05,
+      "loss": 1.4026,
+      "step": 5707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6953858907701894e-05,
+      "loss": 1.575,
+      "step": 5708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.695280809784964e-05,
+      "loss": 1.9203,
+      "step": 5709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.695175711647997e-05,
+      "loss": 0.7779,
+      "step": 5710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.695070596359957e-05,
+      "loss": 0.4403,
+      "step": 5711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6949654639215144e-05,
+      "loss": 0.6952,
+      "step": 5712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.694860314333335e-05,
+      "loss": 0.7791,
+      "step": 5713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.69475514759609e-05,
+      "loss": 0.908,
+      "step": 5714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6946499637104496e-05,
+      "loss": 0.3795,
+      "step": 5715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6945447626770805e-05,
+      "loss": 1.0961,
+      "step": 5716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.694439544496654e-05,
+      "loss": 0.4447,
+      "step": 5717
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6943343091698396e-05,
+      "loss": 0.8668,
+      "step": 5718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.694229056697307e-05,
+      "loss": 0.7153,
+      "step": 5719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.694123787079725e-05,
+      "loss": 1.6123,
+      "step": 5720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6940185003177634e-05,
+      "loss": 0.9787,
+      "step": 5721
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6939131964120944e-05,
+      "loss": 1.0661,
+      "step": 5722
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.693807875363386e-05,
+      "loss": 0.9264,
+      "step": 5723
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6937025371723095e-05,
+      "loss": 0.9515,
+      "step": 5724
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6935971818395345e-05,
+      "loss": 1.6019,
+      "step": 5725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.693491809365731e-05,
+      "loss": 0.3626,
+      "step": 5726
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6933864197515705e-05,
+      "loss": 0.8399,
+      "step": 5727
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6932810129977245e-05,
+      "loss": 1.1431,
+      "step": 5728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.693175589104862e-05,
+      "loss": 0.8841,
+      "step": 5729
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.693070148073655e-05,
+      "loss": 1.7796,
+      "step": 5730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6929646899047726e-05,
+      "loss": 1.4296,
+      "step": 5731
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.692859214598889e-05,
+      "loss": 1.7357,
+      "step": 5732
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.692753722156673e-05,
+      "loss": 1.665,
+      "step": 5733
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6926482125787966e-05,
+      "loss": 1.3204,
+      "step": 5734
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.692542685865932e-05,
+      "loss": 1.8925,
+      "step": 5735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.692437142018749e-05,
+      "loss": 0.5917,
+      "step": 5736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6923315810379215e-05,
+      "loss": 0.5545,
+      "step": 5737
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.69222600292412e-05,
+      "loss": 0.4938,
+      "step": 5738
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.692120407678015e-05,
+      "loss": 1.0391,
+      "step": 5739
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.692014795300281e-05,
+      "loss": 0.8713,
+      "step": 5740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6919091657915885e-05,
+      "loss": 1.2651,
+      "step": 5741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.69180351915261e-05,
+      "loss": 1.4765,
+      "step": 5742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6916978553840194e-05,
+      "loss": 0.7038,
+      "step": 5743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6915921744864855e-05,
+      "loss": 0.4008,
+      "step": 5744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.691486476460684e-05,
+      "loss": 1.1486,
+      "step": 5745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.691380761307287e-05,
+      "loss": 1.2095,
+      "step": 5746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.691275029026966e-05,
+      "loss": 1.1781,
+      "step": 5747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.691169279620394e-05,
+      "loss": 0.6709,
+      "step": 5748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6910635130882446e-05,
+      "loss": 0.6966,
+      "step": 5749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6909577294311916e-05,
+      "loss": 1.0951,
+      "step": 5750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.690851928649907e-05,
+      "loss": 0.8828,
+      "step": 5751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.690746110745064e-05,
+      "loss": 0.2439,
+      "step": 5752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.690640275717337e-05,
+      "loss": 0.3139,
+      "step": 5753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.690534423567398e-05,
+      "loss": 0.2369,
+      "step": 5754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.690428554295922e-05,
+      "loss": 0.3923,
+      "step": 5755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6903226679035826e-05,
+      "loss": 0.6166,
+      "step": 5756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6902167643910525e-05,
+      "loss": 1.3613,
+      "step": 5757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6901108437590067e-05,
+      "loss": 0.4435,
+      "step": 5758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.690004906008118e-05,
+      "loss": 1.3317,
+      "step": 5759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.689898951139062e-05,
+      "loss": 1.5087,
+      "step": 5760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6897929791525125e-05,
+      "loss": 1.2838,
+      "step": 5761
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6896869900491435e-05,
+      "loss": 0.1656,
+      "step": 5762
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.689580983829631e-05,
+      "loss": 1.1415,
+      "step": 5763
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.689474960494647e-05,
+      "loss": 0.6956,
+      "step": 5764
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.689368920044868e-05,
+      "loss": 0.7441,
+      "step": 5765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.689262862480967e-05,
+      "loss": 1.0611,
+      "step": 5766
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6891567878036216e-05,
+      "loss": 1.0622,
+      "step": 5767
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6890506960135044e-05,
+      "loss": 0.2891,
+      "step": 5768
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.688944587111292e-05,
+      "loss": 0.815,
+      "step": 5769
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.68883846109766e-05,
+      "loss": 0.2338,
+      "step": 5770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.688732317973282e-05,
+      "loss": 0.9837,
+      "step": 5771
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.688626157738834e-05,
+      "loss": 0.9247,
+      "step": 5772
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6885199803949924e-05,
+      "loss": 0.4493,
+      "step": 5773
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.688413785942433e-05,
+      "loss": 1.1066,
+      "step": 5774
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6883075743818306e-05,
+      "loss": 0.8022,
+      "step": 5775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.688201345713861e-05,
+      "loss": 0.6386,
+      "step": 5776
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6880950999392006e-05,
+      "loss": 0.644,
+      "step": 5777
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.687988837058526e-05,
+      "loss": 0.8315,
+      "step": 5778
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.687882557072513e-05,
+      "loss": 0.384,
+      "step": 5779
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.687776259981838e-05,
+      "loss": 0.6321,
+      "step": 5780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.687669945787177e-05,
+      "loss": 0.4365,
+      "step": 5781
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6875636144892073e-05,
+      "loss": 0.9721,
+      "step": 5782
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6874572660886055e-05,
+      "loss": 0.838,
+      "step": 5783
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.687350900586048e-05,
+      "loss": 0.7815,
+      "step": 5784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.687244517982211e-05,
+      "loss": 0.9233,
+      "step": 5785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.687138118277772e-05,
+      "loss": 0.7479,
+      "step": 5786
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.687031701473409e-05,
+      "loss": 0.3083,
+      "step": 5787
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6869252675697985e-05,
+      "loss": 0.3645,
+      "step": 5788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6868188165676174e-05,
+      "loss": 0.7661,
+      "step": 5789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.686712348467543e-05,
+      "loss": 0.9,
+      "step": 5790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.686605863270254e-05,
+      "loss": 1.001,
+      "step": 5791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.686499360976427e-05,
+      "loss": 0.2662,
+      "step": 5792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6863928415867406e-05,
+      "loss": 0.5987,
+      "step": 5793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.686286305101872e-05,
+      "loss": 1.5484,
+      "step": 5794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6861797515224985e-05,
+      "loss": 1.2344,
+      "step": 5795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.686073180849299e-05,
+      "loss": 0.8579,
+      "step": 5796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.685966593082952e-05,
+      "loss": 0.6234,
+      "step": 5797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.685859988224136e-05,
+      "loss": 0.4626,
+      "step": 5798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6857533662735274e-05,
+      "loss": 1.5989,
+      "step": 5799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6856467272318074e-05,
+      "loss": 1.7775,
+      "step": 5800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6855400710996526e-05,
+      "loss": 0.5077,
+      "step": 5801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.685433397877742e-05,
+      "loss": 0.9288,
+      "step": 5802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6853267075667564e-05,
+      "loss": 0.7881,
+      "step": 5803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6852200001673714e-05,
+      "loss": 0.3545,
+      "step": 5804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.685113275680269e-05,
+      "loss": 0.5752,
+      "step": 5805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.685006534106127e-05,
+      "loss": 1.136,
+      "step": 5806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.684899775445625e-05,
+      "loss": 0.8839,
+      "step": 5807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.684792999699442e-05,
+      "loss": 1.4557,
+      "step": 5808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.684686206868258e-05,
+      "loss": 1.1723,
+      "step": 5809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.684579396952751e-05,
+      "loss": 0.7178,
+      "step": 5810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.684472569953603e-05,
+      "loss": 0.6496,
+      "step": 5811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6843657258714926e-05,
+      "loss": 1.1902,
+      "step": 5812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6842588647071e-05,
+      "loss": 0.291,
+      "step": 5813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.684151986461105e-05,
+      "loss": 0.4515,
+      "step": 5814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.684045091134188e-05,
+      "loss": 1.1356,
+      "step": 5815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6839381787270285e-05,
+      "loss": 1.0267,
+      "step": 5816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.683831249240308e-05,
+      "loss": 0.7825,
+      "step": 5817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.683724302674707e-05,
+      "loss": 0.48,
+      "step": 5818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6836173390309044e-05,
+      "loss": 0.9207,
+      "step": 5819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.683510358309583e-05,
+      "loss": 1.2187,
+      "step": 5820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6834033605114214e-05,
+      "loss": 1.026,
+      "step": 5821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6832963456371027e-05,
+      "loss": 1.2459,
+      "step": 5822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6831893136873065e-05,
+      "loss": 1.2534,
+      "step": 5823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6830822646627146e-05,
+      "loss": 0.6048,
+      "step": 5824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6829751985640065e-05,
+      "loss": 1.0512,
+      "step": 5825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.682868115391866e-05,
+      "loss": 0.9363,
+      "step": 5826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.682761015146973e-05,
+      "loss": 0.4894,
+      "step": 5827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.68265389783001e-05,
+      "loss": 0.5617,
+      "step": 5828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6825467634416574e-05,
+      "loss": 0.6398,
+      "step": 5829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6824396119825983e-05,
+      "loss": 1.3741,
+      "step": 5830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.682332443453514e-05,
+      "loss": 0.5512,
+      "step": 5831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.682225257855086e-05,
+      "loss": 0.8193,
+      "step": 5832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.682118055187996e-05,
+      "loss": 1.4753,
+      "step": 5833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.682010835452929e-05,
+      "loss": 0.709,
+      "step": 5834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.681903598650564e-05,
+      "loss": 1.5394,
+      "step": 5835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.681796344781586e-05,
+      "loss": 0.6146,
+      "step": 5836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.681689073846675e-05,
+      "loss": 0.3132,
+      "step": 5837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.681581785846515e-05,
+      "loss": 0.8819,
+      "step": 5838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6814744807817894e-05,
+      "loss": 0.9898,
+      "step": 5839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.68136715865318e-05,
+      "loss": 0.3892,
+      "step": 5840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.681259819461369e-05,
+      "loss": 0.5958,
+      "step": 5841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.681152463207042e-05,
+      "loss": 0.4382,
+      "step": 5842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.681045089890881e-05,
+      "loss": 0.4666,
+      "step": 5843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.680937699513567e-05,
+      "loss": 0.8457,
+      "step": 5844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.680830292075786e-05,
+      "loss": 0.4549,
+      "step": 5845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6807228675782215e-05,
+      "loss": 0.8345,
+      "step": 5846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.680615426021556e-05,
+      "loss": 0.7075,
+      "step": 5847
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6805079674064745e-05,
+      "loss": 1.0557,
+      "step": 5848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.680400491733659e-05,
+      "loss": 0.9292,
+      "step": 5849
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6802929990037944e-05,
+      "loss": 1.3479,
+      "step": 5850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.680185489217566e-05,
+      "loss": 0.9998,
+      "step": 5851
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.680077962375656e-05,
+      "loss": 0.6032,
+      "step": 5852
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.67997041847875e-05,
+      "loss": 0.4986,
+      "step": 5853
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6798628575275306e-05,
+      "loss": 0.661,
+      "step": 5854
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6797552795226835e-05,
+      "loss": 0.4311,
+      "step": 5855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.679647684464894e-05,
+      "loss": 1.4179,
+      "step": 5856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.679540072354846e-05,
+      "loss": 1.5904,
+      "step": 5857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.679432443193224e-05,
+      "loss": 0.5479,
+      "step": 5858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.679324796980713e-05,
+      "loss": 0.5094,
+      "step": 5859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6792171337179986e-05,
+      "loss": 0.9289,
+      "step": 5860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6791094534057664e-05,
+      "loss": 0.7946,
+      "step": 5861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6790017560446996e-05,
+      "loss": 0.9128,
+      "step": 5862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.678894041635485e-05,
+      "loss": 1.7565,
+      "step": 5863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.678786310178808e-05,
+      "loss": 0.5971,
+      "step": 5864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.678678561675354e-05,
+      "loss": 1.3655,
+      "step": 5865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6785707961258085e-05,
+      "loss": 1.1436,
+      "step": 5866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.678463013530858e-05,
+      "loss": 0.4513,
+      "step": 5867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.678355213891188e-05,
+      "loss": 0.2947,
+      "step": 5868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.678247397207483e-05,
+      "loss": 0.873,
+      "step": 5869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.678139563480431e-05,
+      "loss": 1.5916,
+      "step": 5870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.678031712710718e-05,
+      "loss": 1.3721,
+      "step": 5871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6779238448990296e-05,
+      "loss": 1.0304,
+      "step": 5872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6778159600460524e-05,
+      "loss": 0.8263,
+      "step": 5873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.677708058152474e-05,
+      "loss": 0.7089,
+      "step": 5874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6776001392189796e-05,
+      "loss": 0.3621,
+      "step": 5875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.677492203246256e-05,
+      "loss": 1.3328,
+      "step": 5876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6773842502349915e-05,
+      "loss": 0.3279,
+      "step": 5877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.677276280185872e-05,
+      "loss": 1.1275,
+      "step": 5878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.677168293099585e-05,
+      "loss": 0.2921,
+      "step": 5879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.677060288976816e-05,
+      "loss": 0.7068,
+      "step": 5880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.676952267818255e-05,
+      "loss": 0.6564,
+      "step": 5881
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.676844229624588e-05,
+      "loss": 1.0692,
+      "step": 5882
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6767361743965024e-05,
+      "loss": 1.3113,
+      "step": 5883
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.676628102134686e-05,
+      "loss": 1.1011,
+      "step": 5884
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.676520012839827e-05,
+      "loss": 0.469,
+      "step": 5885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.676411906512613e-05,
+      "loss": 0.3714,
+      "step": 5886
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6763037831537306e-05,
+      "loss": 1.4667,
+      "step": 5887
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.67619564276387e-05,
+      "loss": 0.2494,
+      "step": 5888
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.676087485343718e-05,
+      "loss": 0.9756,
+      "step": 5889
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.675979310893964e-05,
+      "loss": 0.7266,
+      "step": 5890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.675871119415295e-05,
+      "loss": 1.5261,
+      "step": 5891
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6757629109084e-05,
+      "loss": 1.4315,
+      "step": 5892
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6756546853739677e-05,
+      "loss": 0.2147,
+      "step": 5893
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.675546442812688e-05,
+      "loss": 1.8262,
+      "step": 5894
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.675438183225247e-05,
+      "loss": 0.513,
+      "step": 5895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6753299066123364e-05,
+      "loss": 0.7795,
+      "step": 5896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.675221612974643e-05,
+      "loss": 0.2262,
+      "step": 5897
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6751133023128577e-05,
+      "loss": 0.7952,
+      "step": 5898
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.675004974627669e-05,
+      "loss": 0.5958,
+      "step": 5899
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.674896629919766e-05,
+      "loss": 0.5676,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.674788268189839e-05,
+      "loss": 0.6505,
+      "step": 5901
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6746798894385767e-05,
+      "loss": 0.2099,
+      "step": 5902
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.674571493666669e-05,
+      "loss": 0.4556,
+      "step": 5903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.674463080874805e-05,
+      "loss": 0.4131,
+      "step": 5904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.674354651063677e-05,
+      "loss": 0.5844,
+      "step": 5905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.674246204233972e-05,
+      "loss": 0.514,
+      "step": 5906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.674137740386382e-05,
+      "loss": 1.6101,
+      "step": 5907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.674029259521596e-05,
+      "loss": 0.8004,
+      "step": 5908
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.673920761640307e-05,
+      "loss": 0.4193,
+      "step": 5909
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.673812246743201e-05,
+      "loss": 0.4904,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6737037148309726e-05,
+      "loss": 1.4728,
+      "step": 5911
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6735951659043105e-05,
+      "loss": 0.9755,
+      "step": 5912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6734865999639054e-05,
+      "loss": 0.3369,
+      "step": 5913
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.673378017010449e-05,
+      "loss": 0.2127,
+      "step": 5914
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.673269417044633e-05,
+      "loss": 0.8488,
+      "step": 5915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.673160800067146e-05,
+      "loss": 0.5997,
+      "step": 5916
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.673052166078681e-05,
+      "loss": 1.2389,
+      "step": 5917
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.672943515079928e-05,
+      "loss": 0.7618,
+      "step": 5918
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.672834847071581e-05,
+      "loss": 1.2743,
+      "step": 5919
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.672726162054329e-05,
+      "loss": 0.562,
+      "step": 5920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6726174600288646e-05,
+      "loss": 1.375,
+      "step": 5921
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6725087409958784e-05,
+      "loss": 1.0981,
+      "step": 5922
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6724000049560644e-05,
+      "loss": 0.9115,
+      "step": 5923
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.672291251910114e-05,
+      "loss": 0.5471,
+      "step": 5924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.672182481858717e-05,
+      "loss": 0.7044,
+      "step": 5925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.672073694802568e-05,
+      "loss": 0.3439,
+      "step": 5926
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.671964890742359e-05,
+      "loss": 1.3953,
+      "step": 5927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6718560696787815e-05,
+      "loss": 0.684,
+      "step": 5928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.671747231612528e-05,
+      "loss": 0.9548,
+      "step": 5929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6716383765442924e-05,
+      "loss": 1.0242,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6715295044747655e-05,
+      "loss": 0.502,
+      "step": 5931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.671420615404641e-05,
+      "loss": 1.6149,
+      "step": 5932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.671311709334613e-05,
+      "loss": 1.2437,
+      "step": 5933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6712027862653735e-05,
+      "loss": 0.4356,
+      "step": 5934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.671093846197615e-05,
+      "loss": 2.0876,
+      "step": 5935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.670984889132032e-05,
+      "loss": 1.1936,
+      "step": 5936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.670875915069317e-05,
+      "loss": 0.5022,
+      "step": 5937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.670766924010163e-05,
+      "loss": 1.8642,
+      "step": 5938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.670657915955265e-05,
+      "loss": 1.2173,
+      "step": 5939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6705488909053155e-05,
+      "loss": 0.8038,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.670439848861009e-05,
+      "loss": 0.4372,
+      "step": 5941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.67033078982304e-05,
+      "loss": 0.9632,
+      "step": 5942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6702217137921007e-05,
+      "loss": 1.2591,
+      "step": 5943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.670112620768886e-05,
+      "loss": 1.1659,
+      "step": 5944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.670003510754092e-05,
+      "loss": 1.3879,
+      "step": 5945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.669894383748409e-05,
+      "loss": 1.1379,
+      "step": 5946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.669785239752536e-05,
+      "loss": 1.1201,
+      "step": 5947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.669676078767164e-05,
+      "loss": 0.4084,
+      "step": 5948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.66956690079299e-05,
+      "loss": 1.2045,
+      "step": 5949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.669457705830707e-05,
+      "loss": 0.6868,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.669348493881011e-05,
+      "loss": 0.4063,
+      "step": 5951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.669239264944597e-05,
+      "loss": 0.719,
+      "step": 5952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6691300190221595e-05,
+      "loss": 0.4751,
+      "step": 5953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.669020756114393e-05,
+      "loss": 1.4838,
+      "step": 5954
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6689114762219956e-05,
+      "loss": 2.4898,
+      "step": 5955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.668802179345659e-05,
+      "loss": 0.9804,
+      "step": 5956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6686928654860813e-05,
+      "loss": 0.4448,
+      "step": 5957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.668583534643958e-05,
+      "loss": 1.0114,
+      "step": 5958
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.668474186819984e-05,
+      "loss": 1.5527,
+      "step": 5959
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.668364822014856e-05,
+      "loss": 0.5925,
+      "step": 5960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.668255440229268e-05,
+      "loss": 0.8547,
+      "step": 5961
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.668146041463918e-05,
+      "loss": 0.5173,
+      "step": 5962
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.668036625719502e-05,
+      "loss": 1.2925,
+      "step": 5963
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6679271929967155e-05,
+      "loss": 1.1241,
+      "step": 5964
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.667817743296256e-05,
+      "loss": 1.6034,
+      "step": 5965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6677082766188176e-05,
+      "loss": 0.7199,
+      "step": 5966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.667598792965099e-05,
+      "loss": 1.1353,
+      "step": 5967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.667489292335798e-05,
+      "loss": 0.9507,
+      "step": 5968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6673797747316086e-05,
+      "loss": 0.666,
+      "step": 5969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.66727024015323e-05,
+      "loss": 0.9618,
+      "step": 5970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.667160688601358e-05,
+      "loss": 0.1339,
+      "step": 5971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6670511200766896e-05,
+      "loss": 1.4209,
+      "step": 5972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.666941534579922e-05,
+      "loss": 0.4786,
+      "step": 5973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6668319321117535e-05,
+      "loss": 0.8637,
+      "step": 5974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6667223126728816e-05,
+      "loss": 0.9312,
+      "step": 5975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.666612676264003e-05,
+      "loss": 0.9957,
+      "step": 5976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6665030228858154e-05,
+      "loss": 0.773,
+      "step": 5977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.666393352539018e-05,
+      "loss": 1.6589,
+      "step": 5978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.666283665224307e-05,
+      "loss": 0.8296,
+      "step": 5979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.666173960942381e-05,
+      "loss": 0.6297,
+      "step": 5980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.666064239693938e-05,
+      "loss": 1.2658,
+      "step": 5981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.665954501479676e-05,
+      "loss": 1.0322,
+      "step": 5982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.665844746300295e-05,
+      "loss": 1.6497,
+      "step": 5983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6657349741564915e-05,
+      "loss": 1.1513,
+      "step": 5984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.665625185048965e-05,
+      "loss": 0.5982,
+      "step": 5985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.665515378978413e-05,
+      "loss": 1.5975,
+      "step": 5986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.665405555945536e-05,
+      "loss": 0.3873,
+      "step": 5987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6652957159510314e-05,
+      "loss": 0.4571,
+      "step": 5988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6651858589956e-05,
+      "loss": 1.2526,
+      "step": 5989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.665075985079938e-05,
+      "loss": 0.2897,
+      "step": 5990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.664966094204747e-05,
+      "loss": 1.2023,
+      "step": 5991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6648561863707263e-05,
+      "loss": 1.1513,
+      "step": 5992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6647462615785735e-05,
+      "loss": 0.7323,
+      "step": 5993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6646363198289904e-05,
+      "loss": 0.6973,
+      "step": 5994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6645263611226744e-05,
+      "loss": 1.1355,
+      "step": 5995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.664416385460327e-05,
+      "loss": 1.3136,
+      "step": 5996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.664306392842646e-05,
+      "loss": 0.8591,
+      "step": 5997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.664196383270334e-05,
+      "loss": 0.492,
+      "step": 5998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.66408635674409e-05,
+      "loss": 1.1306,
+      "step": 5999
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.663976313264613e-05,
+      "loss": 1.3323,
+      "step": 6000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.663866252832604e-05,
+      "loss": 0.7347,
+      "step": 6001
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.663756175448764e-05,
+      "loss": 1.1456,
+      "step": 6002
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.663646081113793e-05,
+      "loss": 1.3731,
+      "step": 6003
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.663535969828392e-05,
+      "loss": 0.8794,
+      "step": 6004
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6634258415932614e-05,
+      "loss": 0.2429,
+      "step": 6005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.663315696409101e-05,
+      "loss": 0.7331,
+      "step": 6006
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.663205534276613e-05,
+      "loss": 0.6181,
+      "step": 6007
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.663095355196499e-05,
+      "loss": 0.363,
+      "step": 6008
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6629851591694585e-05,
+      "loss": 1.8659,
+      "step": 6009
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.662874946196193e-05,
+      "loss": 0.9001,
+      "step": 6010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.662764716277406e-05,
+      "loss": 0.3511,
+      "step": 6011
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6626544694137955e-05,
+      "loss": 0.3914,
+      "step": 6012
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6625442056060655e-05,
+      "loss": 0.6525,
+      "step": 6013
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.662433924854917e-05,
+      "loss": 1.0475,
+      "step": 6014
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.662323627161052e-05,
+      "loss": 1.7327,
+      "step": 6015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.662213312525173e-05,
+      "loss": 0.8056,
+      "step": 6016
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.66210298094798e-05,
+      "loss": 0.9889,
+      "step": 6017
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.661992632430177e-05,
+      "loss": 0.2814,
+      "step": 6018
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.661882266972466e-05,
+      "loss": 0.9707,
+      "step": 6019
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6617718845755473e-05,
+      "loss": 0.9816,
+      "step": 6020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.661661485240126e-05,
+      "loss": 0.5029,
+      "step": 6021
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.661551068966904e-05,
+      "loss": 0.5035,
+      "step": 6022
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.661440635756583e-05,
+      "loss": 0.5934,
+      "step": 6023
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.661330185609866e-05,
+      "loss": 1.5582,
+      "step": 6024
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6612197185274564e-05,
+      "loss": 0.2936,
+      "step": 6025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.661109234510058e-05,
+      "loss": 0.5038,
+      "step": 6026
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.660998733558371e-05,
+      "loss": 1.9003,
+      "step": 6027
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.660888215673101e-05,
+      "loss": 1.7493,
+      "step": 6028
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.660777680854951e-05,
+      "loss": 1.5624,
+      "step": 6029
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6606671291046235e-05,
+      "loss": 1.5129,
+      "step": 6030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6605565604228225e-05,
+      "loss": 1.3333,
+      "step": 6031
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.660445974810252e-05,
+      "loss": 1.1664,
+      "step": 6032
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.660335372267615e-05,
+      "loss": 1.4176,
+      "step": 6033
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.660224752795616e-05,
+      "loss": 1.3779,
+      "step": 6034
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.660114116394959e-05,
+      "loss": 1.2812,
+      "step": 6035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.660003463066347e-05,
+      "loss": 0.416,
+      "step": 6036
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.659892792810486e-05,
+      "loss": 0.7885,
+      "step": 6037
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6597821056280785e-05,
+      "loss": 1.368,
+      "step": 6038
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.659671401519829e-05,
+      "loss": 0.5841,
+      "step": 6039
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6595606804864436e-05,
+      "loss": 0.1272,
+      "step": 6040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.659449942528625e-05,
+      "loss": 1.835,
+      "step": 6041
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.659339187647079e-05,
+      "loss": 1.2013,
+      "step": 6042
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.65922841584251e-05,
+      "loss": 1.1956,
+      "step": 6043
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.659117627115623e-05,
+      "loss": 0.8319,
+      "step": 6044
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.659006821467122e-05,
+      "loss": 0.5544,
+      "step": 6045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.658895998897714e-05,
+      "loss": 0.5232,
+      "step": 6046
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.658785159408104e-05,
+      "loss": 0.8796,
+      "step": 6047
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6586743029989956e-05,
+      "loss": 0.6335,
+      "step": 6048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.658563429671096e-05,
+      "loss": 1.5022,
+      "step": 6049
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6584525394251094e-05,
+      "loss": 0.7392,
+      "step": 6050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.658341632261743e-05,
+      "loss": 0.6644,
+      "step": 6051
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.658230708181702e-05,
+      "loss": 1.5677,
+      "step": 6052
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.658119767185691e-05,
+      "loss": 0.9774,
+      "step": 6053
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6580088092744176e-05,
+      "loss": 0.1809,
+      "step": 6054
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.657897834448587e-05,
+      "loss": 0.8408,
+      "step": 6055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.657786842708906e-05,
+      "loss": 1.2159,
+      "step": 6056
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6576758340560806e-05,
+      "loss": 0.9985,
+      "step": 6057
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6575648084908174e-05,
+      "loss": 0.4619,
+      "step": 6058
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.657453766013822e-05,
+      "loss": 0.7533,
+      "step": 6059
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6573427066258024e-05,
+      "loss": 1.0876,
+      "step": 6060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.657231630327464e-05,
+      "loss": 0.3164,
+      "step": 6061
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.657120537119516e-05,
+      "loss": 0.1562,
+      "step": 6062
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6570094270026615e-05,
+      "loss": 1.4003,
+      "step": 6063
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.656898299977611e-05,
+      "loss": 0.9301,
+      "step": 6064
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6567871560450705e-05,
+      "loss": 0.7472,
+      "step": 6065
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6566759952057474e-05,
+      "loss": 0.4716,
+      "step": 6066
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6565648174603486e-05,
+      "loss": 1.3471,
+      "step": 6067
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.656453622809581e-05,
+      "loss": 0.8369,
+      "step": 6068
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.656342411254155e-05,
+      "loss": 0.3544,
+      "step": 6069
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.656231182794775e-05,
+      "loss": 1.2762,
+      "step": 6070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6561199374321505e-05,
+      "loss": 0.8883,
+      "step": 6071
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6560086751669894e-05,
+      "loss": 0.9657,
+      "step": 6072
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6558973959999995e-05,
+      "loss": 1.0775,
+      "step": 6073
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.655786099931889e-05,
+      "loss": 0.6965,
+      "step": 6074
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.655674786963366e-05,
+      "loss": 1.4892,
+      "step": 6075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6555634570951385e-05,
+      "loss": 1.8003,
+      "step": 6076
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.655452110327916e-05,
+      "loss": 0.3875,
+      "step": 6077
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6553407466624066e-05,
+      "loss": 0.8097,
+      "step": 6078
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6552293660993183e-05,
+      "loss": 0.6936,
+      "step": 6079
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.655117968639361e-05,
+      "loss": 0.3351,
+      "step": 6080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.655006554283242e-05,
+      "loss": 0.8507,
+      "step": 6081
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.654895123031672e-05,
+      "loss": 0.837,
+      "step": 6082
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6547836748853595e-05,
+      "loss": 1.2849,
+      "step": 6083
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.654672209845013e-05,
+      "loss": 0.1406,
+      "step": 6084
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6545607279113425e-05,
+      "loss": 1.2844,
+      "step": 6085
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.654449229085058e-05,
+      "loss": 1.6662,
+      "step": 6086
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6543377133668686e-05,
+      "loss": 0.2071,
+      "step": 6087
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.654226180757483e-05,
+      "loss": 0.956,
+      "step": 6088
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6541146312576115e-05,
+      "loss": 1.2552,
+      "step": 6089
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.654003064867965e-05,
+      "loss": 1.0291,
+      "step": 6090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.653891481589252e-05,
+      "loss": 0.2277,
+      "step": 6091
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.653779881422184e-05,
+      "loss": 0.9308,
+      "step": 6092
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.653668264367469e-05,
+      "loss": 1.3527,
+      "step": 6093
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.65355663042582e-05,
+      "loss": 1.1593,
+      "step": 6094
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.653444979597946e-05,
+      "loss": 0.6772,
+      "step": 6095
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6533333118845566e-05,
+      "loss": 1.0868,
+      "step": 6096
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.653221627286364e-05,
+      "loss": 1.1908,
+      "step": 6097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.653109925804078e-05,
+      "loss": 1.0598,
+      "step": 6098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.65299820743841e-05,
+      "loss": 0.7799,
+      "step": 6099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.652886472190071e-05,
+      "loss": 1.0747,
+      "step": 6100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.652774720059771e-05,
+      "loss": 0.9056,
+      "step": 6101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.652662951048223e-05,
+      "loss": 0.5535,
+      "step": 6102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6525511651561354e-05,
+      "loss": 1.1167,
+      "step": 6103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.652439362384222e-05,
+      "loss": 1.3301,
+      "step": 6104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.652327542733194e-05,
+      "loss": 0.6306,
+      "step": 6105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.652215706203763e-05,
+      "loss": 0.8933,
+      "step": 6106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.652103852796639e-05,
+      "loss": 1.5261,
+      "step": 6107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6519919825125354e-05,
+      "loss": 1.5811,
+      "step": 6108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6518800953521636e-05,
+      "loss": 1.0721,
+      "step": 6109
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.651768191316236e-05,
+      "loss": 0.9919,
+      "step": 6110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.651656270405464e-05,
+      "loss": 0.6845,
+      "step": 6111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6515443326205604e-05,
+      "loss": 1.4347,
+      "step": 6112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.651432377962238e-05,
+      "loss": 0.6122,
+      "step": 6113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.651320406431207e-05,
+      "loss": 0.9826,
+      "step": 6114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.651208418028183e-05,
+      "loss": 0.3004,
+      "step": 6115
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6510964127538764e-05,
+      "loss": 1.3043,
+      "step": 6116
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.650984390609001e-05,
+      "loss": 1.4562,
+      "step": 6117
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.65087235159427e-05,
+      "loss": 0.596,
+      "step": 6118
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.650760295710396e-05,
+      "loss": 0.3155,
+      "step": 6119
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6506482229580904e-05,
+      "loss": 1.5468,
+      "step": 6120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.650536133338069e-05,
+      "loss": 0.4311,
+      "step": 6121
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6504240268510437e-05,
+      "loss": 0.272,
+      "step": 6122
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.650311903497728e-05,
+      "loss": 1.6208,
+      "step": 6123
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.650199763278836e-05,
+      "loss": 0.8028,
+      "step": 6124
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6500876061950805e-05,
+      "loss": 1.152,
+      "step": 6125
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.649975432247176e-05,
+      "loss": 0.6293,
+      "step": 6126
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.649863241435835e-05,
+      "loss": 1.0029,
+      "step": 6127
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.649751033761773e-05,
+      "loss": 0.8597,
+      "step": 6128
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.649638809225704e-05,
+      "loss": 1.1461,
+      "step": 6129
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.649526567828341e-05,
+      "loss": 0.2165,
+      "step": 6130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6494143095703985e-05,
+      "loss": 0.6225,
+      "step": 6131
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.649302034452592e-05,
+      "loss": 0.407,
+      "step": 6132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.649189742475635e-05,
+      "loss": 1.0684,
+      "step": 6133
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.649077433640242e-05,
+      "loss": 0.4818,
+      "step": 6134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6489651079471275e-05,
+      "loss": 0.2643,
+      "step": 6135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6488527653970076e-05,
+      "loss": 0.2153,
+      "step": 6136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.648740405990596e-05,
+      "loss": 0.3958,
+      "step": 6137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.648628029728609e-05,
+      "loss": 0.4886,
+      "step": 6138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6485156366117596e-05,
+      "loss": 1.5124,
+      "step": 6139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.648403226640765e-05,
+      "loss": 1.2617,
+      "step": 6140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.648290799816339e-05,
+      "loss": 0.6342,
+      "step": 6141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.648178356139199e-05,
+      "loss": 1.2589,
+      "step": 6142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6480658956100584e-05,
+      "loss": 0.7498,
+      "step": 6143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.647953418229634e-05,
+      "loss": 0.9843,
+      "step": 6144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6478409239986405e-05,
+      "loss": 0.4235,
+      "step": 6145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.647728412917797e-05,
+      "loss": 0.1186,
+      "step": 6146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.647615884987815e-05,
+      "loss": 1.5359,
+      "step": 6147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.647503340209414e-05,
+      "loss": 1.1158,
+      "step": 6148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6473907785833076e-05,
+      "loss": 1.314,
+      "step": 6149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.647278200110214e-05,
+      "loss": 0.6088,
+      "step": 6150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.64716560479085e-05,
+      "loss": 0.6301,
+      "step": 6151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.64705299262593e-05,
+      "loss": 0.6789,
+      "step": 6152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6469403636161717e-05,
+      "loss": 1.3092,
+      "step": 6153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6468277177622924e-05,
+      "loss": 1.2365,
+      "step": 6154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.646715055065008e-05,
+      "loss": 0.4071,
+      "step": 6155
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6466023755250355e-05,
+      "loss": 1.3398,
+      "step": 6156
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.646489679143093e-05,
+      "loss": 0.3797,
+      "step": 6157
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6463769659198975e-05,
+      "loss": 1.0731,
+      "step": 6158
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.646264235856165e-05,
+      "loss": 0.8029,
+      "step": 6159
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.646151488952613e-05,
+      "loss": 1.1356,
+      "step": 6160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.646038725209961e-05,
+      "loss": 1.6227,
+      "step": 6161
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.645925944628924e-05,
+      "loss": 1.3399,
+      "step": 6162
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.645813147210221e-05,
+      "loss": 0.578,
+      "step": 6163
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.64570033295457e-05,
+      "loss": 0.3176,
+      "step": 6164
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6455875018626886e-05,
+      "loss": 0.4335,
+      "step": 6165
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6454746539352945e-05,
+      "loss": 0.368,
+      "step": 6166
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6453617891731056e-05,
+      "loss": 1.4584,
+      "step": 6167
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.645248907576842e-05,
+      "loss": 0.789,
+      "step": 6168
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6451360091472196e-05,
+      "loss": 0.7411,
+      "step": 6169
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.645023093884958e-05,
+      "loss": 0.5136,
+      "step": 6170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6449101617907754e-05,
+      "loss": 1.0173,
+      "step": 6171
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.644797212865391e-05,
+      "loss": 1.5889,
+      "step": 6172
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.644684247109524e-05,
+      "loss": 0.2098,
+      "step": 6173
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6445712645238915e-05,
+      "loss": 1.6506,
+      "step": 6174
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.644458265109214e-05,
+      "loss": 0.9493,
+      "step": 6175
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.64434524886621e-05,
+      "loss": 0.4819,
+      "step": 6176
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6442322157955985e-05,
+      "loss": 0.4348,
+      "step": 6177
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6441191658981e-05,
+      "loss": 0.5789,
+      "step": 6178
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6440060991744324e-05,
+      "loss": 0.9506,
+      "step": 6179
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.643893015625316e-05,
+      "loss": 0.228,
+      "step": 6180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.64377991525147e-05,
+      "loss": 0.7396,
+      "step": 6181
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6436667980536145e-05,
+      "loss": 0.9672,
+      "step": 6182
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6435536640324684e-05,
+      "loss": 1.1125,
+      "step": 6183
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.643440513188754e-05,
+      "loss": 1.36,
+      "step": 6184
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.643327345523189e-05,
+      "loss": 0.3176,
+      "step": 6185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.643214161036494e-05,
+      "loss": 1.4026,
+      "step": 6186
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.64310095972939e-05,
+      "loss": 0.8793,
+      "step": 6187
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.642987741602597e-05,
+      "loss": 1.7102,
+      "step": 6188
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6428745066568354e-05,
+      "loss": 1.2541,
+      "step": 6189
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6427612548928254e-05,
+      "loss": 1.4419,
+      "step": 6190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.642647986311289e-05,
+      "loss": 1.3076,
+      "step": 6191
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6425347009129455e-05,
+      "loss": 1.0822,
+      "step": 6192
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.642421398698517e-05,
+      "loss": 1.1118,
+      "step": 6193
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.642308079668723e-05,
+      "loss": 1.4162,
+      "step": 6194
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6421947438242864e-05,
+      "loss": 0.5287,
+      "step": 6195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.642081391165927e-05,
+      "loss": 0.2226,
+      "step": 6196
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.641968021694367e-05,
+      "loss": 0.7299,
+      "step": 6197
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.641854635410328e-05,
+      "loss": 1.3356,
+      "step": 6198
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6417412323145305e-05,
+      "loss": 0.794,
+      "step": 6199
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.641627812407697e-05,
+      "loss": 1.6208,
+      "step": 6200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.641514375690549e-05,
+      "loss": 0.5447,
+      "step": 6201
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.641400922163808e-05,
+      "loss": 0.7537,
+      "step": 6202
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6412874518281976e-05,
+      "loss": 0.4673,
+      "step": 6203
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6411739646844375e-05,
+      "loss": 0.4501,
+      "step": 6204
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.641060460733252e-05,
+      "loss": 1.2563,
+      "step": 6205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.640946939975361e-05,
+      "loss": 1.8313,
+      "step": 6206
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.640833402411489e-05,
+      "loss": 0.6923,
+      "step": 6207
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.640719848042358e-05,
+      "loss": 0.9366,
+      "step": 6208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6406062768686904e-05,
+      "loss": 1.362,
+      "step": 6209
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6404926888912096e-05,
+      "loss": 0.2759,
+      "step": 6210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.640379084110637e-05,
+      "loss": 0.3592,
+      "step": 6211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.640265462527697e-05,
+      "loss": 0.5104,
+      "step": 6212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.640151824143112e-05,
+      "loss": 0.9671,
+      "step": 6213
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6400381689576045e-05,
+      "loss": 1.0221,
+      "step": 6214
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.639924496971898e-05,
+      "loss": 0.519,
+      "step": 6215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.639810808186718e-05,
+      "loss": 0.4127,
+      "step": 6216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.639697102602785e-05,
+      "loss": 0.6564,
+      "step": 6217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.639583380220824e-05,
+      "loss": 1.4322,
+      "step": 6218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6394696410415584e-05,
+      "loss": 1.1088,
+      "step": 6219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6393558850657123e-05,
+      "loss": 0.4565,
+      "step": 6220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.639242112294009e-05,
+      "loss": 0.623,
+      "step": 6221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.639128322727174e-05,
+      "loss": 0.4743,
+      "step": 6222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.639014516365929e-05,
+      "loss": 1.4127,
+      "step": 6223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.638900693211001e-05,
+      "loss": 1.4016,
+      "step": 6224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.638786853263112e-05,
+      "loss": 1.4884,
+      "step": 6225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6386729965229876e-05,
+      "loss": 0.8859,
+      "step": 6226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6385591229913514e-05,
+      "loss": 0.582,
+      "step": 6227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.638445232668929e-05,
+      "loss": 0.387,
+      "step": 6228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.638331325556445e-05,
+      "loss": 0.9849,
+      "step": 6229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.638217401654624e-05,
+      "loss": 0.4577,
+      "step": 6230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.638103460964192e-05,
+      "loss": 1.2742,
+      "step": 6231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.637989503485872e-05,
+      "loss": 0.2713,
+      "step": 6232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.637875529220391e-05,
+      "loss": 0.6826,
+      "step": 6233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.637761538168473e-05,
+      "loss": 0.3943,
+      "step": 6234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.637647530330845e-05,
+      "loss": 0.2878,
+      "step": 6235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.63753350570823e-05,
+      "loss": 0.1971,
+      "step": 6236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.637419464301356e-05,
+      "loss": 0.1383,
+      "step": 6237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6373054061109475e-05,
+      "loss": 0.6154,
+      "step": 6238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.637191331137732e-05,
+      "loss": 1.2454,
+      "step": 6239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6370772393824326e-05,
+      "loss": 0.5883,
+      "step": 6240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6369631308457776e-05,
+      "loss": 1.1718,
+      "step": 6241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6368490055284926e-05,
+      "loss": 1.0096,
+      "step": 6242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6367348634313026e-05,
+      "loss": 0.6467,
+      "step": 6243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.636620704554936e-05,
+      "loss": 0.9159,
+      "step": 6244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.636506528900118e-05,
+      "loss": 1.5398,
+      "step": 6245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.636392336467576e-05,
+      "loss": 0.634,
+      "step": 6246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6362781272580356e-05,
+      "loss": 0.2874,
+      "step": 6247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.636163901272223e-05,
+      "loss": 0.9195,
+      "step": 6248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6360496585108675e-05,
+      "loss": 0.5631,
+      "step": 6249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.635935398974694e-05,
+      "loss": 0.9091,
+      "step": 6250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6358211226644314e-05,
+      "loss": 1.5332,
+      "step": 6251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6357068295808044e-05,
+      "loss": 0.8041,
+      "step": 6252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6355925197245434e-05,
+      "loss": 0.4109,
+      "step": 6253
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.635478193096373e-05,
+      "loss": 1.3048,
+      "step": 6254
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.635363849697023e-05,
+      "loss": 0.6252,
+      "step": 6255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6352494895272195e-05,
+      "loss": 0.4526,
+      "step": 6256
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6351351125876905e-05,
+      "loss": 0.7332,
+      "step": 6257
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6350207188791644e-05,
+      "loss": 0.7963,
+      "step": 6258
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.634906308402368e-05,
+      "loss": 1.2466,
+      "step": 6259
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6347918811580314e-05,
+      "loss": 0.8124,
+      "step": 6260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6346774371468816e-05,
+      "loss": 0.4097,
+      "step": 6261
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.634562976369646e-05,
+      "loss": 0.9418,
+      "step": 6262
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6344484988270547e-05,
+      "loss": 0.6416,
+      "step": 6263
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.634334004519835e-05,
+      "loss": 1.0823,
+      "step": 6264
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6342194934487155e-05,
+      "loss": 0.5055,
+      "step": 6265
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.634104965614426e-05,
+      "loss": 0.5404,
+      "step": 6266
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6339904210176945e-05,
+      "loss": 0.9994,
+      "step": 6267
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.633875859659249e-05,
+      "loss": 0.5997,
+      "step": 6268
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6337612815398205e-05,
+      "loss": 0.463,
+      "step": 6269
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.633646686660137e-05,
+      "loss": 1.6371,
+      "step": 6270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.633532075020927e-05,
+      "loss": 0.7464,
+      "step": 6271
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.633417446622922e-05,
+      "loss": 1.5413,
+      "step": 6272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.633302801466849e-05,
+      "loss": 0.7445,
+      "step": 6273
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.63318813955344e-05,
+      "loss": 0.7785,
+      "step": 6274
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6330734608834224e-05,
+      "loss": 0.4942,
+      "step": 6275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6329587654575265e-05,
+      "loss": 0.6486,
+      "step": 6276
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.632844053276484e-05,
+      "loss": 0.4273,
+      "step": 6277
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.632729324341023e-05,
+      "loss": 0.821,
+      "step": 6278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6326145786518744e-05,
+      "loss": 0.951,
+      "step": 6279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.632499816209768e-05,
+      "loss": 0.8361,
+      "step": 6280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6323850370154334e-05,
+      "loss": 0.8834,
+      "step": 6281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.632270241069603e-05,
+      "loss": 0.8048,
+      "step": 6282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.632155428373005e-05,
+      "loss": 1.2304,
+      "step": 6283
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6320405989263715e-05,
+      "loss": 1.5066,
+      "step": 6284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.631925752730433e-05,
+      "loss": 0.6731,
+      "step": 6285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.631810889785921e-05,
+      "loss": 1.1658,
+      "step": 6286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.631696010093565e-05,
+      "loss": 0.7289,
+      "step": 6287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.631581113654096e-05,
+      "loss": 0.456,
+      "step": 6288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6314662004682473e-05,
+      "loss": 1.144,
+      "step": 6289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.631351270536748e-05,
+      "loss": 0.7272,
+      "step": 6290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6312363238603305e-05,
+      "loss": 0.984,
+      "step": 6291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.631121360439726e-05,
+      "loss": 1.5511,
+      "step": 6292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.631006380275666e-05,
+      "loss": 1.721,
+      "step": 6293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6308913833688825e-05,
+      "loss": 1.0011,
+      "step": 6294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.630776369720107e-05,
+      "loss": 1.0073,
+      "step": 6295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.630661339330071e-05,
+      "loss": 0.6425,
+      "step": 6296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.630546292199507e-05,
+      "loss": 1.3539,
+      "step": 6297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.630431228329147e-05,
+      "loss": 0.9227,
+      "step": 6298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.630316147719724e-05,
+      "loss": 0.8609,
+      "step": 6299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6302010503719704e-05,
+      "loss": 0.568,
+      "step": 6300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.630085936286617e-05,
+      "loss": 2.0707,
+      "step": 6301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.629970805464397e-05,
+      "loss": 0.4586,
+      "step": 6302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.629855657906044e-05,
+      "loss": 1.3759,
+      "step": 6303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.629740493612289e-05,
+      "loss": 1.0056,
+      "step": 6304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.629625312583866e-05,
+      "loss": 1.6187,
+      "step": 6305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6295101148215084e-05,
+      "loss": 0.3366,
+      "step": 6306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6293949003259485e-05,
+      "loss": 1.9024,
+      "step": 6307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6292796690979195e-05,
+      "loss": 1.2255,
+      "step": 6308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6291644211381567e-05,
+      "loss": 1.0099,
+      "step": 6309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.62904915644739e-05,
+      "loss": 1.6197,
+      "step": 6310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.628933875026355e-05,
+      "loss": 1.3089,
+      "step": 6311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.628818576875784e-05,
+      "loss": 0.5829,
+      "step": 6312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.628703261996413e-05,
+      "loss": 1.3878,
+      "step": 6313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.628587930388973e-05,
+      "loss": 1.1376,
+      "step": 6314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.628472582054201e-05,
+      "loss": 1.0727,
+      "step": 6315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.628357216992828e-05,
+      "loss": 1.0586,
+      "step": 6316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6282418352055906e-05,
+      "loss": 1.3552,
+      "step": 6317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6281264366932214e-05,
+      "loss": 1.3235,
+      "step": 6318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.628011021456456e-05,
+      "loss": 0.9424,
+      "step": 6319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.627895589496028e-05,
+      "loss": 0.7549,
+      "step": 6320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.627780140812672e-05,
+      "loss": 1.6227,
+      "step": 6321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6276646754071225e-05,
+      "loss": 0.6483,
+      "step": 6322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.627549193280115e-05,
+      "loss": 0.4267,
+      "step": 6323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.627433694432384e-05,
+      "loss": 0.3274,
+      "step": 6324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6273181788646644e-05,
+      "loss": 0.8429,
+      "step": 6325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6272026465776916e-05,
+      "loss": 0.8522,
+      "step": 6326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6270870975722e-05,
+      "loss": 1.4739,
+      "step": 6327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.626971531848926e-05,
+      "loss": 0.588,
+      "step": 6328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.626855949408605e-05,
+      "loss": 0.3539,
+      "step": 6329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.626740350251971e-05,
+      "loss": 1.0584,
+      "step": 6330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.626624734379761e-05,
+      "loss": 0.6476,
+      "step": 6331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6265091017927107e-05,
+      "loss": 1.2366,
+      "step": 6332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6263934524915555e-05,
+      "loss": 1.1558,
+      "step": 6333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6262777864770315e-05,
+      "loss": 1.5881,
+      "step": 6334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.626162103749874e-05,
+      "loss": 0.4436,
+      "step": 6335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.626046404310821e-05,
+      "loss": 0.5822,
+      "step": 6336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.625930688160608e-05,
+      "loss": 0.8051,
+      "step": 6337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.62581495529997e-05,
+      "loss": 1.3764,
+      "step": 6338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6256992057296444e-05,
+      "loss": 1.3079,
+      "step": 6339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.625583439450369e-05,
+      "loss": 1.1928,
+      "step": 6340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.625467656462879e-05,
+      "loss": 0.3634,
+      "step": 6341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.625351856767911e-05,
+      "loss": 0.3933,
+      "step": 6342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6252360403662036e-05,
+      "loss": 0.5002,
+      "step": 6343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.625120207258492e-05,
+      "loss": 1.0119,
+      "step": 6344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.625004357445514e-05,
+      "loss": 0.5159,
+      "step": 6345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.624888490928007e-05,
+      "loss": 1.4671,
+      "step": 6346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.624772607706709e-05,
+      "loss": 0.696,
+      "step": 6347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.624656707782355e-05,
+      "loss": 1.6991,
+      "step": 6348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6245407911556855e-05,
+      "loss": 0.5661,
+      "step": 6349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6244248578274364e-05,
+      "loss": 0.5338,
+      "step": 6350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.624308907798346e-05,
+      "loss": 1.6533,
+      "step": 6351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6241929410691514e-05,
+      "loss": 1.1044,
+      "step": 6352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.624076957640592e-05,
+      "loss": 1.0101,
+      "step": 6353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6239609575134044e-05,
+      "loss": 0.111,
+      "step": 6354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623844940688328e-05,
+      "loss": 1.0517,
+      "step": 6355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6237289071661014e-05,
+      "loss": 0.5678,
+      "step": 6356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623612856947461e-05,
+      "loss": 0.3213,
+      "step": 6357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623496790033147e-05,
+      "loss": 0.8694,
+      "step": 6358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623380706423897e-05,
+      "loss": 1.3078,
+      "step": 6359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.62326460612045e-05,
+      "loss": 1.3102,
+      "step": 6360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623148489123545e-05,
+      "loss": 1.0477,
+      "step": 6361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.623032355433922e-05,
+      "loss": 1.2834,
+      "step": 6362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622916205052318e-05,
+      "loss": 0.9289,
+      "step": 6363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622800037979473e-05,
+      "loss": 1.4863,
+      "step": 6364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622683854216127e-05,
+      "loss": 0.3838,
+      "step": 6365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622567653763018e-05,
+      "loss": 1.7641,
+      "step": 6366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622451436620886e-05,
+      "loss": 1.7018,
+      "step": 6367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622335202790472e-05,
+      "loss": 1.6582,
+      "step": 6368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6222189522725135e-05,
+      "loss": 1.0784,
+      "step": 6369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622102685067751e-05,
+      "loss": 0.9793,
+      "step": 6370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621986401176923e-05,
+      "loss": 1.542,
+      "step": 6371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6218701006007734e-05,
+      "loss": 0.284,
+      "step": 6372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621753783340039e-05,
+      "loss": 0.6686,
+      "step": 6373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621637449395461e-05,
+      "loss": 1.4099,
+      "step": 6374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621521098767779e-05,
+      "loss": 0.8683,
+      "step": 6375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621404731457735e-05,
+      "loss": 0.6836,
+      "step": 6376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621288347466068e-05,
+      "loss": 1.5801,
+      "step": 6377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6211719467935194e-05,
+      "loss": 1.3176,
+      "step": 6378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621055529440829e-05,
+      "loss": 1.9497,
+      "step": 6379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.620939095408739e-05,
+      "loss": 1.5606,
+      "step": 6380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6208226446979896e-05,
+      "loss": 1.6519,
+      "step": 6381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.620706177309321e-05,
+      "loss": 1.4848,
+      "step": 6382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6205896932434765e-05,
+      "loss": 1.589,
+      "step": 6383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.620473192501196e-05,
+      "loss": 1.7186,
+      "step": 6384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.62035667508322e-05,
+      "loss": 1.058,
+      "step": 6385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6202401409902925e-05,
+      "loss": 1.2388,
+      "step": 6386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6201235902231525e-05,
+      "loss": 1.3671,
+      "step": 6387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6200070227825424e-05,
+      "loss": 1.7794,
+      "step": 6388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.619890438669205e-05,
+      "loss": 0.9476,
+      "step": 6389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.61977383788388e-05,
+      "loss": 1.1631,
+      "step": 6390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.619657220427313e-05,
+      "loss": 1.0551,
+      "step": 6391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.619540586300243e-05,
+      "loss": 0.8781,
+      "step": 6392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6194239355034136e-05,
+      "loss": 0.8164,
+      "step": 6393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.619307268037566e-05,
+      "loss": 1.2482,
+      "step": 6394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.619190583903444e-05,
+      "loss": 1.3014,
+      "step": 6395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.619073883101789e-05,
+      "loss": 0.2435,
+      "step": 6396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6189571656333456e-05,
+      "loss": 0.4683,
+      "step": 6397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.618840431498854e-05,
+      "loss": 1.4003,
+      "step": 6398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.618723680699058e-05,
+      "loss": 1.537,
+      "step": 6399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.618606913234701e-05,
+      "loss": 1.1421,
+      "step": 6400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6184901291065255e-05,
+      "loss": 1.2,
+      "step": 6401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.618373328315275e-05,
+      "loss": 0.7342,
+      "step": 6402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6182565108616936e-05,
+      "loss": 1.1436,
+      "step": 6403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.618139676746523e-05,
+      "loss": 1.5893,
+      "step": 6404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.618022825970507e-05,
+      "loss": 0.6623,
+      "step": 6405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.61790595853439e-05,
+      "loss": 1.3067,
+      "step": 6406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6177890744389165e-05,
+      "loss": 0.5241,
+      "step": 6407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6176721736848284e-05,
+      "loss": 1.4501,
+      "step": 6408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.61755525627287e-05,
+      "loss": 0.6225,
+      "step": 6409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.617438322203786e-05,
+      "loss": 1.4515,
+      "step": 6410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.61732137147832e-05,
+      "loss": 1.5321,
+      "step": 6411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6172044040972173e-05,
+      "loss": 0.9542,
+      "step": 6412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.61708742006122e-05,
+      "loss": 1.0295,
+      "step": 6413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.616970419371075e-05,
+      "loss": 0.2409,
+      "step": 6414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.616853402027526e-05,
+      "loss": 0.8066,
+      "step": 6415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6167363680313166e-05,
+      "loss": 0.4539,
+      "step": 6416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6166193173831935e-05,
+      "loss": 0.9763,
+      "step": 6417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6165022500838996e-05,
+      "loss": 1.9294,
+      "step": 6418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6163851661341816e-05,
+      "loss": 1.6288,
+      "step": 6419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.616268065534783e-05,
+      "loss": 1.2014,
+      "step": 6420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.616150948286449e-05,
+      "loss": 0.4049,
+      "step": 6421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6160338143899265e-05,
+      "loss": 1.455,
+      "step": 6422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6159166638459605e-05,
+      "loss": 0.5951,
+      "step": 6423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.615799496655295e-05,
+      "loss": 0.7681,
+      "step": 6424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.615682312818677e-05,
+      "loss": 0.203,
+      "step": 6425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.615565112336852e-05,
+      "loss": 1.3271,
+      "step": 6426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6154478952105655e-05,
+      "loss": 0.8279,
+      "step": 6427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.615330661440563e-05,
+      "loss": 1.537,
+      "step": 6428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6152134110275924e-05,
+      "loss": 0.8629,
+      "step": 6429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.615096143972398e-05,
+      "loss": 0.9861,
+      "step": 6430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.614978860275726e-05,
+      "loss": 0.6395,
+      "step": 6431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6148615599383235e-05,
+      "loss": 0.3561,
+      "step": 6432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.614744242960937e-05,
+      "loss": 0.7941,
+      "step": 6433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6146269093443126e-05,
+      "loss": 0.7337,
+      "step": 6434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.614509559089198e-05,
+      "loss": 1.4709,
+      "step": 6435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.614392192196338e-05,
+      "loss": 1.2375,
+      "step": 6436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.614274808666481e-05,
+      "loss": 0.773,
+      "step": 6437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.614157408500374e-05,
+      "loss": 1.3818,
+      "step": 6438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6140399916987636e-05,
+      "loss": 0.4316,
+      "step": 6439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.613922558262398e-05,
+      "loss": 0.3345,
+      "step": 6440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.613805108192022e-05,
+      "loss": 1.4309,
+      "step": 6441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6136876414883855e-05,
+      "loss": 1.2788,
+      "step": 6442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.613570158152235e-05,
+      "loss": 1.1755,
+      "step": 6443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.613452658184318e-05,
+      "loss": 0.7149,
+      "step": 6444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6133351415853834e-05,
+      "loss": 0.4039,
+      "step": 6445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.613217608356177e-05,
+      "loss": 0.5002,
+      "step": 6446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.613100058497448e-05,
+      "loss": 1.2786,
+      "step": 6447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.612982492009945e-05,
+      "loss": 0.4081,
+      "step": 6448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6128649088944156e-05,
+      "loss": 0.6751,
+      "step": 6449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.612747309151606e-05,
+      "loss": 0.7339,
+      "step": 6450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612629692782269e-05,
+      "loss": 1.1287,
+      "step": 6451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612512059787149e-05,
+      "loss": 0.8674,
+      "step": 6452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612394410166997e-05,
+      "loss": 1.1727,
+      "step": 6453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6122767439225595e-05,
+      "loss": 0.7976,
+      "step": 6454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6121590610545876e-05,
+      "loss": 0.7385,
+      "step": 6455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6120413615638286e-05,
+      "loss": 0.8185,
+      "step": 6456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6119236454510325e-05,
+      "loss": 0.4082,
+      "step": 6457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611805912716948e-05,
+      "loss": 1.1989,
+      "step": 6458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611688163362324e-05,
+      "loss": 0.4916,
+      "step": 6459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611570397387911e-05,
+      "loss": 1.145,
+      "step": 6460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611452614794456e-05,
+      "loss": 0.5928,
+      "step": 6461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.61133481558271e-05,
+      "loss": 1.0088,
+      "step": 6462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611216999753424e-05,
+      "loss": 1.0074,
+      "step": 6463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611099167307345e-05,
+      "loss": 0.4567,
+      "step": 6464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610981318245225e-05,
+      "loss": 1.2913,
+      "step": 6465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6108634525678134e-05,
+      "loss": 1.2668,
+      "step": 6466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610745570275859e-05,
+      "loss": 0.1553,
+      "step": 6467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610627671370114e-05,
+      "loss": 0.841,
+      "step": 6468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610509755851327e-05,
+      "loss": 1.2989,
+      "step": 6469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6103918237202496e-05,
+      "loss": 0.5727,
+      "step": 6470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610273874977631e-05,
+      "loss": 0.1386,
+      "step": 6471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610155909624223e-05,
+      "loss": 0.7813,
+      "step": 6472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610037927660775e-05,
+      "loss": 1.296,
+      "step": 6473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.609919929088039e-05,
+      "loss": 0.4123,
+      "step": 6474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6098019139067654e-05,
+      "loss": 0.5474,
+      "step": 6475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6096838821177055e-05,
+      "loss": 0.4822,
+      "step": 6476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60956583372161e-05,
+      "loss": 0.8494,
+      "step": 6477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60944776871923e-05,
+      "loss": 1.192,
+      "step": 6478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.609329687111317e-05,
+      "loss": 0.4175,
+      "step": 6479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.609211588898623e-05,
+      "loss": 0.4691,
+      "step": 6480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6090934740818995e-05,
+      "loss": 0.6943,
+      "step": 6481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.608975342661896e-05,
+      "loss": 0.3324,
+      "step": 6482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.608857194639368e-05,
+      "loss": 1.3532,
+      "step": 6483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.608739030015064e-05,
+      "loss": 1.1129,
+      "step": 6484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6086208487897375e-05,
+      "loss": 1.0825,
+      "step": 6485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60850265096414e-05,
+      "loss": 1.1576,
+      "step": 6486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.608384436539024e-05,
+      "loss": 1.1193,
+      "step": 6487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.608266205515142e-05,
+      "loss": 1.0857,
+      "step": 6488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.608147957893246e-05,
+      "loss": 0.659,
+      "step": 6489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.608029693674088e-05,
+      "loss": 1.0954,
+      "step": 6490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6079114128584224e-05,
+      "loss": 0.449,
+      "step": 6491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6077931154469996e-05,
+      "loss": 1.4029,
+      "step": 6492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607674801440574e-05,
+      "loss": 1.4294,
+      "step": 6493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607556470839898e-05,
+      "loss": 1.2586,
+      "step": 6494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607438123645725e-05,
+      "loss": 0.6851,
+      "step": 6495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607319759858807e-05,
+      "loss": 0.5736,
+      "step": 6496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607201379479898e-05,
+      "loss": 0.4408,
+      "step": 6497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607082982509751e-05,
+      "loss": 1.5142,
+      "step": 6498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60696456894912e-05,
+      "loss": 0.6696,
+      "step": 6499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606846138798758e-05,
+      "loss": 0.2442,
+      "step": 6500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60672769205942e-05,
+      "loss": 1.1904,
+      "step": 6501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6066092287318576e-05,
+      "loss": 0.499,
+      "step": 6502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606490748816826e-05,
+      "loss": 1.1361,
+      "step": 6503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606372252315079e-05,
+      "loss": 0.1875,
+      "step": 6504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606253739227369e-05,
+      "loss": 0.371,
+      "step": 6505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606135209554454e-05,
+      "loss": 0.9752,
+      "step": 6506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6060166632970845e-05,
+      "loss": 1.1526,
+      "step": 6507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605898100456016e-05,
+      "loss": 0.9699,
+      "step": 6508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605779521032004e-05,
+      "loss": 0.2661,
+      "step": 6509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6056609250258025e-05,
+      "loss": 0.5471,
+      "step": 6510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6055423124381654e-05,
+      "loss": 1.2119,
+      "step": 6511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605423683269849e-05,
+      "loss": 1.3029,
+      "step": 6512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605305037521608e-05,
+      "loss": 1.4082,
+      "step": 6513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605186375194196e-05,
+      "loss": 1.1538,
+      "step": 6514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6050676962883686e-05,
+      "loss": 0.667,
+      "step": 6515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.604949000804882e-05,
+      "loss": 0.5963,
+      "step": 6516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.604830288744491e-05,
+      "loss": 0.6737,
+      "step": 6517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.60471156010795e-05,
+      "loss": 0.892,
+      "step": 6518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6045928148960164e-05,
+      "loss": 1.2642,
+      "step": 6519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6044740531094454e-05,
+      "loss": 0.3753,
+      "step": 6520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6043552747489915e-05,
+      "loss": 1.2083,
+      "step": 6521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.604236479815412e-05,
+      "loss": 0.4105,
+      "step": 6522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6041176683094615e-05,
+      "loss": 1.3277,
+      "step": 6523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603998840231897e-05,
+      "loss": 0.8825,
+      "step": 6524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603879995583474e-05,
+      "loss": 1.7342,
+      "step": 6525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603761134364951e-05,
+      "loss": 0.7554,
+      "step": 6526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603642256577081e-05,
+      "loss": 0.6979,
+      "step": 6527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6035233622206224e-05,
+      "loss": 0.7239,
+      "step": 6528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603404451296333e-05,
+      "loss": 0.9194,
+      "step": 6529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6032855238049666e-05,
+      "loss": 2.2672,
+      "step": 6530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603166579747282e-05,
+      "loss": 1.5719,
+      "step": 6531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603047619124036e-05,
+      "loss": 0.9453,
+      "step": 6532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602928641935984e-05,
+      "loss": 1.0777,
+      "step": 6533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602809648183885e-05,
+      "loss": 1.4784,
+      "step": 6534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602690637868496e-05,
+      "loss": 0.4299,
+      "step": 6535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602571610990573e-05,
+      "loss": 1.3521,
+      "step": 6536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602452567550875e-05,
+      "loss": 1.0249,
+      "step": 6537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602333507550158e-05,
+      "loss": 1.6518,
+      "step": 6538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602214430989181e-05,
+      "loss": 2.4295,
+      "step": 6539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602095337868702e-05,
+      "loss": 1.3493,
+      "step": 6540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601976228189477e-05,
+      "loss": 1.7021,
+      "step": 6541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6018571019522666e-05,
+      "loss": 0.3882,
+      "step": 6542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601737959157826e-05,
+      "loss": 1.1686,
+      "step": 6543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601618799806915e-05,
+      "loss": 0.8255,
+      "step": 6544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6014996239002923e-05,
+      "loss": 0.7266,
+      "step": 6545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6013804314387154e-05,
+      "loss": 0.2213,
+      "step": 6546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6012612224229434e-05,
+      "loss": 0.9447,
+      "step": 6547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601141996853734e-05,
+      "loss": 0.8943,
+      "step": 6548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.601022754731847e-05,
+      "loss": 0.6118,
+      "step": 6549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6009034960580395e-05,
+      "loss": 0.5461,
+      "step": 6550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6007842208330726e-05,
+      "loss": 1.0844,
+      "step": 6551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.600664929057704e-05,
+      "loss": 1.6315,
+      "step": 6552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.600545620732693e-05,
+      "loss": 0.8147,
+      "step": 6553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.600426295858799e-05,
+      "loss": 0.7028,
+      "step": 6554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.600306954436782e-05,
+      "loss": 0.6697,
+      "step": 6555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.600187596467401e-05,
+      "loss": 0.6428,
+      "step": 6556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.6000682219514136e-05,
+      "loss": 1.8998,
+      "step": 6557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.599948830889582e-05,
+      "loss": 0.7806,
+      "step": 6558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.599829423282665e-05,
+      "loss": 1.0993,
+      "step": 6559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5997099991314226e-05,
+      "loss": 0.9135,
+      "step": 6560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.599590558436615e-05,
+      "loss": 0.5895,
+      "step": 6561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.599471101199002e-05,
+      "loss": 0.284,
+      "step": 6562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.599351627419344e-05,
+      "loss": 0.3841,
+      "step": 6563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5992321370984005e-05,
+      "loss": 0.9825,
+      "step": 6564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.599112630236933e-05,
+      "loss": 0.7597,
+      "step": 6565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.598993106835701e-05,
+      "loss": 1.4918,
+      "step": 6566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.598873566895466e-05,
+      "loss": 0.398,
+      "step": 6567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5987540104169875e-05,
+      "loss": 1.1451,
+      "step": 6568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.598634437401029e-05,
+      "loss": 1.0337,
+      "step": 6569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5985148478483476e-05,
+      "loss": 1.181,
+      "step": 6570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.598395241759707e-05,
+      "loss": 1.5482,
+      "step": 6571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5982756191358677e-05,
+      "loss": 0.725,
+      "step": 6572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.59815597997759e-05,
+      "loss": 1.1179,
+      "step": 6573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.598036324285637e-05,
+      "loss": 1.2265,
+      "step": 6574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.597916652060769e-05,
+      "loss": 0.4688,
+      "step": 6575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.597796963303747e-05,
+      "loss": 0.1925,
+      "step": 6576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5976772580153336e-05,
+      "loss": 0.3375,
+      "step": 6577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.597557536196291e-05,
+      "loss": 0.4067,
+      "step": 6578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.59743779784738e-05,
+      "loss": 0.4389,
+      "step": 6579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.597318042969363e-05,
+      "loss": 1.5486,
+      "step": 6580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.597198271563002e-05,
+      "loss": 0.3599,
+      "step": 6581
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.597078483629059e-05,
+      "loss": 0.3729,
+      "step": 6582
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5969586791682965e-05,
+      "loss": 1.0455,
+      "step": 6583
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.596838858181477e-05,
+      "loss": 1.1502,
+      "step": 6584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.596719020669363e-05,
+      "loss": 0.4308,
+      "step": 6585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.596599166632718e-05,
+      "loss": 1.0201,
+      "step": 6586
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5964792960723025e-05,
+      "loss": 1.1048,
+      "step": 6587
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.59635940898888e-05,
+      "loss": 1.1499,
+      "step": 6588
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.596239505383215e-05,
+      "loss": 0.9997,
+      "step": 6589
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5961195852560686e-05,
+      "loss": 0.6846,
+      "step": 6590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.595999648608205e-05,
+      "loss": 0.9657,
+      "step": 6591
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.595879695440388e-05,
+      "loss": 0.3483,
+      "step": 6592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5957597257533794e-05,
+      "loss": 0.5793,
+      "step": 6593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5956397395479436e-05,
+      "loss": 0.239,
+      "step": 6594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5955197368248435e-05,
+      "loss": 0.8349,
+      "step": 6595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.595399717584844e-05,
+      "loss": 0.7325,
+      "step": 6596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.595279681828708e-05,
+      "loss": 0.6636,
+      "step": 6597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.595159629557199e-05,
+      "loss": 0.426,
+      "step": 6598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5950395607710813e-05,
+      "loss": 0.4585,
+      "step": 6599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5949194754711194e-05,
+      "loss": 0.6015,
+      "step": 6600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5947993736580774e-05,
+      "loss": 0.5082,
+      "step": 6601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.594679255332718e-05,
+      "loss": 1.1892,
+      "step": 6602
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.594559120495808e-05,
+      "loss": 0.2274,
+      "step": 6603
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.59443896914811e-05,
+      "loss": 0.6143,
+      "step": 6604
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.59431880129039e-05,
+      "loss": 0.4675,
+      "step": 6605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5941986169234116e-05,
+      "loss": 1.5255,
+      "step": 6606
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5940784160479405e-05,
+      "loss": 0.5801,
+      "step": 6607
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.593958198664741e-05,
+      "loss": 1.1085,
+      "step": 6608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.593837964774578e-05,
+      "loss": 0.799,
+      "step": 6609
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.593717714378218e-05,
+      "loss": 0.8621,
+      "step": 6610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.593597447476425e-05,
+      "loss": 0.2916,
+      "step": 6611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.593477164069963e-05,
+      "loss": 1.2889,
+      "step": 6612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5933568641596e-05,
+      "loss": 0.6115,
+      "step": 6613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.593236547746101e-05,
+      "loss": 0.8539,
+      "step": 6614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.59311621483023e-05,
+      "loss": 1.0,
+      "step": 6615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.592995865412754e-05,
+      "loss": 1.3169,
+      "step": 6616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.59287549949444e-05,
+      "loss": 1.4179,
+      "step": 6617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5927551170760516e-05,
+      "loss": 1.2758,
+      "step": 6618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.592634718158356e-05,
+      "loss": 1.1877,
+      "step": 6619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5925143027421196e-05,
+      "loss": 0.6576,
+      "step": 6620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.592393870828108e-05,
+      "loss": 0.3708,
+      "step": 6621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.592273422417089e-05,
+      "loss": 1.258,
+      "step": 6622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5921529575098274e-05,
+      "loss": 1.5016,
+      "step": 6623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.59203247610709e-05,
+      "loss": 0.4456,
+      "step": 6624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5919119782096444e-05,
+      "loss": 1.0385,
+      "step": 6625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5917914638182575e-05,
+      "loss": 0.7887,
+      "step": 6626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.591670932933696e-05,
+      "loss": 1.0313,
+      "step": 6627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5915503855567257e-05,
+      "loss": 1.2183,
+      "step": 6628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.591429821688115e-05,
+      "loss": 0.7771,
+      "step": 6629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5913092413286304e-05,
+      "loss": 0.7827,
+      "step": 6630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5911886444790395e-05,
+      "loss": 0.6964,
+      "step": 6631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5910680311401104e-05,
+      "loss": 0.4145,
+      "step": 6632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5909474013126096e-05,
+      "loss": 0.573,
+      "step": 6633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5908267549973056e-05,
+      "loss": 0.4061,
+      "step": 6634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.590706092194965e-05,
+      "loss": 0.2768,
+      "step": 6635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.590585412906357e-05,
+      "loss": 0.8308,
+      "step": 6636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5904647171322485e-05,
+      "loss": 0.8467,
+      "step": 6637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.590344004873408e-05,
+      "loss": 1.3283,
+      "step": 6638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.590223276130604e-05,
+      "loss": 1.0815,
+      "step": 6639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.590102530904604e-05,
+      "loss": 0.7376,
+      "step": 6640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.589981769196178e-05,
+      "loss": 1.0255,
+      "step": 6641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5898609910060925e-05,
+      "loss": 0.3677,
+      "step": 6642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5897401963351165e-05,
+      "loss": 0.2478,
+      "step": 6643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5896193851840194e-05,
+      "loss": 0.5134,
+      "step": 6644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5894985575535704e-05,
+      "loss": 0.6294,
+      "step": 6645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5893777134445366e-05,
+      "loss": 0.5355,
+      "step": 6646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.5892568528576884e-05,
+      "loss": 0.5519,
+      "step": 6647
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.589135975793795e-05,
+      "loss": 1.1547,
+      "step": 6648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5890150822536245e-05,
+      "loss": 0.6924,
+      "step": 6649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.588894172237948e-05,
+      "loss": 0.6212,
+      "step": 6650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.588773245747532e-05,
+      "loss": 1.8063,
+      "step": 6651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.58865230278315e-05,
+      "loss": 0.3519,
+      "step": 6652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.588531343345568e-05,
+      "loss": 1.265,
+      "step": 6653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.588410367435558e-05,
+      "loss": 1.6286,
+      "step": 6654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.588289375053889e-05,
+      "loss": 1.4976,
+      "step": 6655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.588168366201331e-05,
+      "loss": 1.3441,
+      "step": 6656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.588047340878654e-05,
+      "loss": 1.5602,
+      "step": 6657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.587926299086628e-05,
+      "loss": 0.5258,
+      "step": 6658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.587805240826024e-05,
+      "loss": 0.035,
+      "step": 6659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5876841660976115e-05,
+      "loss": 0.6742,
+      "step": 6660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.587563074902162e-05,
+      "loss": 0.6306,
+      "step": 6661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5874419672404454e-05,
+      "loss": 0.9669,
+      "step": 6662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5873208431132324e-05,
+      "loss": 0.5492,
+      "step": 6663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.587199702521293e-05,
+      "loss": 0.9843,
+      "step": 6664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5870785454654e-05,
+      "loss": 0.3885,
+      "step": 6665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.586957371946323e-05,
+      "loss": 1.1536,
+      "step": 6666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5868361819648327e-05,
+      "loss": 0.5456,
+      "step": 6667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.586714975521701e-05,
+      "loss": 1.7256,
+      "step": 6668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5865937526177e-05,
+      "loss": 0.4804,
+      "step": 6669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5864725132536e-05,
+      "loss": 1.5284,
+      "step": 6670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.586351257430173e-05,
+      "loss": 0.6295,
+      "step": 6671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.586229985148191e-05,
+      "loss": 0.3191,
+      "step": 6672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.586108696408424e-05,
+      "loss": 1.6732,
+      "step": 6673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.585987391211646e-05,
+      "loss": 1.4483,
+      "step": 6674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.585866069558628e-05,
+      "loss": 0.3998,
+      "step": 6675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5857447314501416e-05,
+      "loss": 0.3373,
+      "step": 6676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.58562337688696e-05,
+      "loss": 1.1602,
+      "step": 6677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.585502005869854e-05,
+      "loss": 1.1035,
+      "step": 6678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.585380618399597e-05,
+      "loss": 1.3542,
+      "step": 6679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5852592144769626e-05,
+      "loss": 1.0772,
+      "step": 6680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5851377941027206e-05,
+      "loss": 0.7515,
+      "step": 6681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5850163572776454e-05,
+      "loss": 0.9162,
+      "step": 6682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.584894904002509e-05,
+      "loss": 0.848,
+      "step": 6683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.584773434278086e-05,
+      "loss": 1.1858,
+      "step": 6684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.584651948105147e-05,
+      "loss": 0.3852,
+      "step": 6685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.584530445484467e-05,
+      "loss": 0.6094,
+      "step": 6686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.584408926416818e-05,
+      "loss": 0.7623,
+      "step": 6687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.584287390902975e-05,
+      "loss": 0.6181,
+      "step": 6688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.584165838943709e-05,
+      "loss": 0.7254,
+      "step": 6689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.584044270539795e-05,
+      "loss": 0.8814,
+      "step": 6690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5839226856920055e-05,
+      "loss": 0.4072,
+      "step": 6691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.583801084401116e-05,
+      "loss": 1.3228,
+      "step": 6692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.583679466667899e-05,
+      "loss": 0.3818,
+      "step": 6693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5835578324931276e-05,
+      "loss": 0.97,
+      "step": 6694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.583436181877579e-05,
+      "loss": 0.5019,
+      "step": 6695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5833145148220244e-05,
+      "loss": 1.0001,
+      "step": 6696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5831928313272384e-05,
+      "loss": 0.5888,
+      "step": 6697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.583071131393997e-05,
+      "loss": 0.6838,
+      "step": 6698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.582949415023072e-05,
+      "loss": 1.1049,
+      "step": 6699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.582827682215241e-05,
+      "loss": 0.4327,
+      "step": 6700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.582705932971277e-05,
+      "loss": 0.9336,
+      "step": 6701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.582584167291955e-05,
+      "loss": 1.2149,
+      "step": 6702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5824623851780496e-05,
+      "loss": 0.7307,
+      "step": 6703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.582340586630335e-05,
+      "loss": 0.1788,
+      "step": 6704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.582218771649589e-05,
+      "loss": 1.5952,
+      "step": 6705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5820969402365846e-05,
+      "loss": 0.5463,
+      "step": 6706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.581975092392097e-05,
+      "loss": 0.3929,
+      "step": 6707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5818532281169024e-05,
+      "loss": 0.5124,
+      "step": 6708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.581731347411776e-05,
+      "loss": 1.6207,
+      "step": 6709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.581609450277493e-05,
+      "loss": 1.2481,
+      "step": 6710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5814875367148306e-05,
+      "loss": 1.3283,
+      "step": 6711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5813656067245624e-05,
+      "loss": 0.2363,
+      "step": 6712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5812436603074664e-05,
+      "loss": 0.3779,
+      "step": 6713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.581121697464317e-05,
+      "loss": 0.9896,
+      "step": 6714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5809997181958914e-05,
+      "loss": 1.3792,
+      "step": 6715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.580877722502966e-05,
+      "loss": 0.5194,
+      "step": 6716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.580755710386316e-05,
+      "loss": 0.6583,
+      "step": 6717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.580633681846718e-05,
+      "loss": 1.6219,
+      "step": 6718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.580511636884949e-05,
+      "loss": 1.5013,
+      "step": 6719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.580389575501787e-05,
+      "loss": 1.0613,
+      "step": 6720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.580267497698007e-05,
+      "loss": 0.456,
+      "step": 6721
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5801454034743847e-05,
+      "loss": 1.5988,
+      "step": 6722
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5800232928316996e-05,
+      "loss": 1.13,
+      "step": 6723
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5799011657707276e-05,
+      "loss": 0.9465,
+      "step": 6724
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5797790222922465e-05,
+      "loss": 1.4394,
+      "step": 6725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.579656862397033e-05,
+      "loss": 1.4034,
+      "step": 6726
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.579534686085864e-05,
+      "loss": 1.1475,
+      "step": 6727
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.579412493359518e-05,
+      "loss": 1.0163,
+      "step": 6728
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.579290284218771e-05,
+      "loss": 1.5692,
+      "step": 6729
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.579168058664404e-05,
+      "loss": 1.6803,
+      "step": 6730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.579045816697191e-05,
+      "loss": 0.4754,
+      "step": 6731
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.578923558317912e-05,
+      "loss": 1.1201,
+      "step": 6732
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.578801283527345e-05,
+      "loss": 0.5535,
+      "step": 6733
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5786789923262676e-05,
+      "loss": 0.4212,
+      "step": 6734
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.578556684715457e-05,
+      "loss": 0.4988,
+      "step": 6735
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.578434360695694e-05,
+      "loss": 0.3254,
+      "step": 6736
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.578312020267755e-05,
+      "loss": 1.4116,
+      "step": 6737
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.578189663432419e-05,
+      "loss": 0.6849,
+      "step": 6738
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.578067290190465e-05,
+      "loss": 1.6321,
+      "step": 6739
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5779449005426714e-05,
+      "loss": 0.8629,
+      "step": 6740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.577822494489817e-05,
+      "loss": 1.2138,
+      "step": 6741
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.5777000720326805e-05,
+      "loss": 0.5457,
+      "step": 6742
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.577577633172042e-05,
+      "loss": 0.7467,
+      "step": 6743
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.57745517790868e-05,
+      "loss": 0.6213,
+      "step": 6744
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.577332706243374e-05,
+      "loss": 0.8625,
+      "step": 6745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.577210218176902e-05,
+      "loss": 0.3999,
+      "step": 6746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.577087713710046e-05,
+      "loss": 0.9574,
+      "step": 6747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.576965192843583e-05,
+      "loss": 1.4109,
+      "step": 6748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.576842655578295e-05,
+      "loss": 1.0735,
+      "step": 6749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.57672010191496e-05,
+      "loss": 0.8159,
+      "step": 6750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.576597531854358e-05,
+      "loss": 0.5662,
+      "step": 6751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.57647494539727e-05,
+      "loss": 1.2101,
+      "step": 6752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.576352342544475e-05,
+      "loss": 0.8105,
+      "step": 6753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.576229723296755e-05,
+      "loss": 0.2478,
+      "step": 6754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.576107087654888e-05,
+      "loss": 1.3228,
+      "step": 6755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5759844356196566e-05,
+      "loss": 1.0222,
+      "step": 6756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.57586176719184e-05,
+      "loss": 1.1169,
+      "step": 6757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.575739082372218e-05,
+      "loss": 1.48,
+      "step": 6758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.575616381161573e-05,
+      "loss": 0.9083,
+      "step": 6759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.575493663560685e-05,
+      "loss": 0.4419,
+      "step": 6760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5753709295703355e-05,
+      "loss": 0.4147,
+      "step": 6761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.575248179191305e-05,
+      "loss": 0.717,
+      "step": 6762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.575125412424374e-05,
+      "loss": 0.4649,
+      "step": 6763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.575002629270325e-05,
+      "loss": 1.178,
+      "step": 6764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.574879829729939e-05,
+      "loss": 0.7845,
+      "step": 6765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.574757013803997e-05,
+      "loss": 1.0524,
+      "step": 6766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5746341814932815e-05,
+      "loss": 0.5381,
+      "step": 6767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5745113327985724e-05,
+      "loss": 1.3406,
+      "step": 6768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.574388467720654e-05,
+      "loss": 0.8985,
+      "step": 6769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.574265586260306e-05,
+      "loss": 1.1663,
+      "step": 6770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.574142688418311e-05,
+      "loss": 1.4574,
+      "step": 6771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.574019774195451e-05,
+      "loss": 1.4898,
+      "step": 6772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.573896843592508e-05,
+      "loss": 0.4722,
+      "step": 6773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5737738966102656e-05,
+      "loss": 1.2256,
+      "step": 6774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5736509332495054e-05,
+      "loss": 1.1624,
+      "step": 6775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.57352795351101e-05,
+      "loss": 0.4647,
+      "step": 6776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5734049573955596e-05,
+      "loss": 0.8919,
+      "step": 6777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.573281944903941e-05,
+      "loss": 1.1792,
+      "step": 6778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.573158916036934e-05,
+      "loss": 1.3449,
+      "step": 6779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.573035870795323e-05,
+      "loss": 0.628,
+      "step": 6780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5729128091798915e-05,
+      "loss": 1.3161,
+      "step": 6781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5727897311914195e-05,
+      "loss": 0.3327,
+      "step": 6782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.572666636830694e-05,
+      "loss": 0.8652,
+      "step": 6783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.572543526098496e-05,
+      "loss": 1.0646,
+      "step": 6784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.572420398995611e-05,
+      "loss": 0.4997,
+      "step": 6785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.57229725552282e-05,
+      "loss": 0.9058,
+      "step": 6786
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.572174095680907e-05,
+      "loss": 0.8136,
+      "step": 6787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.572050919470658e-05,
+      "loss": 0.671,
+      "step": 6788
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.571927726892855e-05,
+      "loss": 0.9299,
+      "step": 6789
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.571804517948281e-05,
+      "loss": 0.322,
+      "step": 6790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5716812926377235e-05,
+      "loss": 0.8561,
+      "step": 6791
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.571558050961964e-05,
+      "loss": 0.4681,
+      "step": 6792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.571434792921787e-05,
+      "loss": 2.0253,
+      "step": 6793
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5713115185179774e-05,
+      "loss": 0.8585,
+      "step": 6794
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5711882277513196e-05,
+      "loss": 1.0871,
+      "step": 6795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.571064920622597e-05,
+      "loss": 0.6074,
+      "step": 6796
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5709415971325966e-05,
+      "loss": 0.8199,
+      "step": 6797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.570818257282101e-05,
+      "loss": 1.2441,
+      "step": 6798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5706949010718966e-05,
+      "loss": 1.0695,
+      "step": 6799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5705715285027676e-05,
+      "loss": 0.2298,
+      "step": 6800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5704481395755e-05,
+      "loss": 1.0657,
+      "step": 6801
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.570324734290877e-05,
+      "loss": 0.2195,
+      "step": 6802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.570201312649686e-05,
+      "loss": 1.5363,
+      "step": 6803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5700778746527115e-05,
+      "loss": 1.3029,
+      "step": 6804
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5699544203007395e-05,
+      "loss": 1.2274,
+      "step": 6805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.569830949594554e-05,
+      "loss": 0.5465,
+      "step": 6806
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.569707462534943e-05,
+      "loss": 1.0731,
+      "step": 6807
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5695839591226906e-05,
+      "loss": 0.5973,
+      "step": 6808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5694604393585844e-05,
+      "loss": 1.501,
+      "step": 6809
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5693369032434084e-05,
+      "loss": 0.8963,
+      "step": 6810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.56921335077795e-05,
+      "loss": 1.3337,
+      "step": 6811
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.569089781962996e-05,
+      "loss": 0.4143,
+      "step": 6812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.56896619679933e-05,
+      "loss": 1.7526,
+      "step": 6813
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.568842595287743e-05,
+      "loss": 0.948,
+      "step": 6814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.568718977429018e-05,
+      "loss": 0.3008,
+      "step": 6815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5685953432239415e-05,
+      "loss": 1.2517,
+      "step": 6816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.568471692673303e-05,
+      "loss": 0.8701,
+      "step": 6817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.568348025777887e-05,
+      "loss": 1.3627,
+      "step": 6818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.568224342538481e-05,
+      "loss": 1.431,
+      "step": 6819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.568100642955872e-05,
+      "loss": 1.2611,
+      "step": 6820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5679769270308485e-05,
+      "loss": 0.3546,
+      "step": 6821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.567853194764196e-05,
+      "loss": 0.4261,
+      "step": 6822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.567729446156703e-05,
+      "loss": 0.9388,
+      "step": 6823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5676056812091564e-05,
+      "loss": 1.7609,
+      "step": 6824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5674818999223436e-05,
+      "loss": 1.0226,
+      "step": 6825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.567358102297053e-05,
+      "loss": 1.0928,
+      "step": 6826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5672342883340724e-05,
+      "loss": 0.0844,
+      "step": 6827
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.56711045803419e-05,
+      "loss": 1.2041,
+      "step": 6828
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.566986611398191e-05,
+      "loss": 0.425,
+      "step": 6829
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5668627484268677e-05,
+      "loss": 0.8692,
+      "step": 6830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5667388691210065e-05,
+      "loss": 1.1633,
+      "step": 6831
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.566614973481395e-05,
+      "loss": 0.6814,
+      "step": 6832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.566491061508822e-05,
+      "loss": 2.1678,
+      "step": 6833
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5663671332040765e-05,
+      "loss": 0.7793,
+      "step": 6834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.5662431885679465e-05,
+      "loss": 0.9263,
+      "step": 6835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.566119227601221e-05,
+      "loss": 0.5012,
+      "step": 6836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.56599525030469e-05,
+      "loss": 0.3793,
+      "step": 6837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.56587125667914e-05,
+      "loss": 1.5004,
+      "step": 6838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.565747246725362e-05,
+      "loss": 0.9004,
+      "step": 6839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.565623220444145e-05,
+      "loss": 1.3938,
+      "step": 6840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.565499177836278e-05,
+      "loss": 0.4137,
+      "step": 6841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.56537511890255e-05,
+      "loss": 0.9833,
+      "step": 6842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.56525104364375e-05,
+      "loss": 1.2897,
+      "step": 6843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.565126952060668e-05,
+      "loss": 1.3202,
+      "step": 6844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.565002844154095e-05,
+      "loss": 0.9739,
+      "step": 6845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.564878719924819e-05,
+      "loss": 0.8044,
+      "step": 6846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.564754579373631e-05,
+      "loss": 0.6042,
+      "step": 6847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.56463042250132e-05,
+      "loss": 1.0278,
+      "step": 6848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.564506249308677e-05,
+      "loss": 1.4285,
+      "step": 6849
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.564382059796492e-05,
+      "loss": 1.1268,
+      "step": 6850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.564257853965554e-05,
+      "loss": 0.6814,
+      "step": 6851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.564133631816656e-05,
+      "loss": 0.5567,
+      "step": 6852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.564009393350586e-05,
+      "loss": 0.5883,
+      "step": 6853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5638851385681354e-05,
+      "loss": 0.5838,
+      "step": 6854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.563760867470096e-05,
+      "loss": 0.3098,
+      "step": 6855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.563636580057258e-05,
+      "loss": 1.3047,
+      "step": 6856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.563512276330412e-05,
+      "loss": 0.7914,
+      "step": 6857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.563387956290348e-05,
+      "loss": 1.0754,
+      "step": 6858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.563263619937859e-05,
+      "loss": 0.3486,
+      "step": 6859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.563139267273735e-05,
+      "loss": 0.9506,
+      "step": 6860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5630148982987684e-05,
+      "loss": 0.497,
+      "step": 6861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.56289051301375e-05,
+      "loss": 0.5302,
+      "step": 6862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.562766111419471e-05,
+      "loss": 0.6315,
+      "step": 6863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.562641693516723e-05,
+      "loss": 0.4337,
+      "step": 6864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.562517259306299e-05,
+      "loss": 1.6163,
+      "step": 6865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5623928087889885e-05,
+      "loss": 0.34,
+      "step": 6866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.562268341965586e-05,
+      "loss": 0.9775,
+      "step": 6867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.562143858836882e-05,
+      "loss": 0.9578,
+      "step": 6868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.56201935940367e-05,
+      "loss": 0.7484,
+      "step": 6869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.561894843666741e-05,
+      "loss": 0.6229,
+      "step": 6870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.561770311626889e-05,
+      "loss": 0.4374,
+      "step": 6871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.561645763284903e-05,
+      "loss": 1.3066,
+      "step": 6872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5615211986415796e-05,
+      "loss": 0.2013,
+      "step": 6873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.561396617697709e-05,
+      "loss": 1.3034,
+      "step": 6874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.561272020454085e-05,
+      "loss": 1.2428,
+      "step": 6875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5611474069115005e-05,
+      "loss": 1.4791,
+      "step": 6876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.561022777070747e-05,
+      "loss": 0.6689,
+      "step": 6877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.56089813093262e-05,
+      "loss": 1.1546,
+      "step": 6878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5607734684979104e-05,
+      "loss": 1.5065,
+      "step": 6879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.560648789767414e-05,
+      "loss": 1.053,
+      "step": 6880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.560524094741922e-05,
+      "loss": 0.2877,
+      "step": 6881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.560399383422229e-05,
+      "loss": 1.6477,
+      "step": 6882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.560274655809129e-05,
+      "loss": 1.2981,
+      "step": 6883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.560149911903413e-05,
+      "loss": 1.4068,
+      "step": 6884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5600251517058784e-05,
+      "loss": 0.6455,
+      "step": 6885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.559900375217318e-05,
+      "loss": 0.3772,
+      "step": 6886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.559775582438525e-05,
+      "loss": 1.2083,
+      "step": 6887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.559650773370294e-05,
+      "loss": 1.5028,
+      "step": 6888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.559525948013419e-05,
+      "loss": 0.5938,
+      "step": 6889
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5594011063686944e-05,
+      "loss": 1.2538,
+      "step": 6890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5592762484369154e-05,
+      "loss": 1.436,
+      "step": 6891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5591513742188754e-05,
+      "loss": 0.7956,
+      "step": 6892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5590264837153694e-05,
+      "loss": 1.1416,
+      "step": 6893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.558901576927193e-05,
+      "loss": 0.7964,
+      "step": 6894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5587766538551405e-05,
+      "loss": 0.5113,
+      "step": 6895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.558651714500007e-05,
+      "loss": 0.1793,
+      "step": 6896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5585267588625856e-05,
+      "loss": 0.2384,
+      "step": 6897
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5584017869436753e-05,
+      "loss": 1.1092,
+      "step": 6898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.558276798744068e-05,
+      "loss": 1.2541,
+      "step": 6899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5581517942645616e-05,
+      "loss": 1.6584,
+      "step": 6900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.558026773505949e-05,
+      "loss": 0.4468,
+      "step": 6901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.557901736469028e-05,
+      "loss": 1.777,
+      "step": 6902
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.557776683154593e-05,
+      "loss": 0.1909,
+      "step": 6903
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.55765161356344e-05,
+      "loss": 0.5696,
+      "step": 6904
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5575265276963654e-05,
+      "loss": 1.5832,
+      "step": 6905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.557401425554165e-05,
+      "loss": 0.9129,
+      "step": 6906
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5572763071376346e-05,
+      "loss": 0.4653,
+      "step": 6907
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.557151172447571e-05,
+      "loss": 0.3108,
+      "step": 6908
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.557026021484771e-05,
+      "loss": 0.9693,
+      "step": 6909
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.556900854250028e-05,
+      "loss": 0.4599,
+      "step": 6910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5567756707441416e-05,
+      "loss": 0.5409,
+      "step": 6911
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.556650470967908e-05,
+      "loss": 0.9869,
+      "step": 6912
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5565252549221235e-05,
+      "loss": 0.6921,
+      "step": 6913
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5564000226075837e-05,
+      "loss": 0.8927,
+      "step": 6914
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5562747740250876e-05,
+      "loss": 1.0008,
+      "step": 6915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.55614950917543e-05,
+      "loss": 0.3182,
+      "step": 6916
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.556024228059411e-05,
+      "loss": 0.3258,
+      "step": 6917
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5558989306778255e-05,
+      "loss": 1.6387,
+      "step": 6918
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.555773617031472e-05,
+      "loss": 0.2936,
+      "step": 6919
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.555648287121147e-05,
+      "loss": 1.2429,
+      "step": 6920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.555522940947649e-05,
+      "loss": 1.076,
+      "step": 6921
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.555397578511774e-05,
+      "loss": 0.6693,
+      "step": 6922
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.555272199814322e-05,
+      "loss": 1.3831,
+      "step": 6923
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5551468048560884e-05,
+      "loss": 0.831,
+      "step": 6924
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5550213936378746e-05,
+      "loss": 0.9701,
+      "step": 6925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.554895966160476e-05,
+      "loss": 1.0473,
+      "step": 6926
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.55477052242469e-05,
+      "loss": 1.6604,
+      "step": 6927
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.554645062431317e-05,
+      "loss": 0.9059,
+      "step": 6928
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.554519586181155e-05,
+      "loss": 1.0997,
+      "step": 6929
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.554394093675003e-05,
+      "loss": 0.8012,
+      "step": 6930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5542685849136576e-05,
+      "loss": 0.652,
+      "step": 6931
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5541430598979185e-05,
+      "loss": 0.7191,
+      "step": 6932
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.554017518628585e-05,
+      "loss": 0.4698,
+      "step": 6933
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5538919611064554e-05,
+      "loss": 0.9184,
+      "step": 6934
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.553766387332329e-05,
+      "loss": 0.2947,
+      "step": 6935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5536407973070045e-05,
+      "loss": 1.0942,
+      "step": 6936
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.553515191031282e-05,
+      "loss": 1.7388,
+      "step": 6937
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.553389568505961e-05,
+      "loss": 0.6733,
+      "step": 6938
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.553263929731838e-05,
+      "loss": 0.7658,
+      "step": 6939
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.553138274709717e-05,
+      "loss": 0.6323,
+      "step": 6940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5530126034403934e-05,
+      "loss": 1.7257,
+      "step": 6941
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.55288691592467e-05,
+      "loss": 0.2714,
+      "step": 6942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.552761212163345e-05,
+      "loss": 0.6782,
+      "step": 6943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5526354921572184e-05,
+      "loss": 0.538,
+      "step": 6944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.552509755907091e-05,
+      "loss": 0.4969,
+      "step": 6945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5523840034137625e-05,
+      "loss": 1.389,
+      "step": 6946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.552258234678033e-05,
+      "loss": 1.307,
+      "step": 6947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5521324497007036e-05,
+      "loss": 1.4707,
+      "step": 6948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.552006648482573e-05,
+      "loss": 0.4425,
+      "step": 6949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.551880831024444e-05,
+      "loss": 0.4528,
+      "step": 6950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.551754997327116e-05,
+      "loss": 1.3308,
+      "step": 6951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.55162914739139e-05,
+      "loss": 0.4962,
+      "step": 6952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.551503281218066e-05,
+      "loss": 1.1567,
+      "step": 6953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.551377398807946e-05,
+      "loss": 1.3808,
+      "step": 6954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5512515001618316e-05,
+      "loss": 0.6121,
+      "step": 6955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5511255852805235e-05,
+      "loss": 0.5543,
+      "step": 6956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.550999654164821e-05,
+      "loss": 0.7303,
+      "step": 6957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.550873706815529e-05,
+      "loss": 0.3553,
+      "step": 6958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.550747743233446e-05,
+      "loss": 0.209,
+      "step": 6959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.550621763419375e-05,
+      "loss": 0.2756,
+      "step": 6960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.550495767374117e-05,
+      "loss": 0.9449,
+      "step": 6961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.550369755098475e-05,
+      "loss": 0.9764,
+      "step": 6962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.550243726593249e-05,
+      "loss": 1.7178,
+      "step": 6963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.550117681859243e-05,
+      "loss": 0.818,
+      "step": 6964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.549991620897258e-05,
+      "loss": 0.398,
+      "step": 6965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.549865543708096e-05,
+      "loss": 1.1128,
+      "step": 6966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.549739450292561e-05,
+      "loss": 0.6232,
+      "step": 6967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.549613340651452e-05,
+      "loss": 1.1919,
+      "step": 6968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.549487214785575e-05,
+      "loss": 0.8345,
+      "step": 6969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.549361072695731e-05,
+      "loss": 0.4702,
+      "step": 6970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.549234914382723e-05,
+      "loss": 0.7218,
+      "step": 6971
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.549108739847354e-05,
+      "loss": 0.2592,
+      "step": 6972
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.548982549090426e-05,
+      "loss": 1.1147,
+      "step": 6973
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.548856342112742e-05,
+      "loss": 1.9616,
+      "step": 6974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.548730118915108e-05,
+      "loss": 0.2028,
+      "step": 6975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.548603879498325e-05,
+      "loss": 0.6868,
+      "step": 6976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5484776238631945e-05,
+      "loss": 0.3983,
+      "step": 6977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5483513520105234e-05,
+      "loss": 1.1311,
+      "step": 6978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5482250639411124e-05,
+      "loss": 0.9592,
+      "step": 6979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.548098759655768e-05,
+      "loss": 0.4212,
+      "step": 6980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5479724391552915e-05,
+      "loss": 0.5652,
+      "step": 6981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5478461024404884e-05,
+      "loss": 1.2669,
+      "step": 6982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.547719749512161e-05,
+      "loss": 0.6763,
+      "step": 6983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5475933803711155e-05,
+      "loss": 0.7539,
+      "step": 6984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.547466995018154e-05,
+      "loss": 0.7651,
+      "step": 6985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.547340593454082e-05,
+      "loss": 0.85,
+      "step": 6986
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5472141756797036e-05,
+      "loss": 0.3096,
+      "step": 6987
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.547087741695823e-05,
+      "loss": 1.3079,
+      "step": 6988
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.546961291503245e-05,
+      "loss": 0.8287,
+      "step": 6989
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5468348251027745e-05,
+      "loss": 0.8854,
+      "step": 6990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5467083424952164e-05,
+      "loss": 0.4501,
+      "step": 6991
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5465818436813735e-05,
+      "loss": 0.6999,
+      "step": 6992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.546455328662054e-05,
+      "loss": 0.4037,
+      "step": 6993
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.546328797438061e-05,
+      "loss": 1.4137,
+      "step": 6994
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5462022500102e-05,
+      "loss": 0.4679,
+      "step": 6995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.546075686379278e-05,
+      "loss": 0.5804,
+      "step": 6996
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.545949106546098e-05,
+      "loss": 0.9177,
+      "step": 6997
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.545822510511466e-05,
+      "loss": 1.2433,
+      "step": 6998
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.545695898276188e-05,
+      "loss": 1.0579,
+      "step": 6999
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.54556926984107e-05,
+      "loss": 0.2855,
+      "step": 7000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.545442625206917e-05,
+      "loss": 0.7411,
+      "step": 7001
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.545315964374536e-05,
+      "loss": 0.873,
+      "step": 7002
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5451892873447326e-05,
+      "loss": 1.8317,
+      "step": 7003
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.545062594118313e-05,
+      "loss": 0.3324,
+      "step": 7004
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.544935884696083e-05,
+      "loss": 1.186,
+      "step": 7005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.544809159078849e-05,
+      "loss": 0.722,
+      "step": 7006
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.544682417267417e-05,
+      "loss": 1.1284,
+      "step": 7007
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5445556592625954e-05,
+      "loss": 1.2203,
+      "step": 7008
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5444288850651886e-05,
+      "loss": 0.8377,
+      "step": 7009
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.544302094676005e-05,
+      "loss": 0.3096,
+      "step": 7010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5441752880958504e-05,
+      "loss": 0.8905,
+      "step": 7011
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5440484653255324e-05,
+      "loss": 0.5375,
+      "step": 7012
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.543921626365858e-05,
+      "loss": 0.3191,
+      "step": 7013
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.543794771217634e-05,
+      "loss": 0.1588,
+      "step": 7014
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.543667899881667e-05,
+      "loss": 1.3858,
+      "step": 7015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.543541012358766e-05,
+      "loss": 0.4347,
+      "step": 7016
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.543414108649737e-05,
+      "loss": 1.739,
+      "step": 7017
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.543287188755389e-05,
+      "loss": 0.9179,
+      "step": 7018
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.543160252676528e-05,
+      "loss": 1.5859,
+      "step": 7019
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.543033300413963e-05,
+      "loss": 0.4124,
+      "step": 7020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.542906331968501e-05,
+      "loss": 0.4991,
+      "step": 7021
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.542779347340951e-05,
+      "loss": 1.5182,
+      "step": 7022
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.542652346532121e-05,
+      "loss": 0.6855,
+      "step": 7023
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5425253295428176e-05,
+      "loss": 1.4006,
+      "step": 7024
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.54239829637385e-05,
+      "loss": 0.9883,
+      "step": 7025
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5422712470260285e-05,
+      "loss": 0.4197,
+      "step": 7026
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.542144181500159e-05,
+      "loss": 0.8756,
+      "step": 7027
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.542017099797052e-05,
+      "loss": 0.8349,
+      "step": 7028
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.541890001917513e-05,
+      "loss": 0.5529,
+      "step": 7029
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.541762887862355e-05,
+      "loss": 0.8696,
+      "step": 7030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.541635757632384e-05,
+      "loss": 1.2997,
+      "step": 7031
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.54150861122841e-05,
+      "loss": 0.6513,
+      "step": 7032
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.541381448651242e-05,
+      "loss": 0.8974,
+      "step": 7033
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.54125426990169e-05,
+      "loss": 1.3835,
+      "step": 7034
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.541127074980562e-05,
+      "loss": 0.8045,
+      "step": 7035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.540999863888668e-05,
+      "loss": 0.8937,
+      "step": 7036
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5408726366268174e-05,
+      "loss": 1.7266,
+      "step": 7037
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5407453931958204e-05,
+      "loss": 1.368,
+      "step": 7038
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.540618133596486e-05,
+      "loss": 1.464,
+      "step": 7039
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.540490857829624e-05,
+      "loss": 0.4197,
+      "step": 7040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.5403635658960444e-05,
+      "loss": 1.2378,
+      "step": 7041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.540236257796558e-05,
+      "loss": 0.8301,
+      "step": 7042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.540108933531974e-05,
+      "loss": 0.5708,
+      "step": 7043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.539981593103104e-05,
+      "loss": 0.7135,
+      "step": 7044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5398542365107565e-05,
+      "loss": 0.4143,
+      "step": 7045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5397268637557435e-05,
+      "loss": 0.5538,
+      "step": 7046
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5395994748388744e-05,
+      "loss": 1.1467,
+      "step": 7047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.539472069760961e-05,
+      "loss": 0.7514,
+      "step": 7048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.539344648522812e-05,
+      "loss": 1.0036,
+      "step": 7049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.539217211125241e-05,
+      "loss": 0.3074,
+      "step": 7050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.539089757569057e-05,
+      "loss": 1.1816,
+      "step": 7051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.538962287855072e-05,
+      "loss": 1.6286,
+      "step": 7052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5388348019840965e-05,
+      "loss": 1.1077,
+      "step": 7053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5387072999569426e-05,
+      "loss": 0.535,
+      "step": 7054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5385797817744214e-05,
+      "loss": 0.3429,
+      "step": 7055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5384522474373435e-05,
+      "loss": 0.4094,
+      "step": 7056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.538324696946521e-05,
+      "loss": 1.4067,
+      "step": 7057
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.538197130302766e-05,
+      "loss": 1.4668,
+      "step": 7058
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5380695475068914e-05,
+      "loss": 1.4496,
+      "step": 7059
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.537941948559706e-05,
+      "loss": 1.489,
+      "step": 7060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5378143334620246e-05,
+      "loss": 0.7069,
+      "step": 7061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5376867022146576e-05,
+      "loss": 0.5297,
+      "step": 7062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.537559054818418e-05,
+      "loss": 1.6449,
+      "step": 7063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.537431391274118e-05,
+      "loss": 1.0363,
+      "step": 7064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.537303711582569e-05,
+      "loss": 0.5599,
+      "step": 7065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5371760157445855e-05,
+      "loss": 0.8884,
+      "step": 7066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.537048303760979e-05,
+      "loss": 0.2133,
+      "step": 7067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.536920575632562e-05,
+      "loss": 0.3093,
+      "step": 7068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.536792831360147e-05,
+      "loss": 1.1708,
+      "step": 7069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5366650709445485e-05,
+      "loss": 1.2101,
+      "step": 7070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5365372943865774e-05,
+      "loss": 0.209,
+      "step": 7071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.536409501687049e-05,
+      "loss": 1.2353,
+      "step": 7072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.536281692846775e-05,
+      "loss": 0.6951,
+      "step": 7073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5361538678665687e-05,
+      "loss": 0.3968,
+      "step": 7074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.536026026747245e-05,
+      "loss": 1.1645,
+      "step": 7075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.535898169489615e-05,
+      "loss": 0.3109,
+      "step": 7076
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5357702960944946e-05,
+      "loss": 0.3595,
+      "step": 7077
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5356424065626964e-05,
+      "loss": 1.4257,
+      "step": 7078
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.535514500895035e-05,
+      "loss": 1.5936,
+      "step": 7079
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.535386579092324e-05,
+      "loss": 0.8805,
+      "step": 7080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.535258641155377e-05,
+      "loss": 0.8273,
+      "step": 7081
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.535130687085009e-05,
+      "loss": 0.1925,
+      "step": 7082
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5350027168820325e-05,
+      "loss": 1.1507,
+      "step": 7083
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5348747305472643e-05,
+      "loss": 1.0427,
+      "step": 7084
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.534746728081516e-05,
+      "loss": 1.2519,
+      "step": 7085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5346187094856055e-05,
+      "loss": 1.0971,
+      "step": 7086
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.534490674760345e-05,
+      "loss": 0.3667,
+      "step": 7087
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5343626239065496e-05,
+      "loss": 0.3974,
+      "step": 7088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.534234556925036e-05,
+      "loss": 0.92,
+      "step": 7089
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.534106473816616e-05,
+      "loss": 1.0964,
+      "step": 7090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.533978374582108e-05,
+      "loss": 1.8586,
+      "step": 7091
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.533850259222325e-05,
+      "loss": 1.206,
+      "step": 7092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.533722127738082e-05,
+      "loss": 0.9404,
+      "step": 7093
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.533593980130197e-05,
+      "loss": 0.8921,
+      "step": 7094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.533465816399482e-05,
+      "loss": 1.0224,
+      "step": 7095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.533337636546755e-05,
+      "loss": 0.7905,
+      "step": 7096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.533209440572831e-05,
+      "loss": 0.3539,
+      "step": 7097
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.533081228478526e-05,
+      "loss": 0.2559,
+      "step": 7098
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.532953000264656e-05,
+      "loss": 0.6717,
+      "step": 7099
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5328247559320357e-05,
+      "loss": 0.4438,
+      "step": 7100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5326964954814836e-05,
+      "loss": 0.7181,
+      "step": 7101
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.532568218913813e-05,
+      "loss": 0.3152,
+      "step": 7102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.532439926229842e-05,
+      "loss": 0.6807,
+      "step": 7103
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.532311617430387e-05,
+      "loss": 0.7021,
+      "step": 7104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.532183292516265e-05,
+      "loss": 1.4047,
+      "step": 7105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.532054951488291e-05,
+      "loss": 1.0079,
+      "step": 7106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.531926594347282e-05,
+      "loss": 1.1846,
+      "step": 7107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5317982210940564e-05,
+      "loss": 0.8386,
+      "step": 7108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.531669831729429e-05,
+      "loss": 0.6712,
+      "step": 7109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.531541426254219e-05,
+      "loss": 0.4872,
+      "step": 7110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.531413004669242e-05,
+      "loss": 0.2066,
+      "step": 7111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.531284566975316e-05,
+      "loss": 0.9164,
+      "step": 7112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.531156113173257e-05,
+      "loss": 0.7059,
+      "step": 7113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5310276432638834e-05,
+      "loss": 1.1112,
+      "step": 7114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.530899157248014e-05,
+      "loss": 0.3148,
+      "step": 7115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.530770655126465e-05,
+      "loss": 0.9391,
+      "step": 7116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.530642136900053e-05,
+      "loss": 1.2556,
+      "step": 7117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5305136025695976e-05,
+      "loss": 0.4481,
+      "step": 7118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.530385052135917e-05,
+      "loss": 0.8587,
+      "step": 7119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.530256485599828e-05,
+      "loss": 1.0873,
+      "step": 7120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5301279029621495e-05,
+      "loss": 0.4169,
+      "step": 7121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5299993042236996e-05,
+      "loss": 1.5678,
+      "step": 7122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.529870689385296e-05,
+      "loss": 0.8752,
+      "step": 7123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.529742058447759e-05,
+      "loss": 0.1749,
+      "step": 7124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5296134114119035e-05,
+      "loss": 0.7879,
+      "step": 7125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.529484748278553e-05,
+      "loss": 0.5776,
+      "step": 7126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.529356069048522e-05,
+      "loss": 1.3807,
+      "step": 7127
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5292273737226325e-05,
+      "loss": 1.1882,
+      "step": 7128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.529098662301702e-05,
+      "loss": 0.5871,
+      "step": 7129
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.528969934786549e-05,
+      "loss": 0.3965,
+      "step": 7130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.528841191177994e-05,
+      "loss": 1.5386,
+      "step": 7131
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.528712431476856e-05,
+      "loss": 1.4961,
+      "step": 7132
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5285836556839545e-05,
+      "loss": 1.2395,
+      "step": 7133
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.528454863800107e-05,
+      "loss": 1.5138,
+      "step": 7134
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.528326055826135e-05,
+      "loss": 1.2934,
+      "step": 7135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.528197231762858e-05,
+      "loss": 0.4124,
+      "step": 7136
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.528068391611096e-05,
+      "loss": 1.1477,
+      "step": 7137
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.527939535371668e-05,
+      "loss": 0.3678,
+      "step": 7138
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.5278106630453946e-05,
+      "loss": 0.6193,
+      "step": 7139
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.527681774633096e-05,
+      "loss": 1.1669,
+      "step": 7140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.527552870135592e-05,
+      "loss": 0.3487,
+      "step": 7141
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.527423949553703e-05,
+      "loss": 0.8647,
+      "step": 7142
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5272950128882494e-05,
+      "loss": 1.1614,
+      "step": 7143
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5271660601400527e-05,
+      "loss": 0.5618,
+      "step": 7144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.527037091309932e-05,
+      "loss": 0.4825,
+      "step": 7145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.526908106398708e-05,
+      "loss": 0.6453,
+      "step": 7146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5267791054072024e-05,
+      "loss": 0.4295,
+      "step": 7147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5266500883362357e-05,
+      "loss": 0.9906,
+      "step": 7148
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.526521055186629e-05,
+      "loss": 2.011,
+      "step": 7149
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.526392005959205e-05,
+      "loss": 0.8436,
+      "step": 7150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.52626294065478e-05,
+      "loss": 0.4051,
+      "step": 7151
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5261338592741815e-05,
+      "loss": 0.9076,
+      "step": 7152
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.526004761818227e-05,
+      "loss": 1.0304,
+      "step": 7153
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5258756482877396e-05,
+      "loss": 0.7397,
+      "step": 7154
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.525746518683541e-05,
+      "loss": 0.6432,
+      "step": 7155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5256173730064515e-05,
+      "loss": 0.3718,
+      "step": 7156
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.525488211257294e-05,
+      "loss": 0.5059,
+      "step": 7157
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.52535903343689e-05,
+      "loss": 0.3556,
+      "step": 7158
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.525229839546062e-05,
+      "loss": 1.1792,
+      "step": 7159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.525100629585632e-05,
+      "loss": 1.1312,
+      "step": 7160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.524971403556422e-05,
+      "loss": 0.6264,
+      "step": 7161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.524842161459255e-05,
+      "loss": 1.7106,
+      "step": 7162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.524712903294953e-05,
+      "loss": 1.2104,
+      "step": 7163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.524583629064338e-05,
+      "loss": 1.1801,
+      "step": 7164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.524454338768234e-05,
+      "loss": 1.3597,
+      "step": 7165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.524325032407462e-05,
+      "loss": 0.2944,
+      "step": 7166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5241957099828464e-05,
+      "loss": 0.1444,
+      "step": 7167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.52406637149521e-05,
+      "loss": 1.3542,
+      "step": 7168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.523937016945374e-05,
+      "loss": 0.3489,
+      "step": 7169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5238076463341635e-05,
+      "loss": 0.3052,
+      "step": 7170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.523678259662402e-05,
+      "loss": 0.4693,
+      "step": 7171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5235488569309117e-05,
+      "loss": 0.2865,
+      "step": 7172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5234194381405164e-05,
+      "loss": 1.1279,
+      "step": 7173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.52329000329204e-05,
+      "loss": 1.0331,
+      "step": 7174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.523160552386305e-05,
+      "loss": 1.7445,
+      "step": 7175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5230310854241366e-05,
+      "loss": 1.4827,
+      "step": 7176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.522901602406358e-05,
+      "loss": 0.5408,
+      "step": 7177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.522772103333794e-05,
+      "loss": 0.749,
+      "step": 7178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.522642588207268e-05,
+      "loss": 0.9927,
+      "step": 7179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.522513057027605e-05,
+      "loss": 0.6844,
+      "step": 7180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.522383509795627e-05,
+      "loss": 0.4886,
+      "step": 7181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.52225394651216e-05,
+      "loss": 0.4856,
+      "step": 7182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.522124367178029e-05,
+      "loss": 0.872,
+      "step": 7183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.521994771794058e-05,
+      "loss": 0.3288,
+      "step": 7184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5218651603610714e-05,
+      "loss": 1.1558,
+      "step": 7185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.521735532879895e-05,
+      "loss": 1.1447,
+      "step": 7186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5216058893513514e-05,
+      "loss": 0.9355,
+      "step": 7187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.521476229776269e-05,
+      "loss": 1.2608,
+      "step": 7188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.52134655415547e-05,
+      "loss": 0.9046,
+      "step": 7189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.521216862489781e-05,
+      "loss": 0.3276,
+      "step": 7190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5210871547800274e-05,
+      "loss": 1.1083,
+      "step": 7191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.520957431027033e-05,
+      "loss": 2.0614,
+      "step": 7192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.520827691231626e-05,
+      "loss": 1.6775,
+      "step": 7193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5206979353946294e-05,
+      "loss": 0.2525,
+      "step": 7194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.52056816351687e-05,
+      "loss": 0.3693,
+      "step": 7195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.520438375599175e-05,
+      "loss": 1.3885,
+      "step": 7196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5203085716423685e-05,
+      "loss": 0.6099,
+      "step": 7197
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5201787516472764e-05,
+      "loss": 1.2095,
+      "step": 7198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5200489156147255e-05,
+      "loss": 0.9155,
+      "step": 7199
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5199190635455424e-05,
+      "loss": 0.24,
+      "step": 7200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5197891954405525e-05,
+      "loss": 1.333,
+      "step": 7201
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.519659311300583e-05,
+      "loss": 1.035,
+      "step": 7202
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.519529411126461e-05,
+      "loss": 1.0005,
+      "step": 7203
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.519399494919011e-05,
+      "loss": 1.5008,
+      "step": 7204
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.519269562679062e-05,
+      "loss": 1.1113,
+      "step": 7205
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.519139614407439e-05,
+      "loss": 0.5381,
+      "step": 7206
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.519009650104971e-05,
+      "loss": 1.3021,
+      "step": 7207
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.518879669772483e-05,
+      "loss": 0.3883,
+      "step": 7208
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5187496734108035e-05,
+      "loss": 0.8429,
+      "step": 7209
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.518619661020759e-05,
+      "loss": 1.4151,
+      "step": 7210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.518489632603177e-05,
+      "loss": 0.622,
+      "step": 7211
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.518359588158885e-05,
+      "loss": 0.8837,
+      "step": 7212
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.518229527688711e-05,
+      "loss": 0.8925,
+      "step": 7213
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.518099451193482e-05,
+      "loss": 1.0575,
+      "step": 7214
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.517969358674026e-05,
+      "loss": 0.4703,
+      "step": 7215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5178392501311705e-05,
+      "loss": 0.3966,
+      "step": 7216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5177091255657447e-05,
+      "loss": 1.408,
+      "step": 7217
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5175789849785755e-05,
+      "loss": 0.6846,
+      "step": 7218
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.517448828370491e-05,
+      "loss": 1.3446,
+      "step": 7219
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.51731865574232e-05,
+      "loss": 0.7279,
+      "step": 7220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.51718846709489e-05,
+      "loss": 1.0735,
+      "step": 7221
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.517058262429032e-05,
+      "loss": 0.3261,
+      "step": 7222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5169280417455715e-05,
+      "loss": 0.4093,
+      "step": 7223
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5167978050453386e-05,
+      "loss": 1.6621,
+      "step": 7224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.516667552329163e-05,
+      "loss": 1.4565,
+      "step": 7225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.516537283597871e-05,
+      "loss": 0.4581,
+      "step": 7226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5164069988522934e-05,
+      "loss": 0.943,
+      "step": 7227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.516276698093259e-05,
+      "loss": 0.7166,
+      "step": 7228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5161463813215966e-05,
+      "loss": 0.9479,
+      "step": 7229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5160160485381366e-05,
+      "loss": 0.2588,
+      "step": 7230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.515885699743707e-05,
+      "loss": 1.4632,
+      "step": 7231
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.515755334939138e-05,
+      "loss": 0.7583,
+      "step": 7232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5156249541252596e-05,
+      "loss": 1.0456,
+      "step": 7233
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.515494557302901e-05,
+      "loss": 1.2823,
+      "step": 7234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5153641444728915e-05,
+      "loss": 0.5043,
+      "step": 7235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.515233715636062e-05,
+      "loss": 1.3023,
+      "step": 7236
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.515103270793241e-05,
+      "loss": 1.6947,
+      "step": 7237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5149728099452606e-05,
+      "loss": 0.7058,
+      "step": 7238
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.514842333092949e-05,
+      "loss": 1.4827,
+      "step": 7239
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.514711840237138e-05,
+      "loss": 1.1632,
+      "step": 7240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.514581331378658e-05,
+      "loss": 1.5525,
+      "step": 7241
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.514450806518338e-05,
+      "loss": 0.6164,
+      "step": 7242
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.514320265657011e-05,
+      "loss": 1.1026,
+      "step": 7243
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5141897087955054e-05,
+      "loss": 0.577,
+      "step": 7244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.514059135934652e-05,
+      "loss": 0.7111,
+      "step": 7245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.513928547075284e-05,
+      "loss": 0.5421,
+      "step": 7246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.513797942218231e-05,
+      "loss": 1.2371,
+      "step": 7247
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5136673213643234e-05,
+      "loss": 2.5505,
+      "step": 7248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.513536684514394e-05,
+      "loss": 1.4979,
+      "step": 7249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.513406031669273e-05,
+      "loss": 0.541,
+      "step": 7250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5132753628297924e-05,
+      "loss": 1.6297,
+      "step": 7251
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.513144677996783e-05,
+      "loss": 0.3789,
+      "step": 7252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5130139771710775e-05,
+      "loss": 1.5657,
+      "step": 7253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.512883260353507e-05,
+      "loss": 1.5037,
+      "step": 7254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.512752527544903e-05,
+      "loss": 1.2615,
+      "step": 7255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.512621778746098e-05,
+      "loss": 0.7891,
+      "step": 7256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.512491013957924e-05,
+      "loss": 1.3383,
+      "step": 7257
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.512360233181213e-05,
+      "loss": 1.0505,
+      "step": 7258
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.512229436416797e-05,
+      "loss": 0.5613,
+      "step": 7259
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.512098623665508e-05,
+      "loss": 1.5283,
+      "step": 7260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.51196779492818e-05,
+      "loss": 1.059,
+      "step": 7261
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.511836950205644e-05,
+      "loss": 0.7795,
+      "step": 7262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.511706089498733e-05,
+      "loss": 0.2271,
+      "step": 7263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.511575212808281e-05,
+      "loss": 1.0856,
+      "step": 7264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.511444320135119e-05,
+      "loss": 1.4058,
+      "step": 7265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5113134114800814e-05,
+      "loss": 0.5043,
+      "step": 7266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.511182486843999e-05,
+      "loss": 0.5936,
+      "step": 7267
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.511051546227708e-05,
+      "loss": 0.5414,
+      "step": 7268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.51092058963204e-05,
+      "loss": 0.8094,
+      "step": 7269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5107896170578284e-05,
+      "loss": 1.2924,
+      "step": 7270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5106586285059066e-05,
+      "loss": 0.2085,
+      "step": 7271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.510527623977109e-05,
+      "loss": 0.5964,
+      "step": 7272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5103966034722686e-05,
+      "loss": 0.537,
+      "step": 7273
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5102655669922194e-05,
+      "loss": 0.9855,
+      "step": 7274
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.510134514537794e-05,
+      "loss": 1.0757,
+      "step": 7275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.510003446109828e-05,
+      "loss": 0.2866,
+      "step": 7276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.509872361709155e-05,
+      "loss": 0.8452,
+      "step": 7277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5097412613366086e-05,
+      "loss": 1.4019,
+      "step": 7278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5096101449930246e-05,
+      "loss": 0.42,
+      "step": 7279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5094790126792354e-05,
+      "loss": 0.7051,
+      "step": 7280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.509347864396077e-05,
+      "loss": 1.2753,
+      "step": 7281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.509216700144383e-05,
+      "loss": 1.3615,
+      "step": 7282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.509085519924989e-05,
+      "loss": 0.6427,
+      "step": 7283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.508954323738729e-05,
+      "loss": 1.2987,
+      "step": 7284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5088231115864373e-05,
+      "loss": 1.3577,
+      "step": 7285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.508691883468951e-05,
+      "loss": 1.237,
+      "step": 7286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5085606393871035e-05,
+      "loss": 0.9248,
+      "step": 7287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5084293793417305e-05,
+      "loss": 0.5969,
+      "step": 7288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.508298103333667e-05,
+      "loss": 0.9855,
+      "step": 7289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5081668113637483e-05,
+      "loss": 0.4465,
+      "step": 7290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.508035503432811e-05,
+      "loss": 0.4236,
+      "step": 7291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.507904179541689e-05,
+      "loss": 0.499,
+      "step": 7292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.507772839691219e-05,
+      "loss": 1.9008,
+      "step": 7293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.507641483882237e-05,
+      "loss": 0.4713,
+      "step": 7294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.507510112115578e-05,
+      "loss": 0.3819,
+      "step": 7295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.507378724392079e-05,
+      "loss": 1.591,
+      "step": 7296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.507247320712576e-05,
+      "loss": 0.5914,
+      "step": 7297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.507115901077904e-05,
+      "loss": 1.2186,
+      "step": 7298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.506984465488901e-05,
+      "loss": 0.4954,
+      "step": 7299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.506853013946402e-05,
+      "loss": 1.3613,
+      "step": 7300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5067215464512436e-05,
+      "loss": 0.9348,
+      "step": 7301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.506590063004263e-05,
+      "loss": 0.6012,
+      "step": 7302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.506458563606297e-05,
+      "loss": 0.3912,
+      "step": 7303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.506327048258183e-05,
+      "loss": 0.8297,
+      "step": 7304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.506195516960757e-05,
+      "loss": 0.9288,
+      "step": 7305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5060639697148556e-05,
+      "loss": 1.4175,
+      "step": 7306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.505932406521317e-05,
+      "loss": 1.7531,
+      "step": 7307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.505800827380977e-05,
+      "loss": 1.3611,
+      "step": 7308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.505669232294674e-05,
+      "loss": 1.2297,
+      "step": 7309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.505537621263246e-05,
+      "loss": 1.2011,
+      "step": 7310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.505405994287529e-05,
+      "loss": 0.2656,
+      "step": 7311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.505274351368362e-05,
+      "loss": 1.3839,
+      "step": 7312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5051426925065814e-05,
+      "loss": 1.1269,
+      "step": 7313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.505011017703026e-05,
+      "loss": 0.8002,
+      "step": 7314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.504879326958534e-05,
+      "loss": 1.4843,
+      "step": 7315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.504747620273942e-05,
+      "loss": 1.1399,
+      "step": 7316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.504615897650089e-05,
+      "loss": 0.2812,
+      "step": 7317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.504484159087813e-05,
+      "loss": 0.3666,
+      "step": 7318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.504352404587954e-05,
+      "loss": 0.5798,
+      "step": 7319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5042206341513475e-05,
+      "loss": 0.8566,
+      "step": 7320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5040888477788345e-05,
+      "loss": 0.3414,
+      "step": 7321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.503957045471252e-05,
+      "loss": 0.601,
+      "step": 7322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.50382522722944e-05,
+      "loss": 0.643,
+      "step": 7323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5036933930542364e-05,
+      "loss": 0.8416,
+      "step": 7324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5035615429464806e-05,
+      "loss": 0.3331,
+      "step": 7325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5034296769070115e-05,
+      "loss": 1.4041,
+      "step": 7326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.503297794936668e-05,
+      "loss": 0.8626,
+      "step": 7327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.50316589703629e-05,
+      "loss": 0.6573,
+      "step": 7328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.503033983206717e-05,
+      "loss": 0.5973,
+      "step": 7329
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5029020534487865e-05,
+      "loss": 1.3462,
+      "step": 7330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5027701077633407e-05,
+      "loss": 1.4108,
+      "step": 7331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.502638146151218e-05,
+      "loss": 0.574,
+      "step": 7332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.502506168613258e-05,
+      "loss": 0.4938,
+      "step": 7333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5023741751503005e-05,
+      "loss": 1.0156,
+      "step": 7334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.5022421657631855e-05,
+      "loss": 0.4943,
+      "step": 7335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.502110140452753e-05,
+      "loss": 0.9206,
+      "step": 7336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.501978099219844e-05,
+      "loss": 0.8593,
+      "step": 7337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.5018460420652986e-05,
+      "loss": 0.4293,
+      "step": 7338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.5017139689899566e-05,
+      "loss": 0.4895,
+      "step": 7339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.501581879994658e-05,
+      "loss": 1.3424,
+      "step": 7340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.501449775080245e-05,
+      "loss": 0.9853,
+      "step": 7341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.5013176542475554e-05,
+      "loss": 0.5601,
+      "step": 7342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.501185517497433e-05,
+      "loss": 1.5324,
+      "step": 7343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.5010533648307174e-05,
+      "loss": 0.8319,
+      "step": 7344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.50092119624825e-05,
+      "loss": 1.5009,
+      "step": 7345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.500789011750871e-05,
+      "loss": 1.1178,
+      "step": 7346
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.500656811339423e-05,
+      "loss": 0.7342,
+      "step": 7347
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.500524595014745e-05,
+      "loss": 0.8661,
+      "step": 7348
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.50039236277768e-05,
+      "loss": 0.1422,
+      "step": 7349
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.50026011462907e-05,
+      "loss": 1.4273,
+      "step": 7350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.5001278505697546e-05,
+      "loss": 1.4001,
+      "step": 7351
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4999955706005775e-05,
+      "loss": 0.4429,
+      "step": 7352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4998632747223795e-05,
+      "loss": 1.216,
+      "step": 7353
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.499730962936002e-05,
+      "loss": 0.676,
+      "step": 7354
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4995986352422886e-05,
+      "loss": 1.1608,
+      "step": 7355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.49946629164208e-05,
+      "loss": 0.7192,
+      "step": 7356
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.499333932136218e-05,
+      "loss": 0.3984,
+      "step": 7357
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.499201556725546e-05,
+      "loss": 0.4416,
+      "step": 7358
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4990691654109075e-05,
+      "loss": 0.344,
+      "step": 7359
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4989367581931424e-05,
+      "loss": 0.7133,
+      "step": 7360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4988043350730946e-05,
+      "loss": 0.7942,
+      "step": 7361
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.498671896051606e-05,
+      "loss": 1.5574,
+      "step": 7362
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4985394411295205e-05,
+      "loss": 1.1749,
+      "step": 7363
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.498406970307681e-05,
+      "loss": 0.5732,
+      "step": 7364
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.498274483586928e-05,
+      "loss": 1.4778,
+      "step": 7365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4981419809681085e-05,
+      "loss": 0.7788,
+      "step": 7366
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.498009462452064e-05,
+      "loss": 1.6339,
+      "step": 7367
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4978769280396364e-05,
+      "loss": 0.7673,
+      "step": 7368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.497744377731671e-05,
+      "loss": 0.8582,
+      "step": 7369
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.49761181152901e-05,
+      "loss": 1.2027,
+      "step": 7370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.497479229432498e-05,
+      "loss": 1.483,
+      "step": 7371
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.497346631442978e-05,
+      "loss": 0.4193,
+      "step": 7372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.497214017561294e-05,
+      "loss": 1.1486,
+      "step": 7373
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.49708138778829e-05,
+      "loss": 1.594,
+      "step": 7374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4969487421248106e-05,
+      "loss": 0.7568,
+      "step": 7375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4968160805716995e-05,
+      "loss": 0.6739,
+      "step": 7376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.496683403129799e-05,
+      "loss": 1.152,
+      "step": 7377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.496550709799957e-05,
+      "loss": 1.1865,
+      "step": 7378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4964180005830145e-05,
+      "loss": 0.6176,
+      "step": 7379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4962852754798185e-05,
+      "loss": 0.4857,
+      "step": 7380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.496152534491212e-05,
+      "loss": 1.6724,
+      "step": 7381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4960197776180404e-05,
+      "loss": 0.2651,
+      "step": 7382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.495887004861149e-05,
+      "loss": 0.9353,
+      "step": 7383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.495754216221382e-05,
+      "loss": 0.8185,
+      "step": 7384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.495621411699585e-05,
+      "loss": 1.3677,
+      "step": 7385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.495488591296602e-05,
+      "loss": 0.7358,
+      "step": 7386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.495355755013279e-05,
+      "loss": 1.1783,
+      "step": 7387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4952229028504615e-05,
+      "loss": 0.8779,
+      "step": 7388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.495090034808995e-05,
+      "loss": 0.1161,
+      "step": 7389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.494957150889724e-05,
+      "loss": 0.474,
+      "step": 7390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.494824251093496e-05,
+      "loss": 1.0989,
+      "step": 7391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.494691335421154e-05,
+      "loss": 1.7675,
+      "step": 7392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.494558403873546e-05,
+      "loss": 1.1771,
+      "step": 7393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.494425456451517e-05,
+      "loss": 0.5864,
+      "step": 7394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.494292493155914e-05,
+      "loss": 1.6968,
+      "step": 7395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4941595139875825e-05,
+      "loss": 0.6608,
+      "step": 7396
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4940265189473685e-05,
+      "loss": 0.428,
+      "step": 7397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.493893508036118e-05,
+      "loss": 0.8658,
+      "step": 7398
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4937604812546785e-05,
+      "loss": 0.4771,
+      "step": 7399
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.493627438603896e-05,
+      "loss": 0.608,
+      "step": 7400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.493494380084617e-05,
+      "loss": 0.6793,
+      "step": 7401
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.493361305697688e-05,
+      "loss": 1.3231,
+      "step": 7402
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.493228215443956e-05,
+      "loss": 1.0503,
+      "step": 7403
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.493095109324268e-05,
+      "loss": 0.873,
+      "step": 7404
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.492961987339472e-05,
+      "loss": 1.8314,
+      "step": 7405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.492828849490414e-05,
+      "loss": 1.1116,
+      "step": 7406
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.492695695777942e-05,
+      "loss": 1.6006,
+      "step": 7407
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.492562526202902e-05,
+      "loss": 1.3,
+      "step": 7408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.492429340766142e-05,
+      "loss": 0.4358,
+      "step": 7409
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.492296139468511e-05,
+      "loss": 0.397,
+      "step": 7410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4921629223108545e-05,
+      "loss": 0.6473,
+      "step": 7411
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.492029689294022e-05,
+      "loss": 1.1457,
+      "step": 7412
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.49189644041886e-05,
+      "loss": 0.9215,
+      "step": 7413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.491763175686218e-05,
+      "loss": 0.9728,
+      "step": 7414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.491629895096942e-05,
+      "loss": 0.9329,
+      "step": 7415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.491496598651882e-05,
+      "loss": 0.8327,
+      "step": 7416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.491363286351884e-05,
+      "loss": 1.6604,
+      "step": 7417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.491229958197799e-05,
+      "loss": 1.234,
+      "step": 7418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.491096614190475e-05,
+      "loss": 0.4967,
+      "step": 7419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.490963254330758e-05,
+      "loss": 0.8653,
+      "step": 7420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4908298786194994e-05,
+      "loss": 1.0819,
+      "step": 7421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4906964870575465e-05,
+      "loss": 1.0113,
+      "step": 7422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.49056307964575e-05,
+      "loss": 0.7656,
+      "step": 7423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.490429656384956e-05,
+      "loss": 0.4124,
+      "step": 7424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4902962172760164e-05,
+      "loss": 1.5976,
+      "step": 7425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4901627623197786e-05,
+      "loss": 1.0677,
+      "step": 7426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.490029291517092e-05,
+      "loss": 0.593,
+      "step": 7427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.489895804868805e-05,
+      "loss": 1.2255,
+      "step": 7428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.48976230237577e-05,
+      "loss": 0.6784,
+      "step": 7429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.489628784038834e-05,
+      "loss": 0.4693,
+      "step": 7430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4894952498588486e-05,
+      "loss": 0.9099,
+      "step": 7431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.489361699836662e-05,
+      "loss": 1.0175,
+      "step": 7432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4892281339731245e-05,
+      "loss": 0.8882,
+      "step": 7433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.4890945522690855e-05,
+      "loss": 1.3266,
+      "step": 7434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.488960954725396e-05,
+      "loss": 0.7025,
+      "step": 7435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4888273413429064e-05,
+      "loss": 1.7204,
+      "step": 7436
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.488693712122466e-05,
+      "loss": 0.8148,
+      "step": 7437
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.488560067064926e-05,
+      "loss": 0.3027,
+      "step": 7438
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.488426406171136e-05,
+      "loss": 0.6446,
+      "step": 7439
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.488292729441947e-05,
+      "loss": 1.2667,
+      "step": 7440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4881590368782085e-05,
+      "loss": 0.7546,
+      "step": 7441
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.488025328480775e-05,
+      "loss": 0.2973,
+      "step": 7442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4878916042504924e-05,
+      "loss": 1.2887,
+      "step": 7443
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4877578641882155e-05,
+      "loss": 0.4528,
+      "step": 7444
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.487624108294793e-05,
+      "loss": 0.3147,
+      "step": 7445
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.487490336571078e-05,
+      "loss": 0.4805,
+      "step": 7446
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.48735654901792e-05,
+      "loss": 0.4988,
+      "step": 7447
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4872227456361714e-05,
+      "loss": 1.193,
+      "step": 7448
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.487088926426683e-05,
+      "loss": 1.1394,
+      "step": 7449
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.486955091390308e-05,
+      "loss": 0.6152,
+      "step": 7450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.486821240527896e-05,
+      "loss": 0.7913,
+      "step": 7451
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4866873738403e-05,
+      "loss": 0.7029,
+      "step": 7452
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.486553491328371e-05,
+      "loss": 1.1572,
+      "step": 7453
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4864195929929624e-05,
+      "loss": 1.1858,
+      "step": 7454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.486285678834925e-05,
+      "loss": 1.851,
+      "step": 7455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4861517488551104e-05,
+      "loss": 1.3557,
+      "step": 7456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.486017803054373e-05,
+      "loss": 0.8899,
+      "step": 7457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.485883841433564e-05,
+      "loss": 0.5555,
+      "step": 7458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.485749863993536e-05,
+      "loss": 1.1227,
+      "step": 7459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4856158707351404e-05,
+      "loss": 0.9609,
+      "step": 7460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.485481861659232e-05,
+      "loss": 0.8208,
+      "step": 7461
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.485347836766663e-05,
+      "loss": 0.3911,
+      "step": 7462
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.485213796058285e-05,
+      "loss": 1.361,
+      "step": 7463
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4850797395349516e-05,
+      "loss": 1.7423,
+      "step": 7464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.484945667197517e-05,
+      "loss": 1.2149,
+      "step": 7465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4848115790468326e-05,
+      "loss": 1.2689,
+      "step": 7466
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.484677475083752e-05,
+      "loss": 1.1077,
+      "step": 7467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.484543355309131e-05,
+      "loss": 1.173,
+      "step": 7468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.48440921972382e-05,
+      "loss": 0.3627,
+      "step": 7469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.484275068328674e-05,
+      "loss": 1.0101,
+      "step": 7470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.484140901124548e-05,
+      "loss": 1.2901,
+      "step": 7471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.484006718112292e-05,
+      "loss": 0.7954,
+      "step": 7472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.483872519292763e-05,
+      "loss": 1.0681,
+      "step": 7473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4837383046668146e-05,
+      "loss": 1.2877,
+      "step": 7474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.483604074235301e-05,
+      "loss": 0.4526,
+      "step": 7475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4834698279990745e-05,
+      "loss": 1.334,
+      "step": 7476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4833355659589915e-05,
+      "loss": 0.3523,
+      "step": 7477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.483201288115906e-05,
+      "loss": 0.4896,
+      "step": 7478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.483066994470672e-05,
+      "loss": 1.2445,
+      "step": 7479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.482932685024144e-05,
+      "loss": 0.4842,
+      "step": 7480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.482798359777177e-05,
+      "loss": 1.3156,
+      "step": 7481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4826640187306256e-05,
+      "loss": 0.709,
+      "step": 7482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.482529661885345e-05,
+      "loss": 0.3022,
+      "step": 7483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.482395289242191e-05,
+      "loss": 0.1986,
+      "step": 7484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.482260900802016e-05,
+      "loss": 1.5512,
+      "step": 7485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.482126496565678e-05,
+      "loss": 0.2509,
+      "step": 7486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4819920765340314e-05,
+      "loss": 0.5115,
+      "step": 7487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.48185764070793e-05,
+      "loss": 1.7899,
+      "step": 7488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4817231890882324e-05,
+      "loss": 0.57,
+      "step": 7489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4815887216757916e-05,
+      "loss": 0.7195,
+      "step": 7490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.481454238471464e-05,
+      "loss": 0.407,
+      "step": 7491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4813197394761064e-05,
+      "loss": 1.5386,
+      "step": 7492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.481185224690573e-05,
+      "loss": 1.1244,
+      "step": 7493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4810506941157216e-05,
+      "loss": 1.5645,
+      "step": 7494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.480916147752408e-05,
+      "loss": 0.7817,
+      "step": 7495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.480781585601486e-05,
+      "loss": 1.0056,
+      "step": 7496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.480647007663815e-05,
+      "loss": 0.5909,
+      "step": 7497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.48051241394025e-05,
+      "loss": 1.1215,
+      "step": 7498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.480377804431648e-05,
+      "loss": 1.2559,
+      "step": 7499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4802431791388636e-05,
+      "loss": 0.4832,
+      "step": 7500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.480108538062756e-05,
+      "loss": 0.8764,
+      "step": 7501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.479973881204182e-05,
+      "loss": 0.5197,
+      "step": 7502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.479839208563997e-05,
+      "loss": 1.0768,
+      "step": 7503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.479704520143058e-05,
+      "loss": 2.0009,
+      "step": 7504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.479569815942224e-05,
+      "loss": 1.5703,
+      "step": 7505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4794350959623496e-05,
+      "loss": 1.0272,
+      "step": 7506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4793003602042955e-05,
+      "loss": 1.6256,
+      "step": 7507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.479165608668915e-05,
+      "loss": 0.6903,
+      "step": 7508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4790308413570685e-05,
+      "loss": 1.3513,
+      "step": 7509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4788960582696134e-05,
+      "loss": 0.3763,
+      "step": 7510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.478761259407406e-05,
+      "loss": 0.2484,
+      "step": 7511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.478626444771305e-05,
+      "loss": 1.1988,
+      "step": 7512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.478491614362169e-05,
+      "loss": 0.6532,
+      "step": 7513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4783567681808544e-05,
+      "loss": 0.199,
+      "step": 7514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4782219062282194e-05,
+      "loss": 1.0466,
+      "step": 7515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.478087028505124e-05,
+      "loss": 1.5196,
+      "step": 7516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.477952135012426e-05,
+      "loss": 0.5463,
+      "step": 7517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.477817225750983e-05,
+      "loss": 0.4053,
+      "step": 7518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.477682300721653e-05,
+      "loss": 1.2824,
+      "step": 7519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.477547359925296e-05,
+      "loss": 0.2343,
+      "step": 7520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.477412403362769e-05,
+      "loss": 0.9973,
+      "step": 7521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4772774310349335e-05,
+      "loss": 0.9631,
+      "step": 7522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.477142442942646e-05,
+      "loss": 0.9548,
+      "step": 7523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.477007439086767e-05,
+      "loss": 0.6665,
+      "step": 7524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.476872419468154e-05,
+      "loss": 0.3607,
+      "step": 7525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.476737384087668e-05,
+      "loss": 1.0008,
+      "step": 7526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.476602332946167e-05,
+      "loss": 0.5914,
+      "step": 7527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.476467266044511e-05,
+      "loss": 1.2999,
+      "step": 7528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.476332183383559e-05,
+      "loss": 1.4162,
+      "step": 7529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.476197084964172e-05,
+      "loss": 1.7668,
+      "step": 7530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4760619707872086e-05,
+      "loss": 0.6571,
+      "step": 7531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.4759268408535275e-05,
+      "loss": 0.5498,
+      "step": 7532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.475791695163991e-05,
+      "loss": 1.0757,
+      "step": 7533
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.475656533719458e-05,
+      "loss": 1.1372,
+      "step": 7534
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4755213565207884e-05,
+      "loss": 1.1364,
+      "step": 7535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4753861635688426e-05,
+      "loss": 0.2849,
+      "step": 7536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.47525095486448e-05,
+      "loss": 0.9046,
+      "step": 7537
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4751157304085635e-05,
+      "loss": 0.5815,
+      "step": 7538
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4749804902019514e-05,
+      "loss": 0.9711,
+      "step": 7539
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4748452342455056e-05,
+      "loss": 0.5323,
+      "step": 7540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4747099625400854e-05,
+      "loss": 1.5477,
+      "step": 7541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.474574675086553e-05,
+      "loss": 1.4479,
+      "step": 7542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4744393718857684e-05,
+      "loss": 1.433,
+      "step": 7543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4743040529385926e-05,
+      "loss": 0.7479,
+      "step": 7544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4741687182458875e-05,
+      "loss": 1.8094,
+      "step": 7545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.474033367808514e-05,
+      "loss": 0.7825,
+      "step": 7546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.473898001627334e-05,
+      "loss": 0.55,
+      "step": 7547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.473762619703207e-05,
+      "loss": 0.2618,
+      "step": 7548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.473627222036996e-05,
+      "loss": 1.4449,
+      "step": 7549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.473491808629562e-05,
+      "loss": 1.1887,
+      "step": 7550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.473356379481769e-05,
+      "loss": 1.2635,
+      "step": 7551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.473220934594475e-05,
+      "loss": 0.1796,
+      "step": 7552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.473085473968544e-05,
+      "loss": 0.3097,
+      "step": 7553
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.472949997604839e-05,
+      "loss": 0.3389,
+      "step": 7554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.47281450550422e-05,
+      "loss": 1.3103,
+      "step": 7555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4726789976675496e-05,
+      "loss": 1.4228,
+      "step": 7556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.472543474095692e-05,
+      "loss": 1.0454,
+      "step": 7557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4724079347895076e-05,
+      "loss": 1.4184,
+      "step": 7558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.47227237974986e-05,
+      "loss": 0.7862,
+      "step": 7559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4721368089776114e-05,
+      "loss": 0.5774,
+      "step": 7560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.472001222473625e-05,
+      "loss": 0.7694,
+      "step": 7561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.471865620238763e-05,
+      "loss": 1.6764,
+      "step": 7562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.471730002273888e-05,
+      "loss": 0.2778,
+      "step": 7563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.471594368579864e-05,
+      "loss": 1.4638,
+      "step": 7564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.471458719157553e-05,
+      "loss": 1.0877,
+      "step": 7565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.471323054007821e-05,
+      "loss": 0.4922,
+      "step": 7566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.471187373131527e-05,
+      "loss": 1.2725,
+      "step": 7567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4710516765295374e-05,
+      "loss": 1.9185,
+      "step": 7568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.470915964202715e-05,
+      "loss": 0.5981,
+      "step": 7569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.470780236151923e-05,
+      "loss": 0.2997,
+      "step": 7570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4706444923780254e-05,
+      "loss": 0.7375,
+      "step": 7571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.470508732881887e-05,
+      "loss": 0.291,
+      "step": 7572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.47037295766437e-05,
+      "loss": 0.8458,
+      "step": 7573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4702371667263405e-05,
+      "loss": 0.5251,
+      "step": 7574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.47010136006866e-05,
+      "loss": 0.7405,
+      "step": 7575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.469965537692195e-05,
+      "loss": 1.2186,
+      "step": 7576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.469829699597809e-05,
+      "loss": 1.3863,
+      "step": 7577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.469693845786366e-05,
+      "loss": 0.4555,
+      "step": 7578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.469557976258731e-05,
+      "loss": 0.3635,
+      "step": 7579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.469422091015768e-05,
+      "loss": 0.4563,
+      "step": 7580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.469286190058343e-05,
+      "loss": 0.9036,
+      "step": 7581
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.469150273387319e-05,
+      "loss": 1.143,
+      "step": 7582
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.469014341003563e-05,
+      "loss": 1.4128,
+      "step": 7583
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.468878392907938e-05,
+      "loss": 1.0662,
+      "step": 7584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.468742429101312e-05,
+      "loss": 0.2236,
+      "step": 7585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4686064495845464e-05,
+      "loss": 0.5563,
+      "step": 7586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4684704543585094e-05,
+      "loss": 1.5614,
+      "step": 7587
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.468334443424065e-05,
+      "loss": 1.237,
+      "step": 7588
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4681984167820793e-05,
+      "loss": 0.6686,
+      "step": 7589
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.468062374433418e-05,
+      "loss": 0.2802,
+      "step": 7590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4679263163789476e-05,
+      "loss": 0.8445,
+      "step": 7591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.467790242619531e-05,
+      "loss": 0.6786,
+      "step": 7592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4676541531560374e-05,
+      "loss": 1.199,
+      "step": 7593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.467518047989331e-05,
+      "loss": 1.3418,
+      "step": 7594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.467381927120279e-05,
+      "loss": 0.851,
+      "step": 7595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.467245790549747e-05,
+      "loss": 0.777,
+      "step": 7596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.467109638278601e-05,
+      "loss": 0.8653,
+      "step": 7597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.466973470307708e-05,
+      "loss": 1.3495,
+      "step": 7598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.466837286637933e-05,
+      "loss": 0.4959,
+      "step": 7599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.466701087270146e-05,
+      "loss": 0.6112,
+      "step": 7600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.466564872205211e-05,
+      "loss": 0.5123,
+      "step": 7601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.466428641443995e-05,
+      "loss": 0.4205,
+      "step": 7602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4662923949873654e-05,
+      "loss": 0.4381,
+      "step": 7603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4661561328361895e-05,
+      "loss": 1.2781,
+      "step": 7604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.466019854991334e-05,
+      "loss": 0.9419,
+      "step": 7605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.465883561453667e-05,
+      "loss": 0.387,
+      "step": 7606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.465747252224055e-05,
+      "loss": 0.943,
+      "step": 7607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.465610927303364e-05,
+      "loss": 1.0142,
+      "step": 7608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.465474586692464e-05,
+      "loss": 0.9513,
+      "step": 7609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.465338230392222e-05,
+      "loss": 0.4581,
+      "step": 7610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4652018584035045e-05,
+      "loss": 1.0182,
+      "step": 7611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4650654707271816e-05,
+      "loss": 0.9925,
+      "step": 7612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.464929067364118e-05,
+      "loss": 0.2227,
+      "step": 7613
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.464792648315185e-05,
+      "loss": 0.8632,
+      "step": 7614
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.464656213581249e-05,
+      "loss": 0.4488,
+      "step": 7615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.464519763163178e-05,
+      "loss": 1.0253,
+      "step": 7616
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.464383297061841e-05,
+      "loss": 0.6912,
+      "step": 7617
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.464246815278105e-05,
+      "loss": 0.271,
+      "step": 7618
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4641103178128416e-05,
+      "loss": 0.8689,
+      "step": 7619
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4639738046669166e-05,
+      "loss": 0.501,
+      "step": 7620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4638372758412e-05,
+      "loss": 0.9996,
+      "step": 7621
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4637007313365603e-05,
+      "loss": 0.7385,
+      "step": 7622
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4635641711538665e-05,
+      "loss": 0.4902,
+      "step": 7623
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4634275952939867e-05,
+      "loss": 0.5677,
+      "step": 7624
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4632910037577916e-05,
+      "loss": 0.8376,
+      "step": 7625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.463154396546149e-05,
+      "loss": 0.5541,
+      "step": 7626
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.463017773659929e-05,
+      "loss": 1.4834,
+      "step": 7627
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4628811351000016e-05,
+      "loss": 0.7777,
+      "step": 7628
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.462744480867235e-05,
+      "loss": 0.813,
+      "step": 7629
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4626078109625e-05,
+      "loss": 0.7133,
+      "step": 7630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.462471125386665e-05,
+      "loss": 0.5751,
+      "step": 7631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.4623344241406e-05,
+      "loss": 1.4216,
+      "step": 7632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.462197707225176e-05,
+      "loss": 0.7613,
+      "step": 7633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.462060974641263e-05,
+      "loss": 0.4706,
+      "step": 7634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4619242263897304e-05,
+      "loss": 1.2121,
+      "step": 7635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.461787462471448e-05,
+      "loss": 0.6323,
+      "step": 7636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.461650682887287e-05,
+      "loss": 1.0027,
+      "step": 7637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.461513887638118e-05,
+      "loss": 0.3081,
+      "step": 7638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.46137707672481e-05,
+      "loss": 1.6936,
+      "step": 7639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.461240250148235e-05,
+      "loss": 0.9454,
+      "step": 7640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4611034079092646e-05,
+      "loss": 0.7598,
+      "step": 7641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.460966550008767e-05,
+      "loss": 1.7008,
+      "step": 7642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.460829676447615e-05,
+      "loss": 1.2902,
+      "step": 7643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4606927872266785e-05,
+      "loss": 0.9428,
+      "step": 7644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.46055588234683e-05,
+      "loss": 1.5316,
+      "step": 7645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.460418961808939e-05,
+      "loss": 0.2228,
+      "step": 7646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.460282025613878e-05,
+      "loss": 0.8376,
+      "step": 7647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.460145073762519e-05,
+      "loss": 1.0419,
+      "step": 7648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4600081062557314e-05,
+      "loss": 1.1961,
+      "step": 7649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.459871123094389e-05,
+      "loss": 0.7232,
+      "step": 7650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.459734124279362e-05,
+      "loss": 0.6538,
+      "step": 7651
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.459597109811523e-05,
+      "loss": 0.858,
+      "step": 7652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.459460079691744e-05,
+      "loss": 0.6788,
+      "step": 7653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.459323033920897e-05,
+      "loss": 0.9315,
+      "step": 7654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4591859724998515e-05,
+      "loss": 0.34,
+      "step": 7655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.459048895429484e-05,
+      "loss": 1.7858,
+      "step": 7656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4589118027106645e-05,
+      "loss": 0.9331,
+      "step": 7657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4587746943442656e-05,
+      "loss": 1.2352,
+      "step": 7658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4586375703311596e-05,
+      "loss": 1.3237,
+      "step": 7659
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4585004306722195e-05,
+      "loss": 0.4273,
+      "step": 7660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4583632753683185e-05,
+      "loss": 1.2241,
+      "step": 7661
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.458226104420329e-05,
+      "loss": 1.274,
+      "step": 7662
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.458088917829123e-05,
+      "loss": 0.8864,
+      "step": 7663
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.457951715595575e-05,
+      "loss": 0.5766,
+      "step": 7664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.457814497720556e-05,
+      "loss": 1.2423,
+      "step": 7665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4576772642049415e-05,
+      "loss": 0.6063,
+      "step": 7666
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4575400150496036e-05,
+      "loss": 1.7397,
+      "step": 7667
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.457402750255416e-05,
+      "loss": 0.4742,
+      "step": 7668
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.457265469823252e-05,
+      "loss": 0.8437,
+      "step": 7669
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4571281737539856e-05,
+      "loss": 0.7008,
+      "step": 7670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.45699086204849e-05,
+      "loss": 0.5287,
+      "step": 7671
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.45685353470764e-05,
+      "loss": 0.5809,
+      "step": 7672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.456716191732308e-05,
+      "loss": 1.7316,
+      "step": 7673
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.456578833123368e-05,
+      "loss": 0.5465,
+      "step": 7674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.456441458881696e-05,
+      "loss": 0.8573,
+      "step": 7675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.456304069008165e-05,
+      "loss": 1.332,
+      "step": 7676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.456166663503649e-05,
+      "loss": 1.0754,
+      "step": 7677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.456029242369022e-05,
+      "loss": 0.5583,
+      "step": 7678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4558918056051605e-05,
+      "loss": 0.8916,
+      "step": 7679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.455754353212937e-05,
+      "loss": 0.7886,
+      "step": 7680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.455616885193228e-05,
+      "loss": 1.6418,
+      "step": 7681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.455479401546906e-05,
+      "loss": 0.3761,
+      "step": 7682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.455341902274848e-05,
+      "loss": 0.7092,
+      "step": 7683
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.455204387377928e-05,
+      "loss": 0.4483,
+      "step": 7684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4550668568570214e-05,
+      "loss": 1.1842,
+      "step": 7685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.454929310713003e-05,
+      "loss": 0.7693,
+      "step": 7686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.454791748946749e-05,
+      "loss": 0.9506,
+      "step": 7687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.454654171559133e-05,
+      "loss": 1.4189,
+      "step": 7688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.454516578551033e-05,
+      "loss": 1.3552,
+      "step": 7689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.454378969923322e-05,
+      "loss": 0.537,
+      "step": 7690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4542413456768775e-05,
+      "loss": 0.7817,
+      "step": 7691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.454103705812575e-05,
+      "loss": 0.3059,
+      "step": 7692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.45396605033129e-05,
+      "loss": 0.3333,
+      "step": 7693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.453828379233898e-05,
+      "loss": 1.0496,
+      "step": 7694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.453690692521276e-05,
+      "loss": 2.8696,
+      "step": 7695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4535529901943e-05,
+      "loss": 1.4245,
+      "step": 7696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.453415272253846e-05,
+      "loss": 1.1942,
+      "step": 7697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.453277538700791e-05,
+      "loss": 0.7269,
+      "step": 7698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.453139789536011e-05,
+      "loss": 0.8486,
+      "step": 7699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.453002024760383e-05,
+      "loss": 0.3999,
+      "step": 7700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4528642443747815e-05,
+      "loss": 1.3425,
+      "step": 7701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.452726448380086e-05,
+      "loss": 1.3893,
+      "step": 7702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.452588636777173e-05,
+      "loss": 0.9966,
+      "step": 7703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4524508095669185e-05,
+      "loss": 1.1207,
+      "step": 7704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4523129667502e-05,
+      "loss": 1.0178,
+      "step": 7705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.452175108327894e-05,
+      "loss": 0.6767,
+      "step": 7706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4520372343008795e-05,
+      "loss": 0.5579,
+      "step": 7707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4518993446700326e-05,
+      "loss": 1.0496,
+      "step": 7708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.451761439436231e-05,
+      "loss": 0.2836,
+      "step": 7709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.451623518600352e-05,
+      "loss": 0.3425,
+      "step": 7710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.451485582163274e-05,
+      "loss": 0.4493,
+      "step": 7711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.451347630125874e-05,
+      "loss": 0.4476,
+      "step": 7712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.451209662489031e-05,
+      "loss": 1.0658,
+      "step": 7713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.451071679253621e-05,
+      "loss": 0.9154,
+      "step": 7714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.450933680420524e-05,
+      "loss": 0.4304,
+      "step": 7715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.450795665990617e-05,
+      "loss": 1.011,
+      "step": 7716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4506576359647796e-05,
+      "loss": 0.3095,
+      "step": 7717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4505195903438885e-05,
+      "loss": 0.5686,
+      "step": 7718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4503815291288226e-05,
+      "loss": 0.5335,
+      "step": 7719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4502434523204616e-05,
+      "loss": 1.5615,
+      "step": 7720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.450105359919684e-05,
+      "loss": 0.852,
+      "step": 7721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.449967251927366e-05,
+      "loss": 0.9216,
+      "step": 7722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.44982912834439e-05,
+      "loss": 1.2663,
+      "step": 7723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.449690989171633e-05,
+      "loss": 0.904,
+      "step": 7724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.4495528344099745e-05,
+      "loss": 0.694,
+      "step": 7725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.449414664060294e-05,
+      "loss": 1.317,
+      "step": 7726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.44927647812347e-05,
+      "loss": 0.9592,
+      "step": 7727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.449138276600381e-05,
+      "loss": 1.3575,
+      "step": 7728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.449000059491909e-05,
+      "loss": 0.4803,
+      "step": 7729
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.448861826798932e-05,
+      "loss": 2.281,
+      "step": 7730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4487235785223305e-05,
+      "loss": 1.3058,
+      "step": 7731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.448585314662983e-05,
+      "loss": 1.4282,
+      "step": 7732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.448447035221771e-05,
+      "loss": 0.437,
+      "step": 7733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.448308740199572e-05,
+      "loss": 0.9242,
+      "step": 7734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.448170429597268e-05,
+      "loss": 0.4241,
+      "step": 7735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.448032103415739e-05,
+      "loss": 0.7811,
+      "step": 7736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4478937616558656e-05,
+      "loss": 0.3816,
+      "step": 7737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4477554043185266e-05,
+      "loss": 0.6351,
+      "step": 7738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4476170314046034e-05,
+      "loss": 0.1885,
+      "step": 7739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4474786429149775e-05,
+      "loss": 1.3148,
+      "step": 7740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.447340238850528e-05,
+      "loss": 1.0773,
+      "step": 7741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4472018192121356e-05,
+      "loss": 0.4786,
+      "step": 7742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4470633840006826e-05,
+      "loss": 0.3997,
+      "step": 7743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4469249332170487e-05,
+      "loss": 0.4643,
+      "step": 7744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.446786466862116e-05,
+      "loss": 1.4358,
+      "step": 7745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.446647984936764e-05,
+      "loss": 0.332,
+      "step": 7746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.446509487441876e-05,
+      "loss": 1.4571,
+      "step": 7747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.446370974378332e-05,
+      "loss": 0.3488,
+      "step": 7748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.446232445747013e-05,
+      "loss": 0.6997,
+      "step": 7749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4460939015488024e-05,
+      "loss": 0.6536,
+      "step": 7750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4459553417845806e-05,
+      "loss": 0.5671,
+      "step": 7751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.445816766455229e-05,
+      "loss": 1.0696,
+      "step": 7752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.44567817556163e-05,
+      "loss": 1.0669,
+      "step": 7753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.445539569104665e-05,
+      "loss": 0.7679,
+      "step": 7754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.445400947085218e-05,
+      "loss": 1.1727,
+      "step": 7755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.445262309504169e-05,
+      "loss": 0.6453,
+      "step": 7756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4451236563624e-05,
+      "loss": 0.9781,
+      "step": 7757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.444984987660795e-05,
+      "loss": 0.382,
+      "step": 7758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.444846303400236e-05,
+      "loss": 1.2814,
+      "step": 7759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.444707603581605e-05,
+      "loss": 1.7151,
+      "step": 7760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.444568888205785e-05,
+      "loss": 1.485,
+      "step": 7761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.444430157273657e-05,
+      "loss": 0.5288,
+      "step": 7762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.444291410786107e-05,
+      "loss": 0.8388,
+      "step": 7763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.444152648744016e-05,
+      "loss": 0.9124,
+      "step": 7764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4440138711482676e-05,
+      "loss": 1.5737,
+      "step": 7765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.443875077999744e-05,
+      "loss": 1.1806,
+      "step": 7766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.443736269299329e-05,
+      "loss": 0.4046,
+      "step": 7767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4435974450479065e-05,
+      "loss": 0.1773,
+      "step": 7768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.443458605246359e-05,
+      "loss": 0.5307,
+      "step": 7769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.443319749895571e-05,
+      "loss": 0.968,
+      "step": 7770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4431808789964256e-05,
+      "loss": 0.5755,
+      "step": 7771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.443041992549806e-05,
+      "loss": 0.184,
+      "step": 7772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4429030905565965e-05,
+      "loss": 1.004,
+      "step": 7773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.442764173017682e-05,
+      "loss": 1.0085,
+      "step": 7774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.442625239933945e-05,
+      "loss": 1.4446,
+      "step": 7775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.44248629130627e-05,
+      "loss": 1.8081,
+      "step": 7776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.442347327135542e-05,
+      "loss": 0.7652,
+      "step": 7777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.442208347422644e-05,
+      "loss": 0.5422,
+      "step": 7778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.442069352168462e-05,
+      "loss": 0.5973,
+      "step": 7779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.441930341373879e-05,
+      "loss": 0.6014,
+      "step": 7780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.44179131503978e-05,
+      "loss": 1.1201,
+      "step": 7781
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4416522731670506e-05,
+      "loss": 0.8449,
+      "step": 7782
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.441513215756575e-05,
+      "loss": 1.0216,
+      "step": 7783
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.441374142809238e-05,
+      "loss": 1.3502,
+      "step": 7784
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.441235054325924e-05,
+      "loss": 0.799,
+      "step": 7785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4410959503075194e-05,
+      "loss": 0.6983,
+      "step": 7786
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.440956830754909e-05,
+      "loss": 1.1353,
+      "step": 7787
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.440817695668977e-05,
+      "loss": 0.868,
+      "step": 7788
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4406785450506116e-05,
+      "loss": 1.3584,
+      "step": 7789
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.440539378900695e-05,
+      "loss": 1.3392,
+      "step": 7790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.440400197220114e-05,
+      "loss": 0.4588,
+      "step": 7791
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.440261000009755e-05,
+      "loss": 1.3119,
+      "step": 7792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4401217872705035e-05,
+      "loss": 0.3558,
+      "step": 7793
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4399825590032454e-05,
+      "loss": 0.8545,
+      "step": 7794
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4398433152088656e-05,
+      "loss": 1.1835,
+      "step": 7795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.439704055888252e-05,
+      "loss": 1.311,
+      "step": 7796
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.439564781042289e-05,
+      "loss": 0.4757,
+      "step": 7797
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4394254906718646e-05,
+      "loss": 0.9875,
+      "step": 7798
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.439286184777864e-05,
+      "loss": 0.4688,
+      "step": 7799
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.439146863361174e-05,
+      "loss": 0.4591,
+      "step": 7800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4390075264226816e-05,
+      "loss": 0.2062,
+      "step": 7801
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.438868173963273e-05,
+      "loss": 0.5334,
+      "step": 7802
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.438728805983836e-05,
+      "loss": 0.372,
+      "step": 7803
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4385894224852554e-05,
+      "loss": 1.5262,
+      "step": 7804
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4384500234684194e-05,
+      "loss": 0.8283,
+      "step": 7805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.438310608934215e-05,
+      "loss": 1.0415,
+      "step": 7806
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.43817117888353e-05,
+      "loss": 0.6792,
+      "step": 7807
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4380317333172514e-05,
+      "loss": 0.8211,
+      "step": 7808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4378922722362656e-05,
+      "loss": 0.834,
+      "step": 7809
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4377527956414614e-05,
+      "loss": 1.2949,
+      "step": 7810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.437613303533726e-05,
+      "loss": 1.602,
+      "step": 7811
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.437473795913946e-05,
+      "loss": 0.5808,
+      "step": 7812
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.437334272783011e-05,
+      "loss": 0.4494,
+      "step": 7813
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4371947341418065e-05,
+      "loss": 0.6189,
+      "step": 7814
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.437055179991223e-05,
+      "loss": 1.2179,
+      "step": 7815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.436915610332147e-05,
+      "loss": 1.4154,
+      "step": 7816
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.436776025165467e-05,
+      "loss": 0.5402,
+      "step": 7817
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.436636424492072e-05,
+      "loss": 0.3707,
+      "step": 7818
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4364968083128496e-05,
+      "loss": 0.2659,
+      "step": 7819
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.436357176628688e-05,
+      "loss": 1.3988,
+      "step": 7820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4362175294404764e-05,
+      "loss": 0.7681,
+      "step": 7821
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.436077866749103e-05,
+      "loss": 0.6287,
+      "step": 7822
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.435938188555457e-05,
+      "loss": 0.498,
+      "step": 7823
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4357984948604266e-05,
+      "loss": 0.1572,
+      "step": 7824
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.435658785664902e-05,
+      "loss": 1.0118,
+      "step": 7825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.435519060969771e-05,
+      "loss": 1.5963,
+      "step": 7826
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4353793207759234e-05,
+      "loss": 1.8438,
+      "step": 7827
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4352395650842476e-05,
+      "loss": 0.5332,
+      "step": 7828
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.435099793895635e-05,
+      "loss": 1.4772,
+      "step": 7829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.434960007210972e-05,
+      "loss": 0.7505,
+      "step": 7830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.434820205031151e-05,
+      "loss": 1.0182,
+      "step": 7831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.43468038735706e-05,
+      "loss": 0.589,
+      "step": 7832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.434540554189589e-05,
+      "loss": 0.3422,
+      "step": 7833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.434400705529629e-05,
+      "loss": 1.4574,
+      "step": 7834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4342608413780676e-05,
+      "loss": 0.8071,
+      "step": 7835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.434120961735797e-05,
+      "loss": 0.471,
+      "step": 7836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.433981066603707e-05,
+      "loss": 1.3124,
+      "step": 7837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.433841155982687e-05,
+      "loss": 1.1233,
+      "step": 7838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4337012298736274e-05,
+      "loss": 0.3206,
+      "step": 7839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4335612882774195e-05,
+      "loss": 0.8859,
+      "step": 7840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.433421331194953e-05,
+      "loss": 1.0777,
+      "step": 7841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4332813586271196e-05,
+      "loss": 0.9512,
+      "step": 7842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.433141370574809e-05,
+      "loss": 0.7367,
+      "step": 7843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.433001367038912e-05,
+      "loss": 0.1866,
+      "step": 7844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.43286134802032e-05,
+      "loss": 1.6369,
+      "step": 7845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4327213135199245e-05,
+      "loss": 0.7235,
+      "step": 7846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4325812635386146e-05,
+      "loss": 0.2457,
+      "step": 7847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.432441198077284e-05,
+      "loss": 0.2886,
+      "step": 7848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.432301117136822e-05,
+      "loss": 0.9059,
+      "step": 7849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.432161020718122e-05,
+      "loss": 1.1998,
+      "step": 7850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4320209088220745e-05,
+      "loss": 0.4752,
+      "step": 7851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.431880781449571e-05,
+      "loss": 0.3286,
+      "step": 7852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.431740638601503e-05,
+      "loss": 0.8782,
+      "step": 7853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.431600480278763e-05,
+      "loss": 1.5158,
+      "step": 7854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.431460306482242e-05,
+      "loss": 0.4376,
+      "step": 7855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.431320117212833e-05,
+      "loss": 0.5405,
+      "step": 7856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.431179912471428e-05,
+      "loss": 0.513,
+      "step": 7857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4310396922589184e-05,
+      "loss": 1.1754,
+      "step": 7858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.430899456576198e-05,
+      "loss": 1.1858,
+      "step": 7859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4307592054241564e-05,
+      "loss": 0.361,
+      "step": 7860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4306189388036894e-05,
+      "loss": 1.354,
+      "step": 7861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4304786567156874e-05,
+      "loss": 1.2638,
+      "step": 7862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4303383591610446e-05,
+      "loss": 1.1521,
+      "step": 7863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.430198046140652e-05,
+      "loss": 0.8269,
+      "step": 7864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.430057717655404e-05,
+      "loss": 0.9327,
+      "step": 7865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.429917373706193e-05,
+      "loss": 1.0695,
+      "step": 7866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4297770142939126e-05,
+      "loss": 0.3439,
+      "step": 7867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4296366394194556e-05,
+      "loss": 1.2796,
+      "step": 7868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4294962490837145e-05,
+      "loss": 0.9088,
+      "step": 7869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4293558432875845e-05,
+      "loss": 0.6802,
+      "step": 7870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.429215422031958e-05,
+      "loss": 1.017,
+      "step": 7871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.429074985317728e-05,
+      "loss": 0.471,
+      "step": 7872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4289345331457896e-05,
+      "loss": 0.3953,
+      "step": 7873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.428794065517035e-05,
+      "loss": 1.2691,
+      "step": 7874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.428653582432359e-05,
+      "loss": 1.0207,
+      "step": 7875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.428513083892657e-05,
+      "loss": 1.3453,
+      "step": 7876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.42837256989882e-05,
+      "loss": 1.3783,
+      "step": 7877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4282320404517445e-05,
+      "loss": 0.1248,
+      "step": 7878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.428091495552324e-05,
+      "loss": 0.0738,
+      "step": 7879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4279509352014525e-05,
+      "loss": 1.3072,
+      "step": 7880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.427810359400025e-05,
+      "loss": 0.1925,
+      "step": 7881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4276697681489363e-05,
+      "loss": 0.6924,
+      "step": 7882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4275291614490806e-05,
+      "loss": 1.3856,
+      "step": 7883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4273885393013524e-05,
+      "loss": 1.0657,
+      "step": 7884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.427247901706647e-05,
+      "loss": 0.322,
+      "step": 7885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.427107248665859e-05,
+      "loss": 0.6099,
+      "step": 7886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.426966580179885e-05,
+      "loss": 1.3483,
+      "step": 7887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4268258962496184e-05,
+      "loss": 1.4156,
+      "step": 7888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.426685196875955e-05,
+      "loss": 0.8041,
+      "step": 7889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.42654448205979e-05,
+      "loss": 0.8556,
+      "step": 7890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.426403751802019e-05,
+      "loss": 1.3496,
+      "step": 7891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.426263006103538e-05,
+      "loss": 0.8241,
+      "step": 7892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.426122244965242e-05,
+      "loss": 0.8954,
+      "step": 7893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.425981468388027e-05,
+      "loss": 1.0842,
+      "step": 7894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4258406763727886e-05,
+      "loss": 0.4119,
+      "step": 7895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4256998689204235e-05,
+      "loss": 0.4666,
+      "step": 7896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.425559046031827e-05,
+      "loss": 1.377,
+      "step": 7897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.425418207707895e-05,
+      "loss": 0.8928,
+      "step": 7898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4252773539495236e-05,
+      "loss": 1.0988,
+      "step": 7899
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4251364847576114e-05,
+      "loss": 0.6634,
+      "step": 7900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.424995600133053e-05,
+      "loss": 1.6473,
+      "step": 7901
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4248547000767435e-05,
+      "loss": 0.6737,
+      "step": 7902
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.424713784589582e-05,
+      "loss": 1.1921,
+      "step": 7903
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.424572853672465e-05,
+      "loss": 0.5401,
+      "step": 7904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4244319073262885e-05,
+      "loss": 0.5734,
+      "step": 7905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.424290945551949e-05,
+      "loss": 1.6514,
+      "step": 7906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.424149968350344e-05,
+      "loss": 0.7325,
+      "step": 7907
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4240089757223724e-05,
+      "loss": 1.1355,
+      "step": 7908
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4238679676689284e-05,
+      "loss": 0.9198,
+      "step": 7909
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4237269441909115e-05,
+      "loss": 1.5555,
+      "step": 7910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.423585905289218e-05,
+      "loss": 0.4066,
+      "step": 7911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.423444850964746e-05,
+      "loss": 0.7665,
+      "step": 7912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.423303781218392e-05,
+      "loss": 0.7332,
+      "step": 7913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.423162696051055e-05,
+      "loss": 1.0494,
+      "step": 7914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.423021595463633e-05,
+      "loss": 0.713,
+      "step": 7915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.422880479457023e-05,
+      "loss": 0.6219,
+      "step": 7916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.422739348032124e-05,
+      "loss": 1.335,
+      "step": 7917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.422598201189832e-05,
+      "loss": 0.634,
+      "step": 7918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.422457038931048e-05,
+      "loss": 0.5648,
+      "step": 7919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4223158612566683e-05,
+      "loss": 1.4604,
+      "step": 7920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.422174668167592e-05,
+      "loss": 0.8438,
+      "step": 7921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4220334596647175e-05,
+      "loss": 1.4154,
+      "step": 7922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.421892235748944e-05,
+      "loss": 1.2414,
+      "step": 7923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4217509964211694e-05,
+      "loss": 0.8741,
+      "step": 7924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.4216097416822924e-05,
+      "loss": 1.2684,
+      "step": 7925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.421468471533213e-05,
+      "loss": 1.0277,
+      "step": 7926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.421327185974828e-05,
+      "loss": 0.7958,
+      "step": 7927
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.421185885008039e-05,
+      "loss": 0.8278,
+      "step": 7928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4210445686337436e-05,
+      "loss": 1.2474,
+      "step": 7929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.420903236852842e-05,
+      "loss": 1.2804,
+      "step": 7930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4207618896662324e-05,
+      "loss": 1.8779,
+      "step": 7931
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.420620527074816e-05,
+      "loss": 0.7605,
+      "step": 7932
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.420479149079491e-05,
+      "loss": 0.9035,
+      "step": 7933
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4203377556811576e-05,
+      "loss": 0.7107,
+      "step": 7934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.420196346880715e-05,
+      "loss": 1.2094,
+      "step": 7935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4200549226790644e-05,
+      "loss": 1.6169,
+      "step": 7936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.419913483077104e-05,
+      "loss": 0.3594,
+      "step": 7937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.419772028075735e-05,
+      "loss": 1.2239,
+      "step": 7938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.419630557675858e-05,
+      "loss": 1.4103,
+      "step": 7939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.419489071878372e-05,
+      "loss": 1.5846,
+      "step": 7940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.419347570684178e-05,
+      "loss": 1.6411,
+      "step": 7941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4192060540941765e-05,
+      "loss": 1.4559,
+      "step": 7942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4190645221092674e-05,
+      "loss": 1.1319,
+      "step": 7943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.418922974730354e-05,
+      "loss": 1.2278,
+      "step": 7944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.418781411958333e-05,
+      "loss": 0.95,
+      "step": 7945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4186398337941075e-05,
+      "loss": 1.82,
+      "step": 7946
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.418498240238579e-05,
+      "loss": 1.2915,
+      "step": 7947
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.418356631292647e-05,
+      "loss": 1.1523,
+      "step": 7948
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.418215006957213e-05,
+      "loss": 0.9161,
+      "step": 7949
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.41807336723318e-05,
+      "loss": 1.4016,
+      "step": 7950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.417931712121447e-05,
+      "loss": 1.1108,
+      "step": 7951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4177900416229166e-05,
+      "loss": 1.229,
+      "step": 7952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.417648355738491e-05,
+      "loss": 0.4911,
+      "step": 7953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.417506654469069e-05,
+      "loss": 0.96,
+      "step": 7954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.417364937815556e-05,
+      "loss": 0.7018,
+      "step": 7955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4172232057788515e-05,
+      "loss": 1.0075,
+      "step": 7956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4170814583598584e-05,
+      "loss": 1.5251,
+      "step": 7957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.416939695559478e-05,
+      "loss": 0.7368,
+      "step": 7958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.416797917378613e-05,
+      "loss": 1.21,
+      "step": 7959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.416656123818165e-05,
+      "loss": 0.946,
+      "step": 7960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.416514314879038e-05,
+      "loss": 0.48,
+      "step": 7961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4163724905621325e-05,
+      "loss": 0.4094,
+      "step": 7962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.416230650868352e-05,
+      "loss": 1.0381,
+      "step": 7963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4160887957985986e-05,
+      "loss": 0.241,
+      "step": 7964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.415946925353775e-05,
+      "loss": 1.0136,
+      "step": 7965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.415805039534784e-05,
+      "loss": 0.7864,
+      "step": 7966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4156631383425294e-05,
+      "loss": 1.1305,
+      "step": 7967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.415521221777913e-05,
+      "loss": 0.5577,
+      "step": 7968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4153792898418385e-05,
+      "loss": 1.4346,
+      "step": 7969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.41523734253521e-05,
+      "loss": 0.2732,
+      "step": 7970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.415095379858929e-05,
+      "loss": 1.3386,
+      "step": 7971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4149534018138994e-05,
+      "loss": 0.4168,
+      "step": 7972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.414811408401026e-05,
+      "loss": 1.1688,
+      "step": 7973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.414669399621211e-05,
+      "loss": 0.468,
+      "step": 7974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.414527375475358e-05,
+      "loss": 1.019,
+      "step": 7975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4143853359643725e-05,
+      "loss": 0.4293,
+      "step": 7976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.414243281089156e-05,
+      "loss": 1.3681,
+      "step": 7977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.414101210850614e-05,
+      "loss": 0.4144,
+      "step": 7978
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4139591252496504e-05,
+      "loss": 0.5815,
+      "step": 7979
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4138170242871696e-05,
+      "loss": 1.3386,
+      "step": 7980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.413674907964075e-05,
+      "loss": 1.6683,
+      "step": 7981
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4135327762812716e-05,
+      "loss": 0.4048,
+      "step": 7982
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.413390629239664e-05,
+      "loss": 0.6095,
+      "step": 7983
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4132484668401566e-05,
+      "loss": 0.7967,
+      "step": 7984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.413106289083654e-05,
+      "loss": 0.6255,
+      "step": 7985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.41296409597106e-05,
+      "loss": 0.577,
+      "step": 7986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.412821887503281e-05,
+      "loss": 0.5035,
+      "step": 7987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.412679663681222e-05,
+      "loss": 1.8637,
+      "step": 7988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.412537424505787e-05,
+      "loss": 0.9849,
+      "step": 7989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.412395169977882e-05,
+      "loss": 0.6128,
+      "step": 7990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.412252900098412e-05,
+      "loss": 0.948,
+      "step": 7991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.412110614868281e-05,
+      "loss": 0.7777,
+      "step": 7992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.411968314288397e-05,
+      "loss": 1.3183,
+      "step": 7993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4118259983596635e-05,
+      "loss": 0.9609,
+      "step": 7994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.411683667082987e-05,
+      "loss": 1.5081,
+      "step": 7995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4115413204592734e-05,
+      "loss": 0.4772,
+      "step": 7996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4113989584894275e-05,
+      "loss": 0.3241,
+      "step": 7997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.411256581174356e-05,
+      "loss": 0.3224,
+      "step": 7998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.411114188514965e-05,
+      "loss": 1.313,
+      "step": 7999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.410971780512161e-05,
+      "loss": 1.2215,
+      "step": 8000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.41082935716685e-05,
+      "loss": 0.7152,
+      "step": 8001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.410686918479937e-05,
+      "loss": 1.2261,
+      "step": 8002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.410544464452331e-05,
+      "loss": 0.5345,
+      "step": 8003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.410401995084936e-05,
+      "loss": 1.0952,
+      "step": 8004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4102595103786594e-05,
+      "loss": 0.4921,
+      "step": 8005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.410117010334409e-05,
+      "loss": 0.613,
+      "step": 8006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.409974494953091e-05,
+      "loss": 0.4927,
+      "step": 8007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.409831964235612e-05,
+      "loss": 0.6398,
+      "step": 8008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.409689418182878e-05,
+      "loss": 1.2965,
+      "step": 8009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.409546856795798e-05,
+      "loss": 1.2399,
+      "step": 8010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4094042800752784e-05,
+      "loss": 1.2545,
+      "step": 8011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.409261688022226e-05,
+      "loss": 1.9152,
+      "step": 8012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.40911908063755e-05,
+      "loss": 1.1297,
+      "step": 8013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.408976457922155e-05,
+      "loss": 1.7507,
+      "step": 8014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.408833819876952e-05,
+      "loss": 0.805,
+      "step": 8015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.408691166502846e-05,
+      "loss": 1.3426,
+      "step": 8016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.408548497800747e-05,
+      "loss": 0.8976,
+      "step": 8017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4084058137715595e-05,
+      "loss": 0.331,
+      "step": 8018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4082631144161946e-05,
+      "loss": 0.3786,
+      "step": 8019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.4081203997355594e-05,
+      "loss": 1.7615,
+      "step": 8020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.407977669730563e-05,
+      "loss": 0.3824,
+      "step": 8021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.407834924402111e-05,
+      "loss": 1.5615,
+      "step": 8022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.407692163751114e-05,
+      "loss": 1.2049,
+      "step": 8023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.407549387778481e-05,
+      "loss": 0.3999,
+      "step": 8024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.407406596485118e-05,
+      "loss": 1.3984,
+      "step": 8025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.407263789871937e-05,
+      "loss": 1.1626,
+      "step": 8026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.407120967939844e-05,
+      "loss": 0.7105,
+      "step": 8027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4069781306897496e-05,
+      "loss": 0.3416,
+      "step": 8028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.406835278122561e-05,
+      "loss": 0.3097,
+      "step": 8029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.406692410239189e-05,
+      "loss": 0.7526,
+      "step": 8030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4065495270405415e-05,
+      "loss": 1.2324,
+      "step": 8031
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4064066285275286e-05,
+      "loss": 0.7623,
+      "step": 8032
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.406263714701059e-05,
+      "loss": 0.4622,
+      "step": 8033
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4061207855620434e-05,
+      "loss": 0.8212,
+      "step": 8034
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.40597784111139e-05,
+      "loss": 0.483,
+      "step": 8035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.405834881350009e-05,
+      "loss": 0.3937,
+      "step": 8036
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.40569190627881e-05,
+      "loss": 0.5071,
+      "step": 8037
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.405548915898702e-05,
+      "loss": 1.071,
+      "step": 8038
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4054059102105975e-05,
+      "loss": 0.9234,
+      "step": 8039
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4052628892154035e-05,
+      "loss": 1.2087,
+      "step": 8040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4051198529140326e-05,
+      "loss": 0.273,
+      "step": 8041
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4049768013073924e-05,
+      "loss": 0.8835,
+      "step": 8042
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4048337343963956e-05,
+      "loss": 0.5004,
+      "step": 8043
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.404690652181952e-05,
+      "loss": 0.6181,
+      "step": 8044
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.404547554664971e-05,
+      "loss": 0.515,
+      "step": 8045
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.404404441846365e-05,
+      "loss": 1.4714,
+      "step": 8046
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.404261313727044e-05,
+      "loss": 0.3183,
+      "step": 8047
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.404118170307918e-05,
+      "loss": 0.8924,
+      "step": 8048
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.403975011589899e-05,
+      "loss": 0.9814,
+      "step": 8049
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.403831837573897e-05,
+      "loss": 1.2936,
+      "step": 8050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4036886482608236e-05,
+      "loss": 0.3139,
+      "step": 8051
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4035454436515904e-05,
+      "loss": 0.554,
+      "step": 8052
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4034022237471086e-05,
+      "loss": 0.9922,
+      "step": 8053
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4032589885482895e-05,
+      "loss": 0.4481,
+      "step": 8054
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.403115738056044e-05,
+      "loss": 0.8808,
+      "step": 8055
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.402972472271284e-05,
+      "loss": 0.9388,
+      "step": 8056
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4028291911949215e-05,
+      "loss": 1.3513,
+      "step": 8057
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4026858948278675e-05,
+      "loss": 0.8861,
+      "step": 8058
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.402542583171035e-05,
+      "loss": 0.3328,
+      "step": 8059
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4023992562253357e-05,
+      "loss": 0.4163,
+      "step": 8060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4022559139916813e-05,
+      "loss": 0.8689,
+      "step": 8061
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4021125564709844e-05,
+      "loss": 0.8072,
+      "step": 8062
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.401969183664156e-05,
+      "loss": 0.8861,
+      "step": 8063
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.401825795572111e-05,
+      "loss": 0.3351,
+      "step": 8064
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.401682392195759e-05,
+      "loss": 1.0313,
+      "step": 8065
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4015389735360145e-05,
+      "loss": 1.4335,
+      "step": 8066
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.40139553959379e-05,
+      "loss": 0.4098,
+      "step": 8067
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4012520903699966e-05,
+      "loss": 1.5752,
+      "step": 8068
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4011086258655494e-05,
+      "loss": 0.3557,
+      "step": 8069
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.40096514608136e-05,
+      "loss": 0.9846,
+      "step": 8070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4008216510183414e-05,
+      "loss": 1.1083,
+      "step": 8071
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.400678140677408e-05,
+      "loss": 1.4603,
+      "step": 8072
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4005346150594715e-05,
+      "loss": 0.5966,
+      "step": 8073
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4003910741654466e-05,
+      "loss": 1.2424,
+      "step": 8074
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.400247517996245e-05,
+      "loss": 1.4373,
+      "step": 8075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.4001039465527824e-05,
+      "loss": 1.0376,
+      "step": 8076
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.399960359835971e-05,
+      "loss": 0.2754,
+      "step": 8077
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.3998167578467244e-05,
+      "loss": 0.97,
+      "step": 8078
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.3996731405859566e-05,
+      "loss": 1.3274,
+      "step": 8079
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.399529508054583e-05,
+      "loss": 0.3478,
+      "step": 8080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.3993858602535156e-05,
+      "loss": 0.4266,
+      "step": 8081
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.39924219718367e-05,
+      "loss": 1.2323,
+      "step": 8082
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.39909851884596e-05,
+      "loss": 1.0284,
+      "step": 8083
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.398954825241298e-05,
+      "loss": 0.1579,
+      "step": 8084
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.398811116370601e-05,
+      "loss": 1.4546,
+      "step": 8085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.398667392234783e-05,
+      "loss": 1.075,
+      "step": 8086
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.398523652834758e-05,
+      "loss": 0.5349,
+      "step": 8087
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.398379898171441e-05,
+      "loss": 0.2661,
+      "step": 8088
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.398236128245747e-05,
+      "loss": 1.1196,
+      "step": 8089
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.3980923430585905e-05,
+      "loss": 0.8033,
+      "step": 8090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.397948542610886e-05,
+      "loss": 0.1257,
+      "step": 8091
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.397804726903549e-05,
+      "loss": 1.3423,
+      "step": 8092
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.397660895937495e-05,
+      "loss": 0.7701,
+      "step": 8093
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.39751704971364e-05,
+      "loss": 1.4494,
+      "step": 8094
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.3973731882328986e-05,
+      "loss": 1.1779,
+      "step": 8095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.397229311496186e-05,
+      "loss": 1.5842,
+      "step": 8096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.397085419504418e-05,
+      "loss": 2.1032,
+      "step": 8097
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.39694151225851e-05,
+      "loss": 0.2669,
+      "step": 8098
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.3967975897593774e-05,
+      "loss": 1.0311,
+      "step": 8099
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.396653652007938e-05,
+      "loss": 0.8566,
+      "step": 8100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.396509699005106e-05,
+      "loss": 1.274,
+      "step": 8101
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.3963657307517983e-05,
+      "loss": 0.5878,
+      "step": 8102
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.39622174724893e-05,
+      "loss": 1.3545,
+      "step": 8103
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.396077748497418e-05,
+      "loss": 1.1924,
+      "step": 8104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.39593373449818e-05,
+      "loss": 0.9488,
+      "step": 8105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.39578970525213e-05,
+      "loss": 1.3453,
+      "step": 8106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.3956456607601857e-05,
+      "loss": 1.056,
+      "step": 8107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.3955016010232644e-05,
+      "loss": 0.8688,
+      "step": 8108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.395357526042281e-05,
+      "loss": 1.3315,
+      "step": 8109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.395213435818156e-05,
+      "loss": 1.4185,
+      "step": 8110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.395069330351802e-05,
+      "loss": 1.5236,
+      "step": 8111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.394925209644138e-05,
+      "loss": 1.0306,
+      "step": 8112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.394781073696081e-05,
+      "loss": 0.7501,
+      "step": 8113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.3946369225085485e-05,
+      "loss": 1.3369,
+      "step": 8114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.394492756082458e-05,
+      "loss": 0.6079,
+      "step": 8115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.394348574418725e-05,
+      "loss": 1.7226,
+      "step": 8116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.394204377518271e-05,
+      "loss": 1.4371,
+      "step": 8117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.3940601653820096e-05,
+      "loss": 1.565,
+      "step": 8118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.39391593801086e-05,
+      "loss": 1.0849,
+      "step": 8119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.3937716954057404e-05,
+      "loss": 1.247,
+      "step": 8120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.393627437567568e-05,
+      "loss": 1.67,
+      "step": 8121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.393483164497262e-05,
+      "loss": 0.6812,
+      "step": 8122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.3933388761957386e-05,
+      "loss": 0.3219,
+      "step": 8123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.393194572663918e-05,
+      "loss": 0.9128,
+      "step": 8124
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.393050253902717e-05,
+      "loss": 0.3757,
+      "step": 8125
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3929059199130554e-05,
+      "loss": 0.4561,
+      "step": 8126
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3927615706958495e-05,
+      "loss": 1.0718,
+      "step": 8127
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.39261720625202e-05,
+      "loss": 0.5135,
+      "step": 8128
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.392472826582485e-05,
+      "loss": 0.6971,
+      "step": 8129
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.392328431688162e-05,
+      "loss": 0.2041,
+      "step": 8130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.392184021569972e-05,
+      "loss": 1.3082,
+      "step": 8131
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.392039596228833e-05,
+      "loss": 1.5793,
+      "step": 8132
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.391895155665663e-05,
+      "loss": 1.1927,
+      "step": 8133
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3917506998813835e-05,
+      "loss": 1.3903,
+      "step": 8134
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.391606228876912e-05,
+      "loss": 0.77,
+      "step": 8135
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3914617426531674e-05,
+      "loss": 1.1476,
+      "step": 8136
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.391317241211071e-05,
+      "loss": 0.7352,
+      "step": 8137
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.391172724551541e-05,
+      "loss": 1.6502,
+      "step": 8138
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.391028192675497e-05,
+      "loss": 1.4729,
+      "step": 8139
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.39088364558386e-05,
+      "loss": 0.4136,
+      "step": 8140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.390739083277548e-05,
+      "loss": 1.1907,
+      "step": 8141
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3905945057574827e-05,
+      "loss": 0.571,
+      "step": 8142
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.390449913024583e-05,
+      "loss": 1.0514,
+      "step": 8143
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3903053050797705e-05,
+      "loss": 0.2428,
+      "step": 8144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.390160681923963e-05,
+      "loss": 1.1869,
+      "step": 8145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3900160435580825e-05,
+      "loss": 0.1945,
+      "step": 8146
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.389871389983049e-05,
+      "loss": 0.4948,
+      "step": 8147
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.389726721199784e-05,
+      "loss": 1.0626,
+      "step": 8148
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.389582037209207e-05,
+      "loss": 0.9271,
+      "step": 8149
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.389437338012238e-05,
+      "loss": 0.6779,
+      "step": 8150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3892926236098004e-05,
+      "loss": 1.3031,
+      "step": 8151
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3891478940028116e-05,
+      "loss": 0.3579,
+      "step": 8152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.389003149192196e-05,
+      "loss": 1.3825,
+      "step": 8153
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3888583891788716e-05,
+      "loss": 0.223,
+      "step": 8154
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.388713613963762e-05,
+      "loss": 0.4513,
+      "step": 8155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.388568823547788e-05,
+      "loss": 0.0817,
+      "step": 8156
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.388424017931871e-05,
+      "loss": 1.2321,
+      "step": 8157
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.388279197116931e-05,
+      "loss": 0.9499,
+      "step": 8158
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3881343611038915e-05,
+      "loss": 1.5718,
+      "step": 8159
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.387989509893673e-05,
+      "loss": 0.9514,
+      "step": 8160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.387844643487199e-05,
+      "loss": 0.9839,
+      "step": 8161
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3876997618853884e-05,
+      "loss": 0.7536,
+      "step": 8162
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3875548650891654e-05,
+      "loss": 0.697,
+      "step": 8163
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.387409953099452e-05,
+      "loss": 0.454,
+      "step": 8164
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.387265025917169e-05,
+      "loss": 1.0411,
+      "step": 8165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.387120083543241e-05,
+      "loss": 1.0971,
+      "step": 8166
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.386975125978587e-05,
+      "loss": 1.1411,
+      "step": 8167
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.386830153224133e-05,
+      "loss": 0.5457,
+      "step": 8168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3866851652807984e-05,
+      "loss": 0.773,
+      "step": 8169
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.386540162149508e-05,
+      "loss": 0.5156,
+      "step": 8170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3863951438311846e-05,
+      "loss": 0.2995,
+      "step": 8171
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.386250110326749e-05,
+      "loss": 1.1059,
+      "step": 8172
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3861050616371274e-05,
+      "loss": 0.4383,
+      "step": 8173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3859599977632396e-05,
+      "loss": 0.6272,
+      "step": 8174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.38581491870601e-05,
+      "loss": 1.6697,
+      "step": 8175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.385669824466362e-05,
+      "loss": 0.5715,
+      "step": 8176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.385524715045219e-05,
+      "loss": 0.2641,
+      "step": 8177
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.385379590443503e-05,
+      "loss": 1.2799,
+      "step": 8178
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.38523445066214e-05,
+      "loss": 0.2954,
+      "step": 8179
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.385089295702052e-05,
+      "loss": 0.2329,
+      "step": 8180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.384944125564164e-05,
+      "loss": 1.5349,
+      "step": 8181
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.384798940249398e-05,
+      "loss": 0.8667,
+      "step": 8182
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3846537397586784e-05,
+      "loss": 0.6316,
+      "step": 8183
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.38450852409293e-05,
+      "loss": 1.3842,
+      "step": 8184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.384363293253077e-05,
+      "loss": 0.9685,
+      "step": 8185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3842180472400415e-05,
+      "loss": 1.2023,
+      "step": 8186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.384072786054751e-05,
+      "loss": 0.7942,
+      "step": 8187
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.383927509698127e-05,
+      "loss": 0.8329,
+      "step": 8188
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.383782218171096e-05,
+      "loss": 0.6563,
+      "step": 8189
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3836369114745826e-05,
+      "loss": 0.4649,
+      "step": 8190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3834915896095084e-05,
+      "loss": 0.2362,
+      "step": 8191
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.383346252576803e-05,
+      "loss": 0.7928,
+      "step": 8192
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.383200900377387e-05,
+      "loss": 0.3597,
+      "step": 8193
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.383055533012188e-05,
+      "loss": 0.4464,
+      "step": 8194
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.382910150482129e-05,
+      "loss": 0.6837,
+      "step": 8195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.382764752788136e-05,
+      "loss": 0.8022,
+      "step": 8196
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3826193399311366e-05,
+      "loss": 0.6209,
+      "step": 8197
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.382473911912053e-05,
+      "loss": 1.1217,
+      "step": 8198
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3823284687318115e-05,
+      "loss": 0.886,
+      "step": 8199
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.382183010391339e-05,
+      "loss": 0.3072,
+      "step": 8200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.382037536891559e-05,
+      "loss": 0.699,
+      "step": 8201
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.381892048233398e-05,
+      "loss": 0.6168,
+      "step": 8202
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.381746544417783e-05,
+      "loss": 0.6521,
+      "step": 8203
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.381601025445639e-05,
+      "loss": 0.6118,
+      "step": 8204
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3814554913178914e-05,
+      "loss": 0.9958,
+      "step": 8205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.381309942035468e-05,
+      "loss": 0.546,
+      "step": 8206
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3811643775992934e-05,
+      "loss": 0.4145,
+      "step": 8207
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.381018798010294e-05,
+      "loss": 0.4685,
+      "step": 8208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3808732032693974e-05,
+      "loss": 0.3517,
+      "step": 8209
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.38072759337753e-05,
+      "loss": 1.1924,
+      "step": 8210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3805819683356174e-05,
+      "loss": 0.6019,
+      "step": 8211
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.380436328144586e-05,
+      "loss": 1.1517,
+      "step": 8212
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.380290672805365e-05,
+      "loss": 0.2242,
+      "step": 8213
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.380145002318878e-05,
+      "loss": 0.8152,
+      "step": 8214
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.379999316686055e-05,
+      "loss": 0.4976,
+      "step": 8215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3798536159078204e-05,
+      "loss": 1.383,
+      "step": 8216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3797078999851034e-05,
+      "loss": 1.1082,
+      "step": 8217
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.379562168918831e-05,
+      "loss": 1.332,
+      "step": 8218
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.37941642270993e-05,
+      "loss": 1.2521,
+      "step": 8219
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.3792706613593276e-05,
+      "loss": 0.7029,
+      "step": 8220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.379124884867953e-05,
+      "loss": 0.7549,
+      "step": 8221
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.378979093236731e-05,
+      "loss": 0.6627,
+      "step": 8222
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.378833286466592e-05,
+      "loss": 0.8923,
+      "step": 8223
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.378687464558463e-05,
+      "loss": 0.7376,
+      "step": 8224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.378541627513272e-05,
+      "loss": 0.9852,
+      "step": 8225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3783957753319466e-05,
+      "loss": 0.8107,
+      "step": 8226
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3782499080154144e-05,
+      "loss": 0.5132,
+      "step": 8227
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.378104025564606e-05,
+      "loss": 1.0686,
+      "step": 8228
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.377958127980448e-05,
+      "loss": 0.492,
+      "step": 8229
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3778122152638684e-05,
+      "loss": 1.0281,
+      "step": 8230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.377666287415797e-05,
+      "loss": 0.8901,
+      "step": 8231
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.377520344437162e-05,
+      "loss": 0.9127,
+      "step": 8232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.377374386328891e-05,
+      "loss": 1.2276,
+      "step": 8233
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.377228413091915e-05,
+      "loss": 0.5884,
+      "step": 8234
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.377082424727161e-05,
+      "loss": 1.6628,
+      "step": 8235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.376936421235558e-05,
+      "loss": 1.2376,
+      "step": 8236
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3767904026180374e-05,
+      "loss": 0.5759,
+      "step": 8237
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3766443688755265e-05,
+      "loss": 0.4092,
+      "step": 8238
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.376498320008954e-05,
+      "loss": 1.566,
+      "step": 8239
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.376352256019252e-05,
+      "loss": 1.2276,
+      "step": 8240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.376206176907346e-05,
+      "loss": 1.0346,
+      "step": 8241
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.376060082674169e-05,
+      "loss": 1.1194,
+      "step": 8242
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.37591397332065e-05,
+      "loss": 0.879,
+      "step": 8243
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.375767848847718e-05,
+      "loss": 0.6193,
+      "step": 8244
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.375621709256303e-05,
+      "loss": 0.4481,
+      "step": 8245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3754755545473354e-05,
+      "loss": 1.3265,
+      "step": 8246
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.375329384721745e-05,
+      "loss": 0.8484,
+      "step": 8247
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3751831997804616e-05,
+      "loss": 1.2202,
+      "step": 8248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.375036999724416e-05,
+      "loss": 0.8404,
+      "step": 8249
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.374890784554539e-05,
+      "loss": 0.3952,
+      "step": 8250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.37474455427176e-05,
+      "loss": 1.1035,
+      "step": 8251
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.37459830887701e-05,
+      "loss": 0.9759,
+      "step": 8252
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3744520483712207e-05,
+      "loss": 1.035,
+      "step": 8253
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.374305772755321e-05,
+      "loss": 0.7936,
+      "step": 8254
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.374159482030242e-05,
+      "loss": 1.2119,
+      "step": 8255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.374013176196916e-05,
+      "loss": 1.337,
+      "step": 8256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3738668552562736e-05,
+      "loss": 0.5665,
+      "step": 8257
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.373720519209245e-05,
+      "loss": 1.408,
+      "step": 8258
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.373574168056762e-05,
+      "loss": 1.1812,
+      "step": 8259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3734278017997564e-05,
+      "loss": 0.7788,
+      "step": 8260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3732814204391584e-05,
+      "loss": 1.4514,
+      "step": 8261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.373135023975901e-05,
+      "loss": 0.438,
+      "step": 8262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.372988612410915e-05,
+      "loss": 0.3641,
+      "step": 8263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.372842185745132e-05,
+      "loss": 0.4293,
+      "step": 8264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.372695743979484e-05,
+      "loss": 1.9528,
+      "step": 8265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3725492871149036e-05,
+      "loss": 0.7908,
+      "step": 8266
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.37240281515232e-05,
+      "loss": 0.5781,
+      "step": 8267
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.37225632809267e-05,
+      "loss": 1.3558,
+      "step": 8268
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3721098259368825e-05,
+      "loss": 0.3672,
+      "step": 8269
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.371963308685889e-05,
+      "loss": 1.5859,
+      "step": 8270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3718167763406245e-05,
+      "loss": 0.7089,
+      "step": 8271
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3716702289020204e-05,
+      "loss": 1.5599,
+      "step": 8272
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.371523666371009e-05,
+      "loss": 1.5552,
+      "step": 8273
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.371377088748523e-05,
+      "loss": 1.1627,
+      "step": 8274
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.371230496035496e-05,
+      "loss": 0.8455,
+      "step": 8275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.371083888232859e-05,
+      "loss": 1.3767,
+      "step": 8276
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.370937265341547e-05,
+      "loss": 0.679,
+      "step": 8277
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.370790627362492e-05,
+      "loss": 0.4338,
+      "step": 8278
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3706439742966274e-05,
+      "loss": 0.4543,
+      "step": 8279
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.370497306144886e-05,
+      "loss": 1.3578,
+      "step": 8280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.370350622908202e-05,
+      "loss": 0.4613,
+      "step": 8281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3702039245875086e-05,
+      "loss": 0.2544,
+      "step": 8282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.370057211183739e-05,
+      "loss": 0.9867,
+      "step": 8283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.369910482697826e-05,
+      "loss": 2.096,
+      "step": 8284
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.369763739130705e-05,
+      "loss": 1.6217,
+      "step": 8285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.369616980483309e-05,
+      "loss": 0.4812,
+      "step": 8286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.369470206756572e-05,
+      "loss": 0.6908,
+      "step": 8287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.369323417951428e-05,
+      "loss": 0.4236,
+      "step": 8288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.369176614068811e-05,
+      "loss": 0.3672,
+      "step": 8289
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3690297951096544e-05,
+      "loss": 1.5754,
+      "step": 8290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3688829610748945e-05,
+      "loss": 0.4998,
+      "step": 8291
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.368736111965465e-05,
+      "loss": 1.4565,
+      "step": 8292
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3685892477822993e-05,
+      "loss": 1.0141,
+      "step": 8293
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.368442368526333e-05,
+      "loss": 0.2994,
+      "step": 8294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3682954741985e-05,
+      "loss": 1.6018,
+      "step": 8295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.368148564799736e-05,
+      "loss": 0.8759,
+      "step": 8296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3680016403309755e-05,
+      "loss": 1.6051,
+      "step": 8297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.367854700793153e-05,
+      "loss": 0.3145,
+      "step": 8298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.367707746187204e-05,
+      "loss": 1.167,
+      "step": 8299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.367560776514064e-05,
+      "loss": 1.4416,
+      "step": 8300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3674137917746676e-05,
+      "loss": 0.6476,
+      "step": 8301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.36726679196995e-05,
+      "loss": 1.2872,
+      "step": 8302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3671197771008475e-05,
+      "loss": 0.5675,
+      "step": 8303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.366972747168296e-05,
+      "loss": 0.4553,
+      "step": 8304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.366825702173229e-05,
+      "loss": 0.4087,
+      "step": 8305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.366678642116585e-05,
+      "loss": 1.0998,
+      "step": 8306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.366531566999297e-05,
+      "loss": 1.3622,
+      "step": 8307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.366384476822304e-05,
+      "loss": 0.7,
+      "step": 8308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.366237371586539e-05,
+      "loss": 0.4471,
+      "step": 8309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3660902512929406e-05,
+      "loss": 0.7711,
+      "step": 8310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.365943115942443e-05,
+      "loss": 1.2572,
+      "step": 8311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.365795965535984e-05,
+      "loss": 0.9504,
+      "step": 8312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3656488000744993e-05,
+      "loss": 1.3816,
+      "step": 8313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3655016195589256e-05,
+      "loss": 0.949,
+      "step": 8314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3653544239902e-05,
+      "loss": 1.1444,
+      "step": 8315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3652072133692574e-05,
+      "loss": 0.9097,
+      "step": 8316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.365059987697037e-05,
+      "loss": 1.2181,
+      "step": 8317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.364912746974474e-05,
+      "loss": 0.8571,
+      "step": 8318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3647654912025066e-05,
+      "loss": 0.4328,
+      "step": 8319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.3646182203820703e-05,
+      "loss": 1.5133,
+      "step": 8320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.364470934514103e-05,
+      "loss": 0.8117,
+      "step": 8321
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.364323633599543e-05,
+      "loss": 0.4109,
+      "step": 8322
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3641763176393274e-05,
+      "loss": 0.1737,
+      "step": 8323
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.364028986634392e-05,
+      "loss": 1.0912,
+      "step": 8324
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3638816405856754e-05,
+      "loss": 0.7446,
+      "step": 8325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3637342794941163e-05,
+      "loss": 0.4407,
+      "step": 8326
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.363586903360651e-05,
+      "loss": 1.43,
+      "step": 8327
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.363439512186217e-05,
+      "loss": 0.6075,
+      "step": 8328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.363292105971754e-05,
+      "loss": 1.4124,
+      "step": 8329
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3631446847181986e-05,
+      "loss": 0.9967,
+      "step": 8330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3629972484264894e-05,
+      "loss": 0.8709,
+      "step": 8331
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3628497970975655e-05,
+      "loss": 0.7139,
+      "step": 8332
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3627023307323634e-05,
+      "loss": 0.9695,
+      "step": 8333
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.362554849331823e-05,
+      "loss": 0.8265,
+      "step": 8334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3624073528968816e-05,
+      "loss": 0.4142,
+      "step": 8335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.362259841428479e-05,
+      "loss": 1.3335,
+      "step": 8336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3621123149275534e-05,
+      "loss": 1.1933,
+      "step": 8337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.361964773395044e-05,
+      "loss": 1.5012,
+      "step": 8338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.361817216831889e-05,
+      "loss": 0.7413,
+      "step": 8339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3616696452390275e-05,
+      "loss": 0.6879,
+      "step": 8340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.361522058617399e-05,
+      "loss": 0.9353,
+      "step": 8341
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.361374456967943e-05,
+      "loss": 0.7212,
+      "step": 8342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.361226840291597e-05,
+      "loss": 0.5763,
+      "step": 8343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.361079208589302e-05,
+      "loss": 1.2229,
+      "step": 8344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.360931561861998e-05,
+      "loss": 0.3827,
+      "step": 8345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.360783900110623e-05,
+      "loss": 0.4196,
+      "step": 8346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3606362233361173e-05,
+      "loss": 1.2371,
+      "step": 8347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.360488531539421e-05,
+      "loss": 1.4126,
+      "step": 8348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3603408247214735e-05,
+      "loss": 0.8794,
+      "step": 8349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.360193102883215e-05,
+      "loss": 0.8617,
+      "step": 8350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3600453660255854e-05,
+      "loss": 1.4716,
+      "step": 8351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.359897614149525e-05,
+      "loss": 0.7823,
+      "step": 8352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.359749847255974e-05,
+      "loss": 0.7597,
+      "step": 8353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.359602065345873e-05,
+      "loss": 0.9841,
+      "step": 8354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.359454268420161e-05,
+      "loss": 0.2977,
+      "step": 8355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.359306456479781e-05,
+      "loss": 0.4144,
+      "step": 8356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.359158629525671e-05,
+      "loss": 0.5834,
+      "step": 8357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.359010787558773e-05,
+      "loss": 0.3406,
+      "step": 8358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.358862930580028e-05,
+      "loss": 1.181,
+      "step": 8359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.358715058590377e-05,
+      "loss": 1.2065,
+      "step": 8360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.358567171590761e-05,
+      "loss": 0.8934,
+      "step": 8361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.35841926958212e-05,
+      "loss": 1.4711,
+      "step": 8362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.358271352565396e-05,
+      "loss": 1.3827,
+      "step": 8363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.358123420541531e-05,
+      "loss": 0.7511,
+      "step": 8364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.357975473511465e-05,
+      "loss": 1.1469,
+      "step": 8365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3578275114761404e-05,
+      "loss": 0.3824,
+      "step": 8366
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.357679534436498e-05,
+      "loss": 0.384,
+      "step": 8367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.357531542393481e-05,
+      "loss": 0.6761,
+      "step": 8368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3573835353480294e-05,
+      "loss": 0.6949,
+      "step": 8369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.357235513301086e-05,
+      "loss": 0.8299,
+      "step": 8370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3570874762535916e-05,
+      "loss": 0.3872,
+      "step": 8371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.35693942420649e-05,
+      "loss": 1.6778,
+      "step": 8372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.356791357160723e-05,
+      "loss": 0.6663,
+      "step": 8373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3566432751172314e-05,
+      "loss": 1.1177,
+      "step": 8374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3564951780769586e-05,
+      "loss": 0.7646,
+      "step": 8375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.356347066040847e-05,
+      "loss": 0.7312,
+      "step": 8376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.356198939009839e-05,
+      "loss": 0.6116,
+      "step": 8377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.356050796984878e-05,
+      "loss": 0.5153,
+      "step": 8378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3559026399669065e-05,
+      "loss": 0.8224,
+      "step": 8379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3557544679568656e-05,
+      "loss": 0.8319,
+      "step": 8380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3556062809557e-05,
+      "loss": 1.8766,
+      "step": 8381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.355458078964352e-05,
+      "loss": 0.205,
+      "step": 8382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.355309861983764e-05,
+      "loss": 0.5681,
+      "step": 8383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.355161630014881e-05,
+      "loss": 0.9869,
+      "step": 8384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.355013383058645e-05,
+      "loss": 0.7785,
+      "step": 8385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3548651211160005e-05,
+      "loss": 1.489,
+      "step": 8386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.354716844187889e-05,
+      "loss": 0.5513,
+      "step": 8387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.354568552275256e-05,
+      "loss": 0.9085,
+      "step": 8388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.354420245379045e-05,
+      "loss": 0.7716,
+      "step": 8389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3542719235001976e-05,
+      "loss": 1.0735,
+      "step": 8390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.35412358663966e-05,
+      "loss": 1.4566,
+      "step": 8391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.353975234798376e-05,
+      "loss": 1.1874,
+      "step": 8392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.353826867977289e-05,
+      "loss": 1.1612,
+      "step": 8393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3536784861773436e-05,
+      "loss": 0.3396,
+      "step": 8394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.353530089399483e-05,
+      "loss": 0.9394,
+      "step": 8395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3533816776446516e-05,
+      "loss": 0.645,
+      "step": 8396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.353233250913795e-05,
+      "loss": 0.9081,
+      "step": 8397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3530848092078584e-05,
+      "loss": 1.0612,
+      "step": 8398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.352936352527784e-05,
+      "loss": 0.7362,
+      "step": 8399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3527878808745174e-05,
+      "loss": 1.1324,
+      "step": 8400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.352639394249005e-05,
+      "loss": 1.2015,
+      "step": 8401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3524908926521895e-05,
+      "loss": 0.167,
+      "step": 8402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.352342376085018e-05,
+      "loss": 0.593,
+      "step": 8403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.352193844548434e-05,
+      "loss": 1.1316,
+      "step": 8404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3520452980433825e-05,
+      "loss": 1.0999,
+      "step": 8405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.35189673657081e-05,
+      "loss": 0.3469,
+      "step": 8406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3517481601316616e-05,
+      "loss": 0.6117,
+      "step": 8407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3515995687268825e-05,
+      "loss": 0.4625,
+      "step": 8408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.351450962357418e-05,
+      "loss": 1.0954,
+      "step": 8409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.351302341024214e-05,
+      "loss": 1.8786,
+      "step": 8410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.351153704728217e-05,
+      "loss": 0.4191,
+      "step": 8411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.351005053470372e-05,
+      "loss": 0.8929,
+      "step": 8412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.350856387251625e-05,
+      "loss": 0.5899,
+      "step": 8413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.350707706072922e-05,
+      "loss": 1.6129,
+      "step": 8414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.35055900993521e-05,
+      "loss": 0.5545,
+      "step": 8415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3504102988394344e-05,
+      "loss": 1.1637,
+      "step": 8416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.350261572786542e-05,
+      "loss": 1.0819,
+      "step": 8417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.350112831777478e-05,
+      "loss": 0.6801,
+      "step": 8418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.349964075813192e-05,
+      "loss": 0.6826,
+      "step": 8419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.3498153048946273e-05,
+      "loss": 1.5029,
+      "step": 8420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.349666519022732e-05,
+      "loss": 0.2415,
+      "step": 8421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.349517718198453e-05,
+      "loss": 0.9872,
+      "step": 8422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3493689024227374e-05,
+      "loss": 0.678,
+      "step": 8423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.349220071696531e-05,
+      "loss": 0.794,
+      "step": 8424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.349071226020782e-05,
+      "loss": 0.4654,
+      "step": 8425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3489223653964376e-05,
+      "loss": 0.6321,
+      "step": 8426
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.348773489824444e-05,
+      "loss": 1.2325,
+      "step": 8427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.348624599305751e-05,
+      "loss": 1.5838,
+      "step": 8428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.348475693841303e-05,
+      "loss": 0.9588,
+      "step": 8429
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.34832677343205e-05,
+      "loss": 1.2477,
+      "step": 8430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.348177838078938e-05,
+      "loss": 1.2008,
+      "step": 8431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3480288877829165e-05,
+      "loss": 1.1091,
+      "step": 8432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.347879922544932e-05,
+      "loss": 1.0644,
+      "step": 8433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3477309423659326e-05,
+      "loss": 0.7742,
+      "step": 8434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3475819472468675e-05,
+      "loss": 0.4222,
+      "step": 8435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.347432937188683e-05,
+      "loss": 0.3056,
+      "step": 8436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.347283912192328e-05,
+      "loss": 0.9835,
+      "step": 8437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.347134872258752e-05,
+      "loss": 0.4974,
+      "step": 8438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3469858173889026e-05,
+      "loss": 1.1752,
+      "step": 8439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.346836747583728e-05,
+      "loss": 0.8958,
+      "step": 8440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.346687662844177e-05,
+      "loss": 0.323,
+      "step": 8441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3465385631711985e-05,
+      "loss": 0.6192,
+      "step": 8442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.346389448565741e-05,
+      "loss": 0.1356,
+      "step": 8443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.346240319028754e-05,
+      "loss": 0.9847,
+      "step": 8444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.346091174561186e-05,
+      "loss": 0.6762,
+      "step": 8445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.345942015163986e-05,
+      "loss": 0.3406,
+      "step": 8446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3457928408381044e-05,
+      "loss": 1.4562,
+      "step": 8447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3456436515844885e-05,
+      "loss": 0.8358,
+      "step": 8448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.34549444740409e-05,
+      "loss": 1.4738,
+      "step": 8449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.345345228297856e-05,
+      "loss": 1.906,
+      "step": 8450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.345195994266738e-05,
+      "loss": 0.9956,
+      "step": 8451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3450467453116835e-05,
+      "loss": 0.2785,
+      "step": 8452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.344897481433645e-05,
+      "loss": 0.6076,
+      "step": 8453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.34474820263357e-05,
+      "loss": 1.0478,
+      "step": 8454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3445989089124095e-05,
+      "loss": 0.9223,
+      "step": 8455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3444496002711144e-05,
+      "loss": 0.9917,
+      "step": 8456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.344300276710633e-05,
+      "loss": 0.4175,
+      "step": 8457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3441509382319174e-05,
+      "loss": 0.2222,
+      "step": 8458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.344001584835916e-05,
+      "loss": 0.516,
+      "step": 8459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.34385221652358e-05,
+      "loss": 1.799,
+      "step": 8460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3437028332958616e-05,
+      "loss": 0.7417,
+      "step": 8461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.343553435153709e-05,
+      "loss": 0.6801,
+      "step": 8462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.343404022098074e-05,
+      "loss": 0.875,
+      "step": 8463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3432545941299065e-05,
+      "loss": 0.8025,
+      "step": 8464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.34310515125016e-05,
+      "loss": 0.2095,
+      "step": 8465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3429556934597825e-05,
+      "loss": 0.9834,
+      "step": 8466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3428062207597266e-05,
+      "loss": 0.3541,
+      "step": 8467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.342656733150942e-05,
+      "loss": 1.1525,
+      "step": 8468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3425072306343826e-05,
+      "loss": 1.2495,
+      "step": 8469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.342357713210998e-05,
+      "loss": 0.5237,
+      "step": 8470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3422081808817395e-05,
+      "loss": 0.2721,
+      "step": 8471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3420586336475595e-05,
+      "loss": 1.1047,
+      "step": 8472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.341909071509409e-05,
+      "loss": 1.0645,
+      "step": 8473
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3417594944682414e-05,
+      "loss": 1.3329,
+      "step": 8474
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.341609902525006e-05,
+      "loss": 1.47,
+      "step": 8475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3414602956806554e-05,
+      "loss": 1.13,
+      "step": 8476
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.341310673936143e-05,
+      "loss": 0.6966,
+      "step": 8477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.34116103729242e-05,
+      "loss": 1.3827,
+      "step": 8478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.341011385750439e-05,
+      "loss": 1.3341,
+      "step": 8479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.340861719311152e-05,
+      "loss": 0.9619,
+      "step": 8480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.340712037975512e-05,
+      "loss": 1.1482,
+      "step": 8481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.340562341744471e-05,
+      "loss": 0.8814,
+      "step": 8482
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.340412630618981e-05,
+      "loss": 1.1832,
+      "step": 8483
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.340262904599995e-05,
+      "loss": 1.072,
+      "step": 8484
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3401131636884674e-05,
+      "loss": 0.7486,
+      "step": 8485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.339963407885349e-05,
+      "loss": 0.4645,
+      "step": 8486
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.339813637191594e-05,
+      "loss": 0.4199,
+      "step": 8487
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.339663851608155e-05,
+      "loss": 1.0042,
+      "step": 8488
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.339514051135985e-05,
+      "loss": 0.7424,
+      "step": 8489
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.339364235776038e-05,
+      "loss": 1.6581,
+      "step": 8490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3392144055292664e-05,
+      "loss": 0.5362,
+      "step": 8491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.339064560396625e-05,
+      "loss": 0.1856,
+      "step": 8492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3389147003790664e-05,
+      "loss": 0.4335,
+      "step": 8493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.338764825477544e-05,
+      "loss": 1.0073,
+      "step": 8494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3386149356930124e-05,
+      "loss": 0.8742,
+      "step": 8495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3384650310264246e-05,
+      "loss": 0.8585,
+      "step": 8496
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.338315111478735e-05,
+      "loss": 0.313,
+      "step": 8497
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.338165177050897e-05,
+      "loss": 0.4755,
+      "step": 8498
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3380152277438665e-05,
+      "loss": 1.3651,
+      "step": 8499
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.337865263558596e-05,
+      "loss": 0.3689,
+      "step": 8500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3377152844960396e-05,
+      "loss": 1.313,
+      "step": 8501
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.337565290557153e-05,
+      "loss": 0.4701,
+      "step": 8502
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.337415281742891e-05,
+      "loss": 0.4087,
+      "step": 8503
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.337265258054207e-05,
+      "loss": 1.9655,
+      "step": 8504
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3371152194920554e-05,
+      "loss": 0.4534,
+      "step": 8505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.336965166057392e-05,
+      "loss": 0.7515,
+      "step": 8506
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.336815097751171e-05,
+      "loss": 2.0942,
+      "step": 8507
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.336665014574348e-05,
+      "loss": 0.2163,
+      "step": 8508
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.336514916527878e-05,
+      "loss": 0.9064,
+      "step": 8509
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.336364803612715e-05,
+      "loss": 1.9561,
+      "step": 8510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.336214675829816e-05,
+      "loss": 1.4806,
+      "step": 8511
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3360645331801354e-05,
+      "loss": 0.7158,
+      "step": 8512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3359143756646285e-05,
+      "loss": 1.5858,
+      "step": 8513
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.335764203284252e-05,
+      "loss": 0.5258,
+      "step": 8514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.335614016039959e-05,
+      "loss": 0.4125,
+      "step": 8515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.335463813932709e-05,
+      "loss": 0.6966,
+      "step": 8516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.335313596963455e-05,
+      "loss": 0.3971,
+      "step": 8517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3351633651331524e-05,
+      "loss": 1.2828,
+      "step": 8518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3350131184427595e-05,
+      "loss": 0.6863,
+      "step": 8519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.334862856893232e-05,
+      "loss": 0.4915,
+      "step": 8520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.334712580485525e-05,
+      "loss": 0.4495,
+      "step": 8521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3345622892205946e-05,
+      "loss": 0.635,
+      "step": 8522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.334411983099398e-05,
+      "loss": 0.2127,
+      "step": 8523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.334261662122892e-05,
+      "loss": 1.2459,
+      "step": 8524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3341113262920333e-05,
+      "loss": 1.5985,
+      "step": 8525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3339609756077775e-05,
+      "loss": 0.8442,
+      "step": 8526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.333810610071082e-05,
+      "loss": 0.623,
+      "step": 8527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.333660229682904e-05,
+      "loss": 0.5507,
+      "step": 8528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3335098344442e-05,
+      "loss": 0.8626,
+      "step": 8529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.333359424355926e-05,
+      "loss": 0.3284,
+      "step": 8530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.333208999419041e-05,
+      "loss": 0.6669,
+      "step": 8531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.333058559634502e-05,
+      "loss": 1.1455,
+      "step": 8532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3329081050032656e-05,
+      "loss": 1.36,
+      "step": 8533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.332757635526289e-05,
+      "loss": 0.9497,
+      "step": 8534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3326071512045306e-05,
+      "loss": 1.1202,
+      "step": 8535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3324566520389485e-05,
+      "loss": 1.2617,
+      "step": 8536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3323061380304984e-05,
+      "loss": 1.0476,
+      "step": 8537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.33215560918014e-05,
+      "loss": 1.1937,
+      "step": 8538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.33200506548883e-05,
+      "loss": 1.189,
+      "step": 8539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3318545069575264e-05,
+      "loss": 1.1352,
+      "step": 8540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.331703933587189e-05,
+      "loss": 1.0038,
+      "step": 8541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.331553345378774e-05,
+      "loss": 0.5976,
+      "step": 8542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.331402742333241e-05,
+      "loss": 0.7877,
+      "step": 8543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.331252124451547e-05,
+      "loss": 0.3272,
+      "step": 8544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.331101491734653e-05,
+      "loss": 0.3159,
+      "step": 8545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.330950844183514e-05,
+      "loss": 1.4679,
+      "step": 8546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3308001817990926e-05,
+      "loss": 0.2675,
+      "step": 8547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3306495045823434e-05,
+      "loss": 0.6838,
+      "step": 8548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.330498812534229e-05,
+      "loss": 1.7569,
+      "step": 8549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.330348105655706e-05,
+      "loss": 0.4169,
+      "step": 8550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.330197383947734e-05,
+      "loss": 0.9801,
+      "step": 8551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.330046647411272e-05,
+      "loss": 0.8416,
+      "step": 8552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.32989589604728e-05,
+      "loss": 0.235,
+      "step": 8553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.329745129856717e-05,
+      "loss": 1.1779,
+      "step": 8554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.329594348840542e-05,
+      "loss": 1.1538,
+      "step": 8555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.329443552999715e-05,
+      "loss": 0.9087,
+      "step": 8556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.329292742335195e-05,
+      "loss": 0.7772,
+      "step": 8557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3291419168479435e-05,
+      "loss": 0.3254,
+      "step": 8558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3289910765389176e-05,
+      "loss": 0.7375,
+      "step": 8559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.328840221409079e-05,
+      "loss": 0.836,
+      "step": 8560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.328689351459387e-05,
+      "loss": 0.2472,
+      "step": 8561
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.328538466690801e-05,
+      "loss": 0.5237,
+      "step": 8562
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.328387567104282e-05,
+      "loss": 0.9084,
+      "step": 8563
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.328236652700792e-05,
+      "loss": 1.0607,
+      "step": 8564
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.328085723481289e-05,
+      "loss": 0.3572,
+      "step": 8565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3279347794467335e-05,
+      "loss": 0.7411,
+      "step": 8566
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3277838205980866e-05,
+      "loss": 0.6567,
+      "step": 8567
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.32763284693631e-05,
+      "loss": 0.4957,
+      "step": 8568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.327481858462363e-05,
+      "loss": 0.881,
+      "step": 8569
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.327330855177206e-05,
+      "loss": 1.0526,
+      "step": 8570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.327179837081802e-05,
+      "loss": 0.2502,
+      "step": 8571
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3270288041771095e-05,
+      "loss": 0.9869,
+      "step": 8572
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.326877756464092e-05,
+      "loss": 0.7086,
+      "step": 8573
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.326726693943709e-05,
+      "loss": 0.8163,
+      "step": 8574
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.326575616616923e-05,
+      "loss": 0.7865,
+      "step": 8575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.326424524484695e-05,
+      "loss": 0.6772,
+      "step": 8576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3262734175479864e-05,
+      "loss": 1.1853,
+      "step": 8577
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3261222958077575e-05,
+      "loss": 0.4949,
+      "step": 8578
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.325971159264973e-05,
+      "loss": 0.4286,
+      "step": 8579
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3258200079205906e-05,
+      "loss": 0.9415,
+      "step": 8580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3256688417755764e-05,
+      "loss": 1.3172,
+      "step": 8581
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.32551766083089e-05,
+      "loss": 0.5207,
+      "step": 8582
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.325366465087493e-05,
+      "loss": 1.6625,
+      "step": 8583
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.32521525454635e-05,
+      "loss": 1.1126,
+      "step": 8584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3250640292084205e-05,
+      "loss": 0.9775,
+      "step": 8585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.324912789074668e-05,
+      "loss": 1.1991,
+      "step": 8586
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.324761534146056e-05,
+      "loss": 0.2263,
+      "step": 8587
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.324610264423544e-05,
+      "loss": 1.079,
+      "step": 8588
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.324458979908099e-05,
+      "loss": 0.2535,
+      "step": 8589
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.32430768060068e-05,
+      "loss": 0.3387,
+      "step": 8590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.324156366502251e-05,
+      "loss": 0.5898,
+      "step": 8591
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3240050376137755e-05,
+      "loss": 0.3619,
+      "step": 8592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.323853693936216e-05,
+      "loss": 0.5719,
+      "step": 8593
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3237023354705355e-05,
+      "loss": 0.8759,
+      "step": 8594
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.323550962217697e-05,
+      "loss": 0.9666,
+      "step": 8595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.323399574178664e-05,
+      "loss": 0.589,
+      "step": 8596
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3232481713544014e-05,
+      "loss": 1.2946,
+      "step": 8597
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.32309675374587e-05,
+      "loss": 1.2963,
+      "step": 8598
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.322945321354035e-05,
+      "loss": 0.4245,
+      "step": 8599
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3227938741798593e-05,
+      "loss": 0.381,
+      "step": 8600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.322642412224307e-05,
+      "loss": 0.0991,
+      "step": 8601
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3224909354883424e-05,
+      "loss": 0.3615,
+      "step": 8602
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.322339443972929e-05,
+      "loss": 1.6935,
+      "step": 8603
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.322187937679031e-05,
+      "loss": 0.5715,
+      "step": 8604
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3220364166076115e-05,
+      "loss": 1.1057,
+      "step": 8605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3218848807596365e-05,
+      "loss": 1.182,
+      "step": 8606
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3217333301360695e-05,
+      "loss": 0.9134,
+      "step": 8607
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3215817647378736e-05,
+      "loss": 0.5385,
+      "step": 8608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.321430184566016e-05,
+      "loss": 1.5268,
+      "step": 8609
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3212785896214586e-05,
+      "loss": 1.2683,
+      "step": 8610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.321126979905168e-05,
+      "loss": 0.6503,
+      "step": 8611
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3209753554181075e-05,
+      "loss": 1.0115,
+      "step": 8612
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.320823716161244e-05,
+      "loss": 0.9253,
+      "step": 8613
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3206720621355406e-05,
+      "loss": 1.1875,
+      "step": 8614
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.320520393341963e-05,
+      "loss": 0.8763,
+      "step": 8615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.3203687097814766e-05,
+      "loss": 0.4326,
+      "step": 8616
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.320217011455046e-05,
+      "loss": 1.5884,
+      "step": 8617
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.320065298363638e-05,
+      "loss": 1.3745,
+      "step": 8618
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.319913570508216e-05,
+      "loss": 1.0333,
+      "step": 8619
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3197618278897466e-05,
+      "loss": 1.3999,
+      "step": 8620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.319610070509196e-05,
+      "loss": 0.3954,
+      "step": 8621
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3194582983675284e-05,
+      "loss": 0.6648,
+      "step": 8622
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.319306511465712e-05,
+      "loss": 1.3404,
+      "step": 8623
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.31915470980471e-05,
+      "loss": 0.8706,
+      "step": 8624
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.31900289338549e-05,
+      "loss": 0.3933,
+      "step": 8625
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.318851062209019e-05,
+      "loss": 1.5368,
+      "step": 8626
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.318699216276261e-05,
+      "loss": 0.2797,
+      "step": 8627
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3185473555881825e-05,
+      "loss": 0.379,
+      "step": 8628
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.318395480145752e-05,
+      "loss": 0.3192,
+      "step": 8629
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.318243589949934e-05,
+      "loss": 1.2887,
+      "step": 8630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.318091685001695e-05,
+      "loss": 0.9943,
+      "step": 8631
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.317939765302003e-05,
+      "loss": 1.5074,
+      "step": 8632
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.317787830851825e-05,
+      "loss": 0.995,
+      "step": 8633
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.317635881652126e-05,
+      "loss": 1.1952,
+      "step": 8634
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3174839177038734e-05,
+      "loss": 1.3277,
+      "step": 8635
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3173319390080356e-05,
+      "loss": 0.6386,
+      "step": 8636
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.317179945565579e-05,
+      "loss": 1.0686,
+      "step": 8637
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3170279373774705e-05,
+      "loss": 1.4648,
+      "step": 8638
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.316875914444678e-05,
+      "loss": 0.5397,
+      "step": 8639
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.316723876768168e-05,
+      "loss": 1.1908,
+      "step": 8640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.316571824348909e-05,
+      "loss": 0.834,
+      "step": 8641
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.316419757187868e-05,
+      "loss": 0.7619,
+      "step": 8642
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.316267675286012e-05,
+      "loss": 0.992,
+      "step": 8643
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.316115578644311e-05,
+      "loss": 0.7952,
+      "step": 8644
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3159634672637305e-05,
+      "loss": 0.9075,
+      "step": 8645
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.31581134114524e-05,
+      "loss": 0.4979,
+      "step": 8646
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.315659200289807e-05,
+      "loss": 0.6953,
+      "step": 8647
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3155070446984e-05,
+      "loss": 0.7369,
+      "step": 8648
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3153548743719874e-05,
+      "loss": 0.1854,
+      "step": 8649
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.315202689311537e-05,
+      "loss": 1.2053,
+      "step": 8650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3150504895180167e-05,
+      "loss": 1.3465,
+      "step": 8651
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.314898274992396e-05,
+      "loss": 1.0665,
+      "step": 8652
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.314746045735644e-05,
+      "loss": 0.7616,
+      "step": 8653
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.314593801748728e-05,
+      "loss": 1.663,
+      "step": 8654
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.314441543032618e-05,
+      "loss": 0.7361,
+      "step": 8655
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3142892695882825e-05,
+      "loss": 0.5449,
+      "step": 8656
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3141369814166904e-05,
+      "loss": 1.3313,
+      "step": 8657
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3139846785188105e-05,
+      "loss": 1.1283,
+      "step": 8658
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.313832360895613e-05,
+      "loss": 1.0833,
+      "step": 8659
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.313680028548066e-05,
+      "loss": 1.311,
+      "step": 8660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.31352768147714e-05,
+      "loss": 0.377,
+      "step": 8661
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.313375319683804e-05,
+      "loss": 0.8984,
+      "step": 8662
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.313222943169027e-05,
+      "loss": 0.5059,
+      "step": 8663
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.313070551933779e-05,
+      "loss": 1.468,
+      "step": 8664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.312918145979031e-05,
+      "loss": 1.0401,
+      "step": 8665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.312765725305751e-05,
+      "loss": 0.3747,
+      "step": 8666
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3126132899149096e-05,
+      "loss": 1.2767,
+      "step": 8667
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.312460839807477e-05,
+      "loss": 0.4348,
+      "step": 8668
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.312308374984424e-05,
+      "loss": 0.6982,
+      "step": 8669
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.31215589544672e-05,
+      "loss": 0.5766,
+      "step": 8670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.312003401195335e-05,
+      "loss": 1.2753,
+      "step": 8671
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3118508922312396e-05,
+      "loss": 1.2208,
+      "step": 8672
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.311698368555406e-05,
+      "loss": 0.9879,
+      "step": 8673
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.311545830168802e-05,
+      "loss": 1.0936,
+      "step": 8674
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3113932770724e-05,
+      "loss": 1.7386,
+      "step": 8675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3112407092671716e-05,
+      "loss": 0.9846,
+      "step": 8676
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3110881267540847e-05,
+      "loss": 1.3518,
+      "step": 8677
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.310935529534114e-05,
+      "loss": 1.1701,
+      "step": 8678
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.310782917608227e-05,
+      "loss": 0.5285,
+      "step": 8679
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.310630290977398e-05,
+      "loss": 0.5269,
+      "step": 8680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.310477649642596e-05,
+      "loss": 1.0488,
+      "step": 8681
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3103249936047935e-05,
+      "loss": 0.8648,
+      "step": 8682
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3101723228649615e-05,
+      "loss": 0.989,
+      "step": 8683
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.310019637424072e-05,
+      "loss": 0.2363,
+      "step": 8684
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.309866937283095e-05,
+      "loss": 0.5448,
+      "step": 8685
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.309714222443006e-05,
+      "loss": 1.0885,
+      "step": 8686
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.309561492904772e-05,
+      "loss": 0.3706,
+      "step": 8687
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.309408748669369e-05,
+      "loss": 1.4526,
+      "step": 8688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.309255989737767e-05,
+      "loss": 1.5615,
+      "step": 8689
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3091032161109364e-05,
+      "loss": 1.2609,
+      "step": 8690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.308950427789853e-05,
+      "loss": 0.8662,
+      "step": 8691
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.308797624775487e-05,
+      "loss": 1.0346,
+      "step": 8692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3086448070688117e-05,
+      "loss": 0.9806,
+      "step": 8693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.308491974670798e-05,
+      "loss": 0.6534,
+      "step": 8694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3083391275824214e-05,
+      "loss": 0.6255,
+      "step": 8695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3081862658046514e-05,
+      "loss": 0.2166,
+      "step": 8696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.308033389338463e-05,
+      "loss": 1.6356,
+      "step": 8697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.307880498184827e-05,
+      "loss": 2.0639,
+      "step": 8698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3077275923447183e-05,
+      "loss": 1.4723,
+      "step": 8699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3075746718191094e-05,
+      "loss": 0.575,
+      "step": 8700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.307421736608972e-05,
+      "loss": 0.2657,
+      "step": 8701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.307268786715281e-05,
+      "loss": 1.5434,
+      "step": 8702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.307115822139009e-05,
+      "loss": 1.3821,
+      "step": 8703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.306962842881129e-05,
+      "loss": 1.7784,
+      "step": 8704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.306809848942617e-05,
+      "loss": 0.9313,
+      "step": 8705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.306656840324443e-05,
+      "loss": 1.4978,
+      "step": 8706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.306503817027583e-05,
+      "loss": 0.8073,
+      "step": 8707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.30635077905301e-05,
+      "loss": 1.1686,
+      "step": 8708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.306197726401698e-05,
+      "loss": 0.9419,
+      "step": 8709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.306044659074622e-05,
+      "loss": 0.5803,
+      "step": 8710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.305891577072754e-05,
+      "loss": 0.3925,
+      "step": 8711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.305738480397069e-05,
+      "loss": 0.1291,
+      "step": 8712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.3055853690485415e-05,
+      "loss": 1.4934,
+      "step": 8713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.305432243028146e-05,
+      "loss": 1.0401,
+      "step": 8714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.305279102336857e-05,
+      "loss": 0.8786,
+      "step": 8715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.305125946975648e-05,
+      "loss": 0.8793,
+      "step": 8716
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3049727769454957e-05,
+      "loss": 0.4525,
+      "step": 8717
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3048195922473724e-05,
+      "loss": 0.6013,
+      "step": 8718
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3046663928822544e-05,
+      "loss": 0.5719,
+      "step": 8719
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3045131788511166e-05,
+      "loss": 0.9392,
+      "step": 8720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.304359950154933e-05,
+      "loss": 0.6069,
+      "step": 8721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.304206706794679e-05,
+      "loss": 1.7225,
+      "step": 8722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.304053448771331e-05,
+      "loss": 0.4662,
+      "step": 8723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.303900176085863e-05,
+      "loss": 1.3998,
+      "step": 8724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.30374688873925e-05,
+      "loss": 1.3371,
+      "step": 8725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.303593586732469e-05,
+      "loss": 1.331,
+      "step": 8726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3034402700664956e-05,
+      "loss": 0.5971,
+      "step": 8727
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.303286938742304e-05,
+      "loss": 0.4752,
+      "step": 8728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3031335927608695e-05,
+      "loss": 0.7533,
+      "step": 8729
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.302980232123169e-05,
+      "loss": 0.3201,
+      "step": 8730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3028268568301795e-05,
+      "loss": 0.5268,
+      "step": 8731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.302673466882875e-05,
+      "loss": 0.6059,
+      "step": 8732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.302520062282233e-05,
+      "loss": 1.0474,
+      "step": 8733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.30236664302923e-05,
+      "loss": 0.4715,
+      "step": 8734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.30221320912484e-05,
+      "loss": 1.9104,
+      "step": 8735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.302059760570042e-05,
+      "loss": 1.1856,
+      "step": 8736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.301906297365811e-05,
+      "loss": 1.4771,
+      "step": 8737
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3017528195131246e-05,
+      "loss": 0.2806,
+      "step": 8738
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3015993270129584e-05,
+      "loss": 0.7656,
+      "step": 8739
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3014458198662894e-05,
+      "loss": 1.089,
+      "step": 8740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3012922980740956e-05,
+      "loss": 0.298,
+      "step": 8741
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3011387616373526e-05,
+      "loss": 1.2168,
+      "step": 8742
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3009852105570376e-05,
+      "loss": 0.6159,
+      "step": 8743
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3008316448341284e-05,
+      "loss": 1.5137,
+      "step": 8744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.3006780644696015e-05,
+      "loss": 0.7687,
+      "step": 8745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.300524469464435e-05,
+      "loss": 1.8436,
+      "step": 8746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.300370859819606e-05,
+      "loss": 0.2758,
+      "step": 8747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.300217235536092e-05,
+      "loss": 0.2511,
+      "step": 8748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.30006359661487e-05,
+      "loss": 1.0109,
+      "step": 8749
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2999099430569184e-05,
+      "loss": 1.0968,
+      "step": 8750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.299756274863215e-05,
+      "loss": 0.7085,
+      "step": 8751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2996025920347376e-05,
+      "loss": 1.411,
+      "step": 8752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.299448894572464e-05,
+      "loss": 1.0465,
+      "step": 8753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.299295182477373e-05,
+      "loss": 1.1579,
+      "step": 8754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2991414557504415e-05,
+      "loss": 0.4207,
+      "step": 8755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2989877143926486e-05,
+      "loss": 0.1477,
+      "step": 8756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.298833958404972e-05,
+      "loss": 1.5197,
+      "step": 8757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2986801877883915e-05,
+      "loss": 1.1373,
+      "step": 8758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.298526402543884e-05,
+      "loss": 0.4349,
+      "step": 8759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.298372602672429e-05,
+      "loss": 1.3458,
+      "step": 8760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2982187881750056e-05,
+      "loss": 0.4663,
+      "step": 8761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.298064959052592e-05,
+      "loss": 1.2405,
+      "step": 8762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.297911115306167e-05,
+      "loss": 0.7087,
+      "step": 8763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.29775725693671e-05,
+      "loss": 0.3902,
+      "step": 8764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2976033839451995e-05,
+      "loss": 0.5532,
+      "step": 8765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2974494963326156e-05,
+      "loss": 1.1318,
+      "step": 8766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.297295594099937e-05,
+      "loss": 0.4861,
+      "step": 8767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2971416772481427e-05,
+      "loss": 1.3855,
+      "step": 8768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2969877457782134e-05,
+      "loss": 0.8064,
+      "step": 8769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2968337996911274e-05,
+      "loss": 0.2755,
+      "step": 8770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.296679838987865e-05,
+      "loss": 1.2272,
+      "step": 8771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.296525863669406e-05,
+      "loss": 0.7966,
+      "step": 8772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.29637187373673e-05,
+      "loss": 0.9946,
+      "step": 8773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2962178691908166e-05,
+      "loss": 1.5824,
+      "step": 8774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.296063850032647e-05,
+      "loss": 1.0706,
+      "step": 8775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.295909816263199e-05,
+      "loss": 0.9574,
+      "step": 8776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.295755767883456e-05,
+      "loss": 0.4045,
+      "step": 8777
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2956017048943955e-05,
+      "loss": 0.3512,
+      "step": 8778
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.295447627296999e-05,
+      "loss": 0.788,
+      "step": 8779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.295293535092248e-05,
+      "loss": 0.6182,
+      "step": 8780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2951394282811216e-05,
+      "loss": 1.3137,
+      "step": 8781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2949853068646014e-05,
+      "loss": 0.3243,
+      "step": 8782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.294831170843667e-05,
+      "loss": 0.5123,
+      "step": 8783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2946770202193e-05,
+      "loss": 0.5667,
+      "step": 8784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2945228549924824e-05,
+      "loss": 0.9874,
+      "step": 8785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.294368675164194e-05,
+      "loss": 0.9727,
+      "step": 8786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.294214480735415e-05,
+      "loss": 0.7047,
+      "step": 8787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.294060271707129e-05,
+      "loss": 1.0119,
+      "step": 8788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.293906048080316e-05,
+      "loss": 0.8232,
+      "step": 8789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2937518098559577e-05,
+      "loss": 0.4567,
+      "step": 8790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.293597557035034e-05,
+      "loss": 1.1119,
+      "step": 8791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2934432896185294e-05,
+      "loss": 1.1228,
+      "step": 8792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.293289007607424e-05,
+      "loss": 0.5632,
+      "step": 8793
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.293134711002699e-05,
+      "loss": 1.4215,
+      "step": 8794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.292980399805338e-05,
+      "loss": 0.6716,
+      "step": 8795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.292826074016322e-05,
+      "loss": 0.643,
+      "step": 8796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.292671733636633e-05,
+      "loss": 1.5089,
+      "step": 8797
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2925173786672524e-05,
+      "loss": 0.7985,
+      "step": 8798
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.292363009109164e-05,
+      "loss": 1.2015,
+      "step": 8799
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.29220862496335e-05,
+      "loss": 1.1374,
+      "step": 8800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.292054226230792e-05,
+      "loss": 1.3312,
+      "step": 8801
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.291899812912472e-05,
+      "loss": 0.9082,
+      "step": 8802
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.291745385009374e-05,
+      "loss": 1.1349,
+      "step": 8803
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.291590942522481e-05,
+      "loss": 0.951,
+      "step": 8804
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.291436485452773e-05,
+      "loss": 0.5819,
+      "step": 8805
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2912820138012364e-05,
+      "loss": 0.7954,
+      "step": 8806
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.2911275275688524e-05,
+      "loss": 0.8069,
+      "step": 8807
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.290973026756605e-05,
+      "loss": 0.4893,
+      "step": 8808
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.290818511365476e-05,
+      "loss": 0.7576,
+      "step": 8809
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.29066398139645e-05,
+      "loss": 1.405,
+      "step": 8810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.290509436850509e-05,
+      "loss": 0.5089,
+      "step": 8811
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.290354877728639e-05,
+      "loss": 1.2916,
+      "step": 8812
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.290200304031821e-05,
+      "loss": 0.399,
+      "step": 8813
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.290045715761039e-05,
+      "loss": 0.8151,
+      "step": 8814
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2898911129172775e-05,
+      "loss": 1.0303,
+      "step": 8815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.28973649550152e-05,
+      "loss": 0.4334,
+      "step": 8816
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.28958186351475e-05,
+      "loss": 1.0678,
+      "step": 8817
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.289427216957952e-05,
+      "loss": 0.3638,
+      "step": 8818
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2892725558321114e-05,
+      "loss": 0.2905,
+      "step": 8819
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2891178801382106e-05,
+      "loss": 1.4761,
+      "step": 8820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2889631898772336e-05,
+      "loss": 0.3012,
+      "step": 8821
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.288808485050167e-05,
+      "loss": 1.5339,
+      "step": 8822
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.288653765657993e-05,
+      "loss": 1.4925,
+      "step": 8823
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.288499031701698e-05,
+      "loss": 0.6301,
+      "step": 8824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2883442831822645e-05,
+      "loss": 1.7662,
+      "step": 8825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.288189520100679e-05,
+      "loss": 1.3331,
+      "step": 8826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.288034742457926e-05,
+      "loss": 2.3506,
+      "step": 8827
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.28787995025499e-05,
+      "loss": 1.1578,
+      "step": 8828
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.287725143492857e-05,
+      "loss": 1.1053,
+      "step": 8829
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.287570322172511e-05,
+      "loss": 0.9858,
+      "step": 8830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.287415486294938e-05,
+      "loss": 1.1212,
+      "step": 8831
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.287260635861123e-05,
+      "loss": 1.1469,
+      "step": 8832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.287105770872051e-05,
+      "loss": 0.4533,
+      "step": 8833
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.286950891328708e-05,
+      "loss": 1.0604,
+      "step": 8834
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.28679599723208e-05,
+      "loss": 1.4104,
+      "step": 8835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.286641088583152e-05,
+      "loss": 1.0967,
+      "step": 8836
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.28648616538291e-05,
+      "loss": 0.3835,
+      "step": 8837
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2863312276323396e-05,
+      "loss": 1.5914,
+      "step": 8838
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.286176275332427e-05,
+      "loss": 0.2631,
+      "step": 8839
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2860213084841586e-05,
+      "loss": 1.0917,
+      "step": 8840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2858663270885204e-05,
+      "loss": 1.1273,
+      "step": 8841
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.285711331146498e-05,
+      "loss": 0.4735,
+      "step": 8842
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.285556320659079e-05,
+      "loss": 1.4781,
+      "step": 8843
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2854012956272474e-05,
+      "loss": 1.0526,
+      "step": 8844
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.285246256051993e-05,
+      "loss": 0.9881,
+      "step": 8845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2850912019343006e-05,
+      "loss": 0.2752,
+      "step": 8846
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.284936133275156e-05,
+      "loss": 0.6091,
+      "step": 8847
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2847810500755483e-05,
+      "loss": 1.2725,
+      "step": 8848
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2846259523364626e-05,
+      "loss": 1.3827,
+      "step": 8849
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.284470840058886e-05,
+      "loss": 0.9733,
+      "step": 8850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.284315713243807e-05,
+      "loss": 1.6768,
+      "step": 8851
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2841605718922114e-05,
+      "loss": 1.1631,
+      "step": 8852
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.284005416005087e-05,
+      "loss": 0.9271,
+      "step": 8853
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.283850245583421e-05,
+      "loss": 1.356,
+      "step": 8854
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2836950606282005e-05,
+      "loss": 1.2213,
+      "step": 8855
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2835398611404136e-05,
+      "loss": 0.4237,
+      "step": 8856
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.283384647121047e-05,
+      "loss": 0.9159,
+      "step": 8857
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2832294185710906e-05,
+      "loss": 0.8184,
+      "step": 8858
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2830741754915296e-05,
+      "loss": 0.6115,
+      "step": 8859
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.282918917883354e-05,
+      "loss": 0.9838,
+      "step": 8860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.28276364574755e-05,
+      "loss": 0.7537,
+      "step": 8861
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.282608359085107e-05,
+      "loss": 0.5085,
+      "step": 8862
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2824530578970124e-05,
+      "loss": 0.4322,
+      "step": 8863
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2822977421842556e-05,
+      "loss": 1.3102,
+      "step": 8864
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.282142411947823e-05,
+      "loss": 1.3863,
+      "step": 8865
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2819870671887056e-05,
+      "loss": 0.1738,
+      "step": 8866
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2818317079078896e-05,
+      "loss": 0.6667,
+      "step": 8867
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.281676334106366e-05,
+      "loss": 0.8854,
+      "step": 8868
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.281520945785121e-05,
+      "loss": 0.8235,
+      "step": 8869
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2813655429451446e-05,
+      "loss": 0.6613,
+      "step": 8870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.281210125587426e-05,
+      "loss": 1.0052,
+      "step": 8871
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.281054693712954e-05,
+      "loss": 1.2996,
+      "step": 8872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.280899247322718e-05,
+      "loss": 0.529,
+      "step": 8873
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.280743786417706e-05,
+      "loss": 0.7576,
+      "step": 8874
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.280588310998909e-05,
+      "loss": 0.5204,
+      "step": 8875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2804328210673156e-05,
+      "loss": 0.4918,
+      "step": 8876
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.280277316623915e-05,
+      "loss": 0.382,
+      "step": 8877
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.280121797669697e-05,
+      "loss": 1.4663,
+      "step": 8878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2799662642056515e-05,
+      "loss": 0.4323,
+      "step": 8879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2798107162327675e-05,
+      "loss": 0.6713,
+      "step": 8880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2796551537520354e-05,
+      "loss": 0.7243,
+      "step": 8881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.279499576764446e-05,
+      "loss": 0.9232,
+      "step": 8882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2793439852709886e-05,
+      "loss": 0.7566,
+      "step": 8883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.279188379272652e-05,
+      "loss": 1.2127,
+      "step": 8884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.279032758770428e-05,
+      "loss": 1.0054,
+      "step": 8885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.278877123765306e-05,
+      "loss": 0.3863,
+      "step": 8886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.278721474258279e-05,
+      "loss": 0.5101,
+      "step": 8887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2785658102503337e-05,
+      "loss": 1.6634,
+      "step": 8888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.278410131742462e-05,
+      "loss": 0.4758,
+      "step": 8889
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2782544387356554e-05,
+      "loss": 0.7815,
+      "step": 8890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.278098731230904e-05,
+      "loss": 1.1249,
+      "step": 8891
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2779430092292e-05,
+      "loss": 1.1297,
+      "step": 8892
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.277787272731533e-05,
+      "loss": 0.967,
+      "step": 8893
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.277631521738894e-05,
+      "loss": 1.0263,
+      "step": 8894
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2774757562522736e-05,
+      "loss": 1.9381,
+      "step": 8895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.277319976272665e-05,
+      "loss": 1.3704,
+      "step": 8896
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.277164181801058e-05,
+      "loss": 0.8538,
+      "step": 8897
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.277008372838445e-05,
+      "loss": 0.8346,
+      "step": 8898
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.276852549385816e-05,
+      "loss": 0.8937,
+      "step": 8899
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.276696711444164e-05,
+      "loss": 0.59,
+      "step": 8900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2765408590144805e-05,
+      "loss": 1.1348,
+      "step": 8901
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.276384992097756e-05,
+      "loss": 1.6267,
+      "step": 8902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2762291106949846e-05,
+      "loss": 1.3771,
+      "step": 8903
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.276073214807157e-05,
+      "loss": 0.8343,
+      "step": 8904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.275917304435264e-05,
+      "loss": 0.7975,
+      "step": 8905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2757613795803006e-05,
+      "loss": 1.0168,
+      "step": 8906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2756054402432565e-05,
+      "loss": 0.5149,
+      "step": 8907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.275449486425126e-05,
+      "loss": 0.6633,
+      "step": 8908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2752935181269e-05,
+      "loss": 0.2915,
+      "step": 8909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.275137535349572e-05,
+      "loss": 1.4685,
+      "step": 8910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2749815380941334e-05,
+      "loss": 0.2085,
+      "step": 8911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.274825526361578e-05,
+      "loss": 1.0505,
+      "step": 8912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.274669500152899e-05,
+      "loss": 0.3694,
+      "step": 8913
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.274513459469088e-05,
+      "loss": 0.4124,
+      "step": 8914
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2743574043111385e-05,
+      "loss": 1.3168,
+      "step": 8915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.274201334680044e-05,
+      "loss": 1.3299,
+      "step": 8916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.274045250576798e-05,
+      "loss": 0.7681,
+      "step": 8917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.273889152002392e-05,
+      "loss": 1.3976,
+      "step": 8918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.273733038957821e-05,
+      "loss": 0.3105,
+      "step": 8919
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.273576911444078e-05,
+      "loss": 0.5565,
+      "step": 8920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2734207694621566e-05,
+      "loss": 0.4298,
+      "step": 8921
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2732646130130495e-05,
+      "loss": 0.8933,
+      "step": 8922
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.273108442097752e-05,
+      "loss": 1.2773,
+      "step": 8923
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.272952256717257e-05,
+      "loss": 1.6664,
+      "step": 8924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.272796056872558e-05,
+      "loss": 1.5357,
+      "step": 8925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2726398425646496e-05,
+      "loss": 0.4457,
+      "step": 8926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.272483613794526e-05,
+      "loss": 1.2902,
+      "step": 8927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.272327370563181e-05,
+      "loss": 0.5826,
+      "step": 8928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.27217111287161e-05,
+      "loss": 1.4559,
+      "step": 8929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2720148407208055e-05,
+      "loss": 0.9352,
+      "step": 8930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2718585541117626e-05,
+      "loss": 0.3802,
+      "step": 8931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2717022530454767e-05,
+      "loss": 0.2862,
+      "step": 8932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.271545937522941e-05,
+      "loss": 0.8186,
+      "step": 8933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.271389607545153e-05,
+      "loss": 0.4052,
+      "step": 8934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2712332631131047e-05,
+      "loss": 1.3829,
+      "step": 8935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.271076904227791e-05,
+      "loss": 1.2602,
+      "step": 8936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.270920530890208e-05,
+      "loss": 0.2728,
+      "step": 8937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.270764143101352e-05,
+      "loss": 1.5871,
+      "step": 8938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2706077408622155e-05,
+      "loss": 0.7477,
+      "step": 8939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.270451324173796e-05,
+      "loss": 0.6348,
+      "step": 8940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.270294893037087e-05,
+      "loss": 1.0546,
+      "step": 8941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2701384474530855e-05,
+      "loss": 0.9659,
+      "step": 8942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.269981987422787e-05,
+      "loss": 0.2231,
+      "step": 8943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.269825512947186e-05,
+      "loss": 0.3356,
+      "step": 8944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.26966902402728e-05,
+      "loss": 0.8337,
+      "step": 8945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2695125206640624e-05,
+      "loss": 1.0444,
+      "step": 8946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2693560028585306e-05,
+      "loss": 1.0983,
+      "step": 8947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.269199470611681e-05,
+      "loss": 0.6675,
+      "step": 8948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.269042923924509e-05,
+      "loss": 1.2636,
+      "step": 8949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.268886362798012e-05,
+      "loss": 0.6617,
+      "step": 8950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.268729787233184e-05,
+      "loss": 0.5473,
+      "step": 8951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.268573197231023e-05,
+      "loss": 0.7573,
+      "step": 8952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.268416592792526e-05,
+      "loss": 0.8589,
+      "step": 8953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.268259973918688e-05,
+      "loss": 1.3842,
+      "step": 8954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2681033406105064e-05,
+      "loss": 1.0212,
+      "step": 8955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.267946692868978e-05,
+      "loss": 1.5402,
+      "step": 8956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2677900306951e-05,
+      "loss": 0.8937,
+      "step": 8957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2676333540898685e-05,
+      "loss": 0.8236,
+      "step": 8958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.267476663054282e-05,
+      "loss": 0.8544,
+      "step": 8959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.267319957589335e-05,
+      "loss": 0.8008,
+      "step": 8960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.267163237696027e-05,
+      "loss": 0.2774,
+      "step": 8961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.267006503375355e-05,
+      "loss": 0.5305,
+      "step": 8962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2668497546283156e-05,
+      "loss": 0.2403,
+      "step": 8963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.266692991455907e-05,
+      "loss": 1.0951,
+      "step": 8964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.266536213859126e-05,
+      "loss": 0.547,
+      "step": 8965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.266379421838971e-05,
+      "loss": 1.2244,
+      "step": 8966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2662226153964396e-05,
+      "loss": 0.3114,
+      "step": 8967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.26606579453253e-05,
+      "loss": 1.069,
+      "step": 8968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.26590895924824e-05,
+      "loss": 1.4219,
+      "step": 8969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2657521095445664e-05,
+      "loss": 1.433,
+      "step": 8970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2655952454225085e-05,
+      "loss": 0.5106,
+      "step": 8971
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.265438366883065e-05,
+      "loss": 0.2291,
+      "step": 8972
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.265281473927233e-05,
+      "loss": 1.3646,
+      "step": 8973
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2651245665560114e-05,
+      "loss": 1.2571,
+      "step": 8974
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.264967644770399e-05,
+      "loss": 1.7006,
+      "step": 8975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2648107085713935e-05,
+      "loss": 1.3921,
+      "step": 8976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.264653757959995e-05,
+      "loss": 1.2693,
+      "step": 8977
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2644967929372015e-05,
+      "loss": 1.3703,
+      "step": 8978
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.264339813504012e-05,
+      "loss": 0.9159,
+      "step": 8979
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.264182819661424e-05,
+      "loss": 1.2254,
+      "step": 8980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2640258114104394e-05,
+      "loss": 0.2748,
+      "step": 8981
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.263868788752056e-05,
+      "loss": 1.7443,
+      "step": 8982
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2637117516872716e-05,
+      "loss": 1.1554,
+      "step": 8983
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2635547002170874e-05,
+      "loss": 0.4669,
+      "step": 8984
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.263397634342502e-05,
+      "loss": 1.3705,
+      "step": 8985
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2632405540645146e-05,
+      "loss": 1.6229,
+      "step": 8986
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2630834593841264e-05,
+      "loss": 0.7592,
+      "step": 8987
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2629263503023357e-05,
+      "loss": 0.5683,
+      "step": 8988
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.262769226820142e-05,
+      "loss": 0.4523,
+      "step": 8989
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2626120889385464e-05,
+      "loss": 0.8036,
+      "step": 8990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2624549366585476e-05,
+      "loss": 0.2639,
+      "step": 8991
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2622977699811456e-05,
+      "loss": 0.8009,
+      "step": 8992
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2621405889073426e-05,
+      "loss": 0.3436,
+      "step": 8993
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2619833934381374e-05,
+      "loss": 0.9218,
+      "step": 8994
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.26182618357453e-05,
+      "loss": 1.4242,
+      "step": 8995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.261668959317521e-05,
+      "loss": 1.3436,
+      "step": 8996
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.261511720668112e-05,
+      "loss": 1.162,
+      "step": 8997
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.261354467627301e-05,
+      "loss": 0.8268,
+      "step": 8998
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2611972001960916e-05,
+      "loss": 0.9813,
+      "step": 8999
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.261039918375483e-05,
+      "loss": 1.2191,
+      "step": 9000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.260882622166477e-05,
+      "loss": 0.4748,
+      "step": 9001
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.260725311570074e-05,
+      "loss": 0.9121,
+      "step": 9002
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2605679865872755e-05,
+      "loss": 1.4946,
+      "step": 9003
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.260410647219081e-05,
+      "loss": 0.2715,
+      "step": 9004
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.260253293466495e-05,
+      "loss": 1.427,
+      "step": 9005
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.260095925330515e-05,
+      "loss": 0.953,
+      "step": 9006
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2599385428121455e-05,
+      "loss": 0.724,
+      "step": 9007
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2597811459123865e-05,
+      "loss": 1.5922,
+      "step": 9008
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2596237346322395e-05,
+      "loss": 0.4602,
+      "step": 9009
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.2594663089727066e-05,
+      "loss": 1.4475,
+      "step": 9010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.25930886893479e-05,
+      "loss": 1.77,
+      "step": 9011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2591514145194916e-05,
+      "loss": 0.676,
+      "step": 9012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.258993945727813e-05,
+      "loss": 1.5272,
+      "step": 9013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.258836462560756e-05,
+      "loss": 1.1446,
+      "step": 9014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.258678965019322e-05,
+      "loss": 1.2161,
+      "step": 9015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.258521453104517e-05,
+      "loss": 0.8314,
+      "step": 9016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2583639268173385e-05,
+      "loss": 0.3141,
+      "step": 9017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.258206386158792e-05,
+      "loss": 1.3058,
+      "step": 9018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.258048831129879e-05,
+      "loss": 0.2445,
+      "step": 9019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.257891261731602e-05,
+      "loss": 0.968,
+      "step": 9020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2577336779649634e-05,
+      "loss": 0.3787,
+      "step": 9021
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.257576079830967e-05,
+      "loss": 1.5701,
+      "step": 9022
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2574184673306156e-05,
+      "loss": 1.1299,
+      "step": 9023
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.257260840464912e-05,
+      "loss": 1.2656,
+      "step": 9024
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2571031992348585e-05,
+      "loss": 0.6583,
+      "step": 9025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2569455436414585e-05,
+      "loss": 1.8331,
+      "step": 9026
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2567878736857165e-05,
+      "loss": 0.3843,
+      "step": 9027
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.256630189368635e-05,
+      "loss": 0.9113,
+      "step": 9028
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2564724906912174e-05,
+      "loss": 1.4699,
+      "step": 9029
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.256314777654467e-05,
+      "loss": 0.6012,
+      "step": 9030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.256157050259388e-05,
+      "loss": 0.873,
+      "step": 9031
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2559993085069835e-05,
+      "loss": 1.4037,
+      "step": 9032
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.255841552398258e-05,
+      "loss": 1.11,
+      "step": 9033
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.255683781934214e-05,
+      "loss": 0.8857,
+      "step": 9034
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2555259971158574e-05,
+      "loss": 1.2527,
+      "step": 9035
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2553681979441916e-05,
+      "loss": 1.236,
+      "step": 9036
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2552103844202207e-05,
+      "loss": 0.5017,
+      "step": 9037
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2550525565449474e-05,
+      "loss": 1.9464,
+      "step": 9038
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.254894714319379e-05,
+      "loss": 1.4055,
+      "step": 9039
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.254736857744518e-05,
+      "loss": 0.249,
+      "step": 9040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2545789868213695e-05,
+      "loss": 2.212,
+      "step": 9041
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.254421101550938e-05,
+      "loss": 0.5562,
+      "step": 9042
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.254263201934227e-05,
+      "loss": 0.7717,
+      "step": 9043
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.254105287972244e-05,
+      "loss": 0.2851,
+      "step": 9044
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.253947359665992e-05,
+      "loss": 1.6245,
+      "step": 9045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2537894170164766e-05,
+      "loss": 0.6508,
+      "step": 9046
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2536314600247024e-05,
+      "loss": 1.5624,
+      "step": 9047
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.253473488691676e-05,
+      "loss": 1.203,
+      "step": 9048
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2533155030184e-05,
+      "loss": 1.5275,
+      "step": 9049
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.253157503005883e-05,
+      "loss": 0.4007,
+      "step": 9050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.252999488655127e-05,
+      "loss": 0.7301,
+      "step": 9051
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2528414599671403e-05,
+      "loss": 1.4098,
+      "step": 9052
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2526834169429276e-05,
+      "loss": 1.3062,
+      "step": 9053
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.252525359583495e-05,
+      "loss": 1.1551,
+      "step": 9054
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.252367287889847e-05,
+      "loss": 1.1522,
+      "step": 9055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.252209201862991e-05,
+      "loss": 0.2927,
+      "step": 9056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2520511015039324e-05,
+      "loss": 0.6505,
+      "step": 9057
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.251892986813677e-05,
+      "loss": 1.1236,
+      "step": 9058
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.251734857793231e-05,
+      "loss": 1.2071,
+      "step": 9059
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2515767144436024e-05,
+      "loss": 1.4233,
+      "step": 9060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2514185567657946e-05,
+      "loss": 0.7891,
+      "step": 9061
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.251260384760816e-05,
+      "loss": 0.5789,
+      "step": 9062
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2511021984296734e-05,
+      "loss": 1.9225,
+      "step": 9063
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.250943997773372e-05,
+      "loss": 0.6397,
+      "step": 9064
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.250785782792919e-05,
+      "loss": 1.2291,
+      "step": 9065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.250627553489323e-05,
+      "loss": 0.5315,
+      "step": 9066
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.250469309863588e-05,
+      "loss": 1.2944,
+      "step": 9067
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.250311051916723e-05,
+      "loss": 0.7556,
+      "step": 9068
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.250152779649734e-05,
+      "loss": 0.3066,
+      "step": 9069
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.24999449306363e-05,
+      "loss": 1.4759,
+      "step": 9070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.249836192159416e-05,
+      "loss": 0.5405,
+      "step": 9071
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.249677876938101e-05,
+      "loss": 1.0194,
+      "step": 9072
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2495195474006906e-05,
+      "loss": 0.6526,
+      "step": 9073
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.249361203548195e-05,
+      "loss": 1.2072,
+      "step": 9074
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.24920284538162e-05,
+      "loss": 0.7979,
+      "step": 9075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.249044472901973e-05,
+      "loss": 0.8848,
+      "step": 9076
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2488860861102634e-05,
+      "loss": 0.4019,
+      "step": 9077
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.248727685007498e-05,
+      "loss": 0.4762,
+      "step": 9078
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.248569269594684e-05,
+      "loss": 0.6421,
+      "step": 9079
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2484108398728315e-05,
+      "loss": 1.2451,
+      "step": 9080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.248252395842948e-05,
+      "loss": 0.5422,
+      "step": 9081
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.248093937506042e-05,
+      "loss": 0.9763,
+      "step": 9082
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2479354648631206e-05,
+      "loss": 0.9476,
+      "step": 9083
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2477769779151926e-05,
+      "loss": 1.6218,
+      "step": 9084
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2476184766632675e-05,
+      "loss": 1.0697,
+      "step": 9085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2474599611083535e-05,
+      "loss": 1.0921,
+      "step": 9086
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.24730143125146e-05,
+      "loss": 1.1023,
+      "step": 9087
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2471428870935936e-05,
+      "loss": 1.3467,
+      "step": 9088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.246984328635766e-05,
+      "loss": 1.1369,
+      "step": 9089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.246825755878985e-05,
+      "loss": 1.3417,
+      "step": 9090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2466671688242585e-05,
+      "loss": 0.4622,
+      "step": 9091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.246508567472598e-05,
+      "loss": 0.2983,
+      "step": 9092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.24634995182501e-05,
+      "loss": 1.0136,
+      "step": 9093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.246191321882507e-05,
+      "loss": 1.678,
+      "step": 9094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.246032677646096e-05,
+      "loss": 0.3785,
+      "step": 9095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2458740191167885e-05,
+      "loss": 0.8939,
+      "step": 9096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2457153462955915e-05,
+      "loss": 0.7554,
+      "step": 9097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.245556659183518e-05,
+      "loss": 0.398,
+      "step": 9098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2453979577815755e-05,
+      "loss": 1.487,
+      "step": 9099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.245239242090775e-05,
+      "loss": 0.2675,
+      "step": 9100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2450805121121256e-05,
+      "loss": 0.9447,
+      "step": 9101
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.244921767846637e-05,
+      "loss": 0.4605,
+      "step": 9102
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.244763009295321e-05,
+      "loss": 0.7205,
+      "step": 9103
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.244604236459188e-05,
+      "loss": 0.8534,
+      "step": 9104
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2444454493392473e-05,
+      "loss": 1.0087,
+      "step": 9105
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2442866479365096e-05,
+      "loss": 0.6857,
+      "step": 9106
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2441278322519845e-05,
+      "loss": 0.783,
+      "step": 9107
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2439690022866845e-05,
+      "loss": 0.2813,
+      "step": 9108
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.2438101580416186e-05,
+      "loss": 1.6909,
+      "step": 9109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.243651299517799e-05,
+      "loss": 0.5392,
+      "step": 9110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.243492426716236e-05,
+      "loss": 1.5544,
+      "step": 9111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.24333353963794e-05,
+      "loss": 1.4295,
+      "step": 9112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.243174638283924e-05,
+      "loss": 0.9942,
+      "step": 9113
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.243015722655197e-05,
+      "loss": 1.4108,
+      "step": 9114
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.242856792752772e-05,
+      "loss": 1.4451,
+      "step": 9115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.242697848577659e-05,
+      "loss": 0.5996,
+      "step": 9116
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2425388901308706e-05,
+      "loss": 0.9739,
+      "step": 9117
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.242379917413418e-05,
+      "loss": 0.7195,
+      "step": 9118
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.242220930426312e-05,
+      "loss": 0.3031,
+      "step": 9119
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.242061929170565e-05,
+      "loss": 1.3052,
+      "step": 9120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.241902913647189e-05,
+      "loss": 0.5649,
+      "step": 9121
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.241743883857195e-05,
+      "loss": 0.3519,
+      "step": 9122
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.241584839801596e-05,
+      "loss": 0.6406,
+      "step": 9123
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.241425781481405e-05,
+      "loss": 0.6562,
+      "step": 9124
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.241266708897632e-05,
+      "loss": 0.7615,
+      "step": 9125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.24110762205129e-05,
+      "loss": 0.9924,
+      "step": 9126
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2409485209433923e-05,
+      "loss": 1.4847,
+      "step": 9127
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.240789405574951e-05,
+      "loss": 0.635,
+      "step": 9128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.240630275946977e-05,
+      "loss": 0.4786,
+      "step": 9129
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.240471132060486e-05,
+      "loss": 1.6586,
+      "step": 9130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.240311973916488e-05,
+      "loss": 0.3803,
+      "step": 9131
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.240152801515997e-05,
+      "loss": 1.1271,
+      "step": 9132
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.239993614860026e-05,
+      "loss": 0.8611,
+      "step": 9133
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2398344139495874e-05,
+      "loss": 0.8747,
+      "step": 9134
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.239675198785695e-05,
+      "loss": 0.8829,
+      "step": 9135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2395159693693615e-05,
+      "loss": 0.4235,
+      "step": 9136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.239356725701601e-05,
+      "loss": 1.4292,
+      "step": 9137
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.239197467783425e-05,
+      "loss": 1.1647,
+      "step": 9138
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2390381956158486e-05,
+      "loss": 0.9386,
+      "step": 9139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.238878909199885e-05,
+      "loss": 0.7377,
+      "step": 9140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.238719608536548e-05,
+      "loss": 0.6963,
+      "step": 9141
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2385602936268506e-05,
+      "loss": 0.3261,
+      "step": 9142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.238400964471808e-05,
+      "loss": 0.5432,
+      "step": 9143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.238241621072432e-05,
+      "loss": 0.9908,
+      "step": 9144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.238082263429738e-05,
+      "loss": 0.8207,
+      "step": 9145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2379228915447403e-05,
+      "loss": 0.5277,
+      "step": 9146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.237763505418453e-05,
+      "loss": 0.433,
+      "step": 9147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2376041050518895e-05,
+      "loss": 1.2384,
+      "step": 9148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.237444690446064e-05,
+      "loss": 1.418,
+      "step": 9149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.237285261601993e-05,
+      "loss": 1.3065,
+      "step": 9150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.237125818520689e-05,
+      "loss": 0.2452,
+      "step": 9151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.236966361203167e-05,
+      "loss": 0.5139,
+      "step": 9152
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.236806889650443e-05,
+      "loss": 0.2986,
+      "step": 9153
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.236647403863531e-05,
+      "loss": 1.8616,
+      "step": 9154
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2364879038434445e-05,
+      "loss": 0.848,
+      "step": 9155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2363283895912e-05,
+      "loss": 1.462,
+      "step": 9156
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2361688611078134e-05,
+      "loss": 1.2683,
+      "step": 9157
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.236009318394298e-05,
+      "loss": 0.9099,
+      "step": 9158
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.23584976145167e-05,
+      "loss": 0.7706,
+      "step": 9159
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.235690190280945e-05,
+      "loss": 0.5262,
+      "step": 9160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.235530604883138e-05,
+      "loss": 1.5169,
+      "step": 9161
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2353710052592656e-05,
+      "loss": 1.6796,
+      "step": 9162
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.235211391410341e-05,
+      "loss": 0.4546,
+      "step": 9163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.235051763337382e-05,
+      "loss": 0.178,
+      "step": 9164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.234892121041404e-05,
+      "loss": 0.7998,
+      "step": 9165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2347324645234224e-05,
+      "loss": 0.3213,
+      "step": 9166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.234572793784454e-05,
+      "loss": 2.0099,
+      "step": 9167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2344131088255145e-05,
+      "loss": 1.8673,
+      "step": 9168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.23425340964762e-05,
+      "loss": 1.3644,
+      "step": 9169
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.234093696251786e-05,
+      "loss": 0.7414,
+      "step": 9170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.23393396863903e-05,
+      "loss": 1.1463,
+      "step": 9171
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2337742268103675e-05,
+      "loss": 1.0375,
+      "step": 9172
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.233614470766816e-05,
+      "loss": 1.1955,
+      "step": 9173
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.233454700509392e-05,
+      "loss": 0.4331,
+      "step": 9174
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.233294916039112e-05,
+      "loss": 0.9388,
+      "step": 9175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.233135117356993e-05,
+      "loss": 1.1269,
+      "step": 9176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.232975304464051e-05,
+      "loss": 0.7144,
+      "step": 9177
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.232815477361304e-05,
+      "loss": 1.0065,
+      "step": 9178
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.232655636049768e-05,
+      "loss": 1.4296,
+      "step": 9179
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2324957805304615e-05,
+      "loss": 1.2285,
+      "step": 9180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.232335910804401e-05,
+      "loss": 0.8864,
+      "step": 9181
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.232176026872604e-05,
+      "loss": 0.4105,
+      "step": 9182
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.232016128736088e-05,
+      "loss": 1.6309,
+      "step": 9183
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2318562163958704e-05,
+      "loss": 0.9614,
+      "step": 9184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2316962898529694e-05,
+      "loss": 1.4208,
+      "step": 9185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.231536349108402e-05,
+      "loss": 0.5556,
+      "step": 9186
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.231376394163185e-05,
+      "loss": 1.8698,
+      "step": 9187
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2312164250183384e-05,
+      "loss": 1.2483,
+      "step": 9188
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.23105644167488e-05,
+      "loss": 1.7317,
+      "step": 9189
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2308964441338255e-05,
+      "loss": 0.3493,
+      "step": 9190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.230736432396196e-05,
+      "loss": 1.2244,
+      "step": 9191
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2305764064630076e-05,
+      "loss": 0.7977,
+      "step": 9192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.23041636633528e-05,
+      "loss": 0.4889,
+      "step": 9193
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2302563120140316e-05,
+      "loss": 0.4647,
+      "step": 9194
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.23009624350028e-05,
+      "loss": 1.3975,
+      "step": 9195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.229936160795044e-05,
+      "loss": 0.9286,
+      "step": 9196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2297760638993425e-05,
+      "loss": 0.7366,
+      "step": 9197
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2296159528141946e-05,
+      "loss": 0.499,
+      "step": 9198
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2294558275406194e-05,
+      "loss": 1.4431,
+      "step": 9199
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.229295688079635e-05,
+      "loss": 1.0149,
+      "step": 9200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.229135534432261e-05,
+      "loss": 0.5099,
+      "step": 9201
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2289753665995166e-05,
+      "loss": 0.2506,
+      "step": 9202
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.228815184582421e-05,
+      "loss": 0.7027,
+      "step": 9203
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.228654988381993e-05,
+      "loss": 1.0922,
+      "step": 9204
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.2284947779992526e-05,
+      "loss": 1.4646,
+      "step": 9205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.228334553435219e-05,
+      "loss": 1.5018,
+      "step": 9206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.228174314690913e-05,
+      "loss": 0.4234,
+      "step": 9207
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2280140617673526e-05,
+      "loss": 0.7628,
+      "step": 9208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.227853794665558e-05,
+      "loss": 0.4332,
+      "step": 9209
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2276935133865504e-05,
+      "loss": 0.1316,
+      "step": 9210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.227533217931348e-05,
+      "loss": 0.3806,
+      "step": 9211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.227372908300971e-05,
+      "loss": 1.3854,
+      "step": 9212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.227212584496441e-05,
+      "loss": 0.7921,
+      "step": 9213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2270522465187764e-05,
+      "loss": 1.1903,
+      "step": 9214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.226891894369e-05,
+      "loss": 0.5612,
+      "step": 9215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.22673152804813e-05,
+      "loss": 1.1227,
+      "step": 9216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.226571147557188e-05,
+      "loss": 1.6045,
+      "step": 9217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.226410752897193e-05,
+      "loss": 0.9788,
+      "step": 9218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2262503440691684e-05,
+      "loss": 0.8055,
+      "step": 9219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2260899210741327e-05,
+      "loss": 0.5635,
+      "step": 9220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.225929483913108e-05,
+      "loss": 0.8346,
+      "step": 9221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2257690325871144e-05,
+      "loss": 1.3221,
+      "step": 9222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2256085670971737e-05,
+      "loss": 0.6405,
+      "step": 9223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.225448087444307e-05,
+      "loss": 0.9413,
+      "step": 9224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2252875936295353e-05,
+      "loss": 1.4452,
+      "step": 9225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2251270856538795e-05,
+      "loss": 1.1516,
+      "step": 9226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.224966563518362e-05,
+      "loss": 0.9392,
+      "step": 9227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.224806027224003e-05,
+      "loss": 0.3912,
+      "step": 9228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.224645476771825e-05,
+      "loss": 1.3754,
+      "step": 9229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2244849121628496e-05,
+      "loss": 0.4059,
+      "step": 9230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.224324333398098e-05,
+      "loss": 0.4721,
+      "step": 9231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.224163740478593e-05,
+      "loss": 0.2775,
+      "step": 9232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.224003133405356e-05,
+      "loss": 0.326,
+      "step": 9233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2238425121794095e-05,
+      "loss": 1.5397,
+      "step": 9234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.223681876801774e-05,
+      "loss": 0.9822,
+      "step": 9235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.223521227273473e-05,
+      "loss": 1.1306,
+      "step": 9236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2233605635955306e-05,
+      "loss": 1.4208,
+      "step": 9237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2231998857689664e-05,
+      "loss": 0.9126,
+      "step": 9238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.223039193794803e-05,
+      "loss": 1.2747,
+      "step": 9239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2228784876740656e-05,
+      "loss": 0.4148,
+      "step": 9240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.222717767407774e-05,
+      "loss": 1.1851,
+      "step": 9241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2225570329969526e-05,
+      "loss": 0.7391,
+      "step": 9242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.222396284442623e-05,
+      "loss": 1.4322,
+      "step": 9243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.222235521745809e-05,
+      "loss": 0.5651,
+      "step": 9244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.222074744907535e-05,
+      "loss": 1.3433,
+      "step": 9245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.221913953928821e-05,
+      "loss": 0.9353,
+      "step": 9246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.221753148810692e-05,
+      "loss": 1.8434,
+      "step": 9247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2215923295541714e-05,
+      "loss": 1.5776,
+      "step": 9248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.221431496160283e-05,
+      "loss": 0.57,
+      "step": 9249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.221270648630049e-05,
+      "loss": 0.887,
+      "step": 9250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2211097869644936e-05,
+      "loss": 1.0261,
+      "step": 9251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.22094891116464e-05,
+      "loss": 0.706,
+      "step": 9252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2207880212315135e-05,
+      "loss": 1.3507,
+      "step": 9253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.220627117166137e-05,
+      "loss": 0.9499,
+      "step": 9254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.220466198969534e-05,
+      "loss": 0.9164,
+      "step": 9255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.220305266642728e-05,
+      "loss": 1.5605,
+      "step": 9256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.220144320186744e-05,
+      "loss": 0.7523,
+      "step": 9257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.219983359602607e-05,
+      "loss": 1.183,
+      "step": 9258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.21982238489134e-05,
+      "loss": 0.5825,
+      "step": 9259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.219661396053967e-05,
+      "loss": 0.6448,
+      "step": 9260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.219500393091514e-05,
+      "loss": 1.4241,
+      "step": 9261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.219339376005005e-05,
+      "loss": 0.8386,
+      "step": 9262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2191783447954645e-05,
+      "loss": 1.3191,
+      "step": 9263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2190172994639176e-05,
+      "loss": 1.0695,
+      "step": 9264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.218856240011388e-05,
+      "loss": 0.8367,
+      "step": 9265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.218695166438902e-05,
+      "loss": 1.1173,
+      "step": 9266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2185340787474835e-05,
+      "loss": 1.5143,
+      "step": 9267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.218372976938158e-05,
+      "loss": 0.7593,
+      "step": 9268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.218211861011951e-05,
+      "loss": 1.3815,
+      "step": 9269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.218050730969887e-05,
+      "loss": 0.4972,
+      "step": 9270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.217889586812994e-05,
+      "loss": 0.5378,
+      "step": 9271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.217728428542294e-05,
+      "loss": 1.0166,
+      "step": 9272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2175672561588136e-05,
+      "loss": 0.2729,
+      "step": 9273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2174060696635786e-05,
+      "loss": 1.28,
+      "step": 9274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.217244869057616e-05,
+      "loss": 0.8253,
+      "step": 9275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2170836543419493e-05,
+      "loss": 0.3146,
+      "step": 9276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.216922425517607e-05,
+      "loss": 0.7488,
+      "step": 9277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.216761182585612e-05,
+      "loss": 1.0938,
+      "step": 9278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.216599925546993e-05,
+      "loss": 1.0029,
+      "step": 9279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.216438654402777e-05,
+      "loss": 1.3227,
+      "step": 9280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.216277369153986e-05,
+      "loss": 0.9495,
+      "step": 9281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2161160698016496e-05,
+      "loss": 0.3403,
+      "step": 9282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.215954756346795e-05,
+      "loss": 0.7143,
+      "step": 9283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.215793428790446e-05,
+      "loss": 1.2945,
+      "step": 9284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.215632087133631e-05,
+      "loss": 0.9278,
+      "step": 9285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2154707313773756e-05,
+      "loss": 1.2882,
+      "step": 9286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.215309361522708e-05,
+      "loss": 0.3137,
+      "step": 9287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2151479775706545e-05,
+      "loss": 1.5972,
+      "step": 9288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.214986579522241e-05,
+      "loss": 0.1135,
+      "step": 9289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2148251673784965e-05,
+      "loss": 1.1302,
+      "step": 9290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.214663741140447e-05,
+      "loss": 1.0133,
+      "step": 9291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.21450230080912e-05,
+      "loss": 1.1104,
+      "step": 9292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.214340846385543e-05,
+      "loss": 1.7541,
+      "step": 9293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.2141793778707424e-05,
+      "loss": 0.7414,
+      "step": 9294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.214017895265747e-05,
+      "loss": 0.9572,
+      "step": 9295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.213856398571584e-05,
+      "loss": 0.5402,
+      "step": 9296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.213694887789281e-05,
+      "loss": 0.6889,
+      "step": 9297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.213533362919866e-05,
+      "loss": 1.5859,
+      "step": 9298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.213371823964366e-05,
+      "loss": 0.8493,
+      "step": 9299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.21321027092381e-05,
+      "loss": 0.7474,
+      "step": 9300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.213048703799227e-05,
+      "loss": 1.0385,
+      "step": 9301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.212887122591643e-05,
+      "loss": 0.3738,
+      "step": 9302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.212725527302088e-05,
+      "loss": 0.8126,
+      "step": 9303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.212563917931588e-05,
+      "loss": 1.0017,
+      "step": 9304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.212402294481173e-05,
+      "loss": 0.8813,
+      "step": 9305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.212240656951873e-05,
+      "loss": 1.5942,
+      "step": 9306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.212079005344714e-05,
+      "loss": 1.8198,
+      "step": 9307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2119173396607253e-05,
+      "loss": 1.4598,
+      "step": 9308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.211755659900937e-05,
+      "loss": 0.579,
+      "step": 9309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.211593966066377e-05,
+      "loss": 1.5321,
+      "step": 9310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.211432258158075e-05,
+      "loss": 0.8542,
+      "step": 9311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.211270536177058e-05,
+      "loss": 0.492,
+      "step": 9312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.211108800124357e-05,
+      "loss": 1.1744,
+      "step": 9313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2109470500010004e-05,
+      "loss": 0.8243,
+      "step": 9314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.210785285808018e-05,
+      "loss": 0.9317,
+      "step": 9315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.210623507546439e-05,
+      "loss": 1.0162,
+      "step": 9316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.210461715217295e-05,
+      "loss": 0.5932,
+      "step": 9317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.210299908821611e-05,
+      "loss": 1.2023,
+      "step": 9318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.21013808836042e-05,
+      "loss": 0.5222,
+      "step": 9319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2099762538347505e-05,
+      "loss": 0.5327,
+      "step": 9320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.209814405245633e-05,
+      "loss": 0.5074,
+      "step": 9321
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.209652542594098e-05,
+      "loss": 0.2668,
+      "step": 9322
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.209490665881175e-05,
+      "loss": 1.1127,
+      "step": 9323
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.209328775107893e-05,
+      "loss": 1.019,
+      "step": 9324
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.209166870275284e-05,
+      "loss": 1.017,
+      "step": 9325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.209004951384377e-05,
+      "loss": 0.2418,
+      "step": 9326
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.208843018436202e-05,
+      "loss": 0.3299,
+      "step": 9327
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.208681071431791e-05,
+      "loss": 1.3808,
+      "step": 9328
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.208519110372174e-05,
+      "loss": 0.4408,
+      "step": 9329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.208357135258382e-05,
+      "loss": 1.1576,
+      "step": 9330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.208195146091445e-05,
+      "loss": 1.3646,
+      "step": 9331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.208033142872394e-05,
+      "loss": 1.0283,
+      "step": 9332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2078711256022596e-05,
+      "loss": 1.3717,
+      "step": 9333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.207709094282074e-05,
+      "loss": 0.6106,
+      "step": 9334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.207547048912867e-05,
+      "loss": 0.2562,
+      "step": 9335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2073849894956706e-05,
+      "loss": 1.3189,
+      "step": 9336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2072229160315167e-05,
+      "loss": 1.072,
+      "step": 9337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.207060828521435e-05,
+      "loss": 0.6247,
+      "step": 9338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.206898726966458e-05,
+      "loss": 0.2452,
+      "step": 9339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.206736611367617e-05,
+      "loss": 0.1633,
+      "step": 9340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.206574481725943e-05,
+      "loss": 0.7155,
+      "step": 9341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.206412338042469e-05,
+      "loss": 1.0541,
+      "step": 9342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.206250180318227e-05,
+      "loss": 1.4499,
+      "step": 9343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.206088008554247e-05,
+      "loss": 0.6212,
+      "step": 9344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.205925822751563e-05,
+      "loss": 1.2714,
+      "step": 9345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.205763622911205e-05,
+      "loss": 0.2695,
+      "step": 9346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.205601409034208e-05,
+      "loss": 1.2524,
+      "step": 9347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.205439181121601e-05,
+      "loss": 1.3333,
+      "step": 9348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.205276939174419e-05,
+      "loss": 0.7565,
+      "step": 9349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2051146831936936e-05,
+      "loss": 1.1277,
+      "step": 9350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.204952413180458e-05,
+      "loss": 0.821,
+      "step": 9351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.204790129135742e-05,
+      "loss": 0.5083,
+      "step": 9352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.204627831060582e-05,
+      "loss": 1.6287,
+      "step": 9353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.204465518956008e-05,
+      "loss": 0.765,
+      "step": 9354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.204303192823055e-05,
+      "loss": 1.8056,
+      "step": 9355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2041408526627544e-05,
+      "loss": 1.1256,
+      "step": 9356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.203978498476141e-05,
+      "loss": 1.0763,
+      "step": 9357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.203816130264245e-05,
+      "loss": 1.5271,
+      "step": 9358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2036537480281034e-05,
+      "loss": 0.8029,
+      "step": 9359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.203491351768747e-05,
+      "loss": 0.7394,
+      "step": 9360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.20332894148721e-05,
+      "loss": 0.5542,
+      "step": 9361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.203166517184525e-05,
+      "loss": 0.4019,
+      "step": 9362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.203004078861728e-05,
+      "loss": 1.2052,
+      "step": 9363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2028416265198496e-05,
+      "loss": 0.9673,
+      "step": 9364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.202679160159926e-05,
+      "loss": 0.8838,
+      "step": 9365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.202516679782989e-05,
+      "loss": 1.0275,
+      "step": 9366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.202354185390075e-05,
+      "loss": 1.2125,
+      "step": 9367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.202191676982216e-05,
+      "loss": 0.679,
+      "step": 9368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.202029154560447e-05,
+      "loss": 1.4272,
+      "step": 9369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.201866618125803e-05,
+      "loss": 0.8241,
+      "step": 9370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2017040676793165e-05,
+      "loss": 0.8922,
+      "step": 9371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.201541503222023e-05,
+      "loss": 1.301,
+      "step": 9372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2013789247549577e-05,
+      "loss": 0.9078,
+      "step": 9373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.201216332279153e-05,
+      "loss": 1.1279,
+      "step": 9374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2010537257956465e-05,
+      "loss": 0.4406,
+      "step": 9375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.200891105305471e-05,
+      "loss": 1.4764,
+      "step": 9376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2007284708096615e-05,
+      "loss": 0.9131,
+      "step": 9377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.200565822309253e-05,
+      "loss": 1.1854,
+      "step": 9378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.200403159805281e-05,
+      "loss": 0.5784,
+      "step": 9379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.20024048329878e-05,
+      "loss": 0.4045,
+      "step": 9380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.200077792790785e-05,
+      "loss": 0.6539,
+      "step": 9381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.199915088282334e-05,
+      "loss": 1.2983,
+      "step": 9382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.199752369774458e-05,
+      "loss": 1.6962,
+      "step": 9383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.199589637268196e-05,
+      "loss": 0.7831,
+      "step": 9384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.199426890764582e-05,
+      "loss": 0.3353,
+      "step": 9385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.1992641302646525e-05,
+      "loss": 1.5914,
+      "step": 9386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.199101355769442e-05,
+      "loss": 0.8058,
+      "step": 9387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.198938567279987e-05,
+      "loss": 0.7855,
+      "step": 9388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.198775764797324e-05,
+      "loss": 0.5302,
+      "step": 9389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.198612948322488e-05,
+      "loss": 1.3224,
+      "step": 9390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.198450117856516e-05,
+      "loss": 0.5286,
+      "step": 9391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.1982872734004434e-05,
+      "loss": 0.9811,
+      "step": 9392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.1981244149553076e-05,
+      "loss": 0.8411,
+      "step": 9393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.1979615425221436e-05,
+      "loss": 0.6797,
+      "step": 9394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.197798656101988e-05,
+      "loss": 0.4352,
+      "step": 9395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.1976357556958784e-05,
+      "loss": 1.4353,
+      "step": 9396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.1974728413048514e-05,
+      "loss": 0.3998,
+      "step": 9397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.1973099129299426e-05,
+      "loss": 1.4069,
+      "step": 9398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.1971469705721896e-05,
+      "loss": 0.3133,
+      "step": 9399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.19698401423263e-05,
+      "loss": 0.4594,
+      "step": 9400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.1968210439122983e-05,
+      "loss": 0.8068,
+      "step": 9401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.196658059612233e-05,
+      "loss": 1.0244,
+      "step": 9402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.196495061333473e-05,
+      "loss": 0.6425,
+      "step": 9403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.196332049077053e-05,
+      "loss": 0.5793,
+      "step": 9404
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.196169022844012e-05,
+      "loss": 0.167,
+      "step": 9405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.196005982635386e-05,
+      "loss": 0.8849,
+      "step": 9406
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.195842928452214e-05,
+      "loss": 0.3344,
+      "step": 9407
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.195679860295533e-05,
+      "loss": 0.6725,
+      "step": 9408
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1955167781663806e-05,
+      "loss": 1.0257,
+      "step": 9409
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.195353682065794e-05,
+      "loss": 1.1112,
+      "step": 9410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.195190571994812e-05,
+      "loss": 1.0032,
+      "step": 9411
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.195027447954473e-05,
+      "loss": 0.5755,
+      "step": 9412
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.194864309945814e-05,
+      "loss": 0.8744,
+      "step": 9413
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.194701157969874e-05,
+      "loss": 0.9723,
+      "step": 9414
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.19453799202769e-05,
+      "loss": 0.3438,
+      "step": 9415
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1943748121203e-05,
+      "loss": 1.2005,
+      "step": 9416
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.194211618248745e-05,
+      "loss": 0.7818,
+      "step": 9417
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.194048410414061e-05,
+      "loss": 0.5942,
+      "step": 9418
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.193885188617289e-05,
+      "loss": 1.8797,
+      "step": 9419
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.193721952859465e-05,
+      "loss": 0.7385,
+      "step": 9420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.193558703141629e-05,
+      "loss": 1.1164,
+      "step": 9421
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.193395439464821e-05,
+      "loss": 1.2405,
+      "step": 9422
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.193232161830078e-05,
+      "loss": 0.859,
+      "step": 9423
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.19306887023844e-05,
+      "loss": 0.9709,
+      "step": 9424
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.192905564690946e-05,
+      "loss": 0.634,
+      "step": 9425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.192742245188636e-05,
+      "loss": 0.2272,
+      "step": 9426
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.192578911732547e-05,
+      "loss": 0.8561,
+      "step": 9427
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.19241556432372e-05,
+      "loss": 1.5499,
+      "step": 9428
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.192252202963196e-05,
+      "loss": 0.2306,
+      "step": 9429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.192088827652012e-05,
+      "loss": 1.2188,
+      "step": 9430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.191925438391209e-05,
+      "loss": 1.0327,
+      "step": 9431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.191762035181825e-05,
+      "loss": 1.2183,
+      "step": 9432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.191598618024903e-05,
+      "loss": 1.3315,
+      "step": 9433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.191435186921481e-05,
+      "loss": 0.9823,
+      "step": 9434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1912717418725975e-05,
+      "loss": 1.2583,
+      "step": 9435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.191108282879295e-05,
+      "loss": 0.7431,
+      "step": 9436
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.190944809942613e-05,
+      "loss": 0.637,
+      "step": 9437
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.190781323063592e-05,
+      "loss": 0.8823,
+      "step": 9438
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1906178222432716e-05,
+      "loss": 1.3934,
+      "step": 9439
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.190454307482693e-05,
+      "loss": 0.3935,
+      "step": 9440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.190290778782897e-05,
+      "loss": 0.8117,
+      "step": 9441
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.190127236144923e-05,
+      "loss": 1.1562,
+      "step": 9442
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.189963679569813e-05,
+      "loss": 1.6289,
+      "step": 9443
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.189800109058606e-05,
+      "loss": 1.2711,
+      "step": 9444
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.189636524612345e-05,
+      "loss": 1.6769,
+      "step": 9445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1894729262320705e-05,
+      "loss": 0.72,
+      "step": 9446
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1893093139188235e-05,
+      "loss": 0.77,
+      "step": 9447
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.189145687673644e-05,
+      "loss": 0.5718,
+      "step": 9448
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1889820474975745e-05,
+      "loss": 1.1961,
+      "step": 9449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1888183933916555e-05,
+      "loss": 1.1234,
+      "step": 9450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.188654725356929e-05,
+      "loss": 1.3516,
+      "step": 9451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1884910433944373e-05,
+      "loss": 0.815,
+      "step": 9452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1883273475052206e-05,
+      "loss": 0.9356,
+      "step": 9453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.188163637690321e-05,
+      "loss": 0.1239,
+      "step": 9454
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1879999139507806e-05,
+      "loss": 1.1321,
+      "step": 9455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.187836176287641e-05,
+      "loss": 0.5891,
+      "step": 9456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.187672424701944e-05,
+      "loss": 0.4845,
+      "step": 9457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.187508659194733e-05,
+      "loss": 0.5946,
+      "step": 9458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.187344879767048e-05,
+      "loss": 1.3222,
+      "step": 9459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.187181086419933e-05,
+      "loss": 0.4406,
+      "step": 9460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1870172791544293e-05,
+      "loss": 1.5507,
+      "step": 9461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.186853457971579e-05,
+      "loss": 0.6741,
+      "step": 9462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.186689622872426e-05,
+      "loss": 1.6075,
+      "step": 9463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.186525773858012e-05,
+      "loss": 0.4757,
+      "step": 9464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.186361910929379e-05,
+      "loss": 1.44,
+      "step": 9465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1861980340875715e-05,
+      "loss": 0.4532,
+      "step": 9466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.186034143333631e-05,
+      "loss": 0.257,
+      "step": 9467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.185870238668601e-05,
+      "loss": 1.1001,
+      "step": 9468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1857063200935244e-05,
+      "loss": 1.3327,
+      "step": 9469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.185542387609444e-05,
+      "loss": 0.3364,
+      "step": 9470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1853784412174036e-05,
+      "loss": 1.1547,
+      "step": 9471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.185214480918446e-05,
+      "loss": 0.4335,
+      "step": 9472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1850505067136144e-05,
+      "loss": 1.2574,
+      "step": 9473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1848865186039527e-05,
+      "loss": 1.3169,
+      "step": 9474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.184722516590504e-05,
+      "loss": 0.3332,
+      "step": 9475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.184558500674313e-05,
+      "loss": 0.2686,
+      "step": 9476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.184394470856423e-05,
+      "loss": 1.2821,
+      "step": 9477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.184230427137877e-05,
+      "loss": 0.5542,
+      "step": 9478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1840663695197194e-05,
+      "loss": 1.2793,
+      "step": 9479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1839022980029944e-05,
+      "loss": 1.2526,
+      "step": 9480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.183738212588747e-05,
+      "loss": 0.5104,
+      "step": 9481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.183574113278019e-05,
+      "loss": 0.9514,
+      "step": 9482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.183410000071857e-05,
+      "loss": 0.9115,
+      "step": 9483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.183245872971304e-05,
+      "loss": 0.9116,
+      "step": 9484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1830817319774045e-05,
+      "loss": 1.3409,
+      "step": 9485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.182917577091203e-05,
+      "loss": 0.9508,
+      "step": 9486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1827534083137454e-05,
+      "loss": 1.7372,
+      "step": 9487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1825892256460746e-05,
+      "loss": 0.8797,
+      "step": 9488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1824250290892366e-05,
+      "loss": 0.2964,
+      "step": 9489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.182260818644276e-05,
+      "loss": 0.4564,
+      "step": 9490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.182096594312238e-05,
+      "loss": 0.4207,
+      "step": 9491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.181932356094167e-05,
+      "loss": 1.5769,
+      "step": 9492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.181768103991109e-05,
+      "loss": 0.2515,
+      "step": 9493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.181603838004109e-05,
+      "loss": 1.2359,
+      "step": 9494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1814395581342106e-05,
+      "loss": 0.7672,
+      "step": 9495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.181275264382462e-05,
+      "loss": 0.2322,
+      "step": 9496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.181110956749908e-05,
+      "loss": 1.5011,
+      "step": 9497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.180946635237592e-05,
+      "loss": 0.4165,
+      "step": 9498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.180782299846562e-05,
+      "loss": 0.7958,
+      "step": 9499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1806179505778634e-05,
+      "loss": 0.9906,
+      "step": 9500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.180453587432541e-05,
+      "loss": 0.363,
+      "step": 9501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.180289210411642e-05,
+      "loss": 0.5523,
+      "step": 9502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.180124819516212e-05,
+      "loss": 0.3675,
+      "step": 9503
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.179960414747296e-05,
+      "loss": 1.8386,
+      "step": 9504
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.179795996105942e-05,
+      "loss": 1.118,
+      "step": 9505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.179631563593196e-05,
+      "loss": 0.7019,
+      "step": 9506
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.179467117210103e-05,
+      "loss": 1.457,
+      "step": 9507
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.179302656957711e-05,
+      "loss": 1.6583,
+      "step": 9508
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.179138182837066e-05,
+      "loss": 0.251,
+      "step": 9509
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.178973694849214e-05,
+      "loss": 1.1674,
+      "step": 9510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1788091929952026e-05,
+      "loss": 0.3161,
+      "step": 9511
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1786446772760784e-05,
+      "loss": 1.4493,
+      "step": 9512
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.178480147692888e-05,
+      "loss": 0.825,
+      "step": 9513
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.178315604246679e-05,
+      "loss": 0.6046,
+      "step": 9514
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.178151046938498e-05,
+      "loss": 0.7481,
+      "step": 9515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1779864757693926e-05,
+      "loss": 1.2064,
+      "step": 9516
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.17782189074041e-05,
+      "loss": 0.6017,
+      "step": 9517
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.177657291852597e-05,
+      "loss": 0.89,
+      "step": 9518
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.177492679107001e-05,
+      "loss": 0.5963,
+      "step": 9519
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.177328052504671e-05,
+      "loss": 1.0645,
+      "step": 9520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.177163412046653e-05,
+      "loss": 1.7362,
+      "step": 9521
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1769987577339946e-05,
+      "loss": 1.2048,
+      "step": 9522
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1768340895677446e-05,
+      "loss": 1.4417,
+      "step": 9523
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.176669407548951e-05,
+      "loss": 0.5712,
+      "step": 9524
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.176504711678662e-05,
+      "loss": 0.6358,
+      "step": 9525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.176340001957924e-05,
+      "loss": 0.9451,
+      "step": 9526
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1761752783877864e-05,
+      "loss": 0.5711,
+      "step": 9527
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.176010540969298e-05,
+      "loss": 0.212,
+      "step": 9528
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.175845789703505e-05,
+      "loss": 1.4155,
+      "step": 9529
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.175681024591457e-05,
+      "loss": 1.3472,
+      "step": 9530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1755162456342034e-05,
+      "loss": 0.2936,
+      "step": 9531
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.175351452832792e-05,
+      "loss": 0.9938,
+      "step": 9532
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.175186646188272e-05,
+      "loss": 0.5665,
+      "step": 9533
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.17502182570169e-05,
+      "loss": 0.3097,
+      "step": 9534
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1748569913740976e-05,
+      "loss": 0.9427,
+      "step": 9535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.174692143206542e-05,
+      "loss": 1.2111,
+      "step": 9536
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1745272812000746e-05,
+      "loss": 0.5763,
+      "step": 9537
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1743624053557415e-05,
+      "loss": 0.4022,
+      "step": 9538
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.174197515674594e-05,
+      "loss": 0.4813,
+      "step": 9539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.17403261215768e-05,
+      "loss": 0.8326,
+      "step": 9540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.173867694806049e-05,
+      "loss": 0.4002,
+      "step": 9541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.173702763620752e-05,
+      "loss": 1.1904,
+      "step": 9542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.173537818602837e-05,
+      "loss": 0.8321,
+      "step": 9543
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1733728597533544e-05,
+      "loss": 1.5736,
+      "step": 9544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.173207887073353e-05,
+      "loss": 0.2619,
+      "step": 9545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.173042900563885e-05,
+      "loss": 1.0467,
+      "step": 9546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.172877900225997e-05,
+      "loss": 1.067,
+      "step": 9547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1727128860607414e-05,
+      "loss": 0.7854,
+      "step": 9548
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.172547858069167e-05,
+      "loss": 1.2759,
+      "step": 9549
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1723828162523245e-05,
+      "loss": 0.3659,
+      "step": 9550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.172217760611265e-05,
+      "loss": 0.7365,
+      "step": 9551
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.172052691147037e-05,
+      "loss": 0.3654,
+      "step": 9552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1718876078606926e-05,
+      "loss": 1.2956,
+      "step": 9553
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.171722510753281e-05,
+      "loss": 0.7628,
+      "step": 9554
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.171557399825855e-05,
+      "loss": 0.8533,
+      "step": 9555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.171392275079463e-05,
+      "loss": 0.5982,
+      "step": 9556
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1712271365151564e-05,
+      "loss": 1.5803,
+      "step": 9557
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.171061984133986e-05,
+      "loss": 0.8061,
+      "step": 9558
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1708968179370036e-05,
+      "loss": 0.5928,
+      "step": 9559
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.17073163792526e-05,
+      "loss": 0.8754,
+      "step": 9560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.170566444099805e-05,
+      "loss": 1.8494,
+      "step": 9561
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.170401236461692e-05,
+      "loss": 0.9685,
+      "step": 9562
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.170236015011971e-05,
+      "loss": 0.8769,
+      "step": 9563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.170070779751693e-05,
+      "loss": 1.3737,
+      "step": 9564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.16990553068191e-05,
+      "loss": 0.8722,
+      "step": 9565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1697402678036745e-05,
+      "loss": 1.6677,
+      "step": 9566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.169574991118037e-05,
+      "loss": 0.5448,
+      "step": 9567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.169409700626049e-05,
+      "loss": 0.8019,
+      "step": 9568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1692443963287634e-05,
+      "loss": 0.6113,
+      "step": 9569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1690790782272314e-05,
+      "loss": 0.4637,
+      "step": 9570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.168913746322506e-05,
+      "loss": 1.1905,
+      "step": 9571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1687484006156383e-05,
+      "loss": 0.6645,
+      "step": 9572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.168583041107681e-05,
+      "loss": 0.7581,
+      "step": 9573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.168417667799686e-05,
+      "loss": 1.0021,
+      "step": 9574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1682522806927054e-05,
+      "loss": 1.1021,
+      "step": 9575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1680868797877925e-05,
+      "loss": 1.5758,
+      "step": 9576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.167921465086e-05,
+      "loss": 1.2901,
+      "step": 9577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1677560365883795e-05,
+      "loss": 0.9724,
+      "step": 9578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.167590594295984e-05,
+      "loss": 1.199,
+      "step": 9579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.167425138209867e-05,
+      "loss": 0.7664,
+      "step": 9580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1672596683310816e-05,
+      "loss": 1.0848,
+      "step": 9581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.167094184660679e-05,
+      "loss": 1.3555,
+      "step": 9582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1669286871997134e-05,
+      "loss": 0.6705,
+      "step": 9583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.166763175949238e-05,
+      "loss": 0.6672,
+      "step": 9584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1665976509103065e-05,
+      "loss": 0.4661,
+      "step": 9585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1664321120839716e-05,
+      "loss": 1.2991,
+      "step": 9586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1662665594712864e-05,
+      "loss": 0.359,
+      "step": 9587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1661009930733054e-05,
+      "loss": 0.3798,
+      "step": 9588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.165935412891082e-05,
+      "loss": 1.4588,
+      "step": 9589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.16576981892567e-05,
+      "loss": 0.6304,
+      "step": 9590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.165604211178121e-05,
+      "loss": 0.4929,
+      "step": 9591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.165438589649492e-05,
+      "loss": 0.6867,
+      "step": 9592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.165272954340835e-05,
+      "loss": 1.3112,
+      "step": 9593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.165107305253206e-05,
+      "loss": 0.3117,
+      "step": 9594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.164941642387656e-05,
+      "loss": 1.4872,
+      "step": 9595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1647759657452416e-05,
+      "loss": 0.4116,
+      "step": 9596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1646102753270166e-05,
+      "loss": 1.3974,
+      "step": 9597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1644445711340356e-05,
+      "loss": 0.9462,
+      "step": 9598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.164278853167353e-05,
+      "loss": 0.2833,
+      "step": 9599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.164113121428022e-05,
+      "loss": 1.5098,
+      "step": 9600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1639473759170984e-05,
+      "loss": 1.3405,
+      "step": 9601
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1637816166356376e-05,
+      "loss": 0.759,
+      "step": 9602
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.163615843584694e-05,
+      "loss": 1.4082,
+      "step": 9603
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.163450056765321e-05,
+      "loss": 1.1367,
+      "step": 9604
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.163284256178576e-05,
+      "loss": 0.7303,
+      "step": 9605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.163118441825512e-05,
+      "loss": 0.5455,
+      "step": 9606
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.162952613707186e-05,
+      "loss": 1.2521,
+      "step": 9607
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.162786771824652e-05,
+      "loss": 0.4107,
+      "step": 9608
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.162620916178966e-05,
+      "loss": 0.333,
+      "step": 9609
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.162455046771183e-05,
+      "loss": 1.1975,
+      "step": 9610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1622891636023595e-05,
+      "loss": 0.8829,
+      "step": 9611
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1621232666735487e-05,
+      "loss": 0.4806,
+      "step": 9612
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1619573559858084e-05,
+      "loss": 0.665,
+      "step": 9613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.161791431540195e-05,
+      "loss": 0.9461,
+      "step": 9614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1616254933377626e-05,
+      "loss": 0.3718,
+      "step": 9615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.161459541379568e-05,
+      "loss": 1.078,
+      "step": 9616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.161293575666667e-05,
+      "loss": 1.8842,
+      "step": 9617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1611275962001166e-05,
+      "loss": 1.0842,
+      "step": 9618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.160961602980971e-05,
+      "loss": 0.8823,
+      "step": 9619
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.160795596010289e-05,
+      "loss": 0.4129,
+      "step": 9620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.160629575289125e-05,
+      "loss": 1.3962,
+      "step": 9621
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.160463540818537e-05,
+      "loss": 1.062,
+      "step": 9622
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.160297492599581e-05,
+      "loss": 1.0955,
+      "step": 9623
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.160131430633312e-05,
+      "loss": 1.3604,
+      "step": 9624
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.15996535492079e-05,
+      "loss": 1.0935,
+      "step": 9625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1597992654630695e-05,
+      "loss": 1.026,
+      "step": 9626
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.159633162261209e-05,
+      "loss": 0.7001,
+      "step": 9627
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1594670453162634e-05,
+      "loss": 0.1525,
+      "step": 9628
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.159300914629292e-05,
+      "loss": 0.3052,
+      "step": 9629
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1591347702013504e-05,
+      "loss": 1.4164,
+      "step": 9630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.158968612033496e-05,
+      "loss": 0.5024,
+      "step": 9631
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.158802440126788e-05,
+      "loss": 0.4127,
+      "step": 9632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1586362544822814e-05,
+      "loss": 0.9235,
+      "step": 9633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1584700551010356e-05,
+      "loss": 1.2838,
+      "step": 9634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.158303841984107e-05,
+      "loss": 1.6532,
+      "step": 9635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.158137615132554e-05,
+      "loss": 0.6589,
+      "step": 9636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.157971374547434e-05,
+      "loss": 1.4652,
+      "step": 9637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.157805120229806e-05,
+      "loss": 1.21,
+      "step": 9638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.157638852180726e-05,
+      "loss": 1.5413,
+      "step": 9639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.157472570401253e-05,
+      "loss": 0.605,
+      "step": 9640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.157306274892447e-05,
+      "loss": 0.9825,
+      "step": 9641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.157139965655363e-05,
+      "loss": 0.3103,
+      "step": 9642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1569736426910616e-05,
+      "loss": 0.4315,
+      "step": 9643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1568073060006e-05,
+      "loss": 1.2204,
+      "step": 9644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1566409555850375e-05,
+      "loss": 0.449,
+      "step": 9645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.156474591445432e-05,
+      "loss": 0.6308,
+      "step": 9646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.156308213582843e-05,
+      "loss": 0.1714,
+      "step": 9647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.156141821998329e-05,
+      "loss": 0.4504,
+      "step": 9648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.155975416692948e-05,
+      "loss": 1.4226,
+      "step": 9649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.15580899766776e-05,
+      "loss": 1.7002,
+      "step": 9650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.155642564923824e-05,
+      "loss": 0.6787,
+      "step": 9651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.155476118462198e-05,
+      "loss": 0.3269,
+      "step": 9652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.155309658283942e-05,
+      "loss": 0.6628,
+      "step": 9653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.155143184390116e-05,
+      "loss": 1.0312,
+      "step": 9654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1549766967817785e-05,
+      "loss": 1.0238,
+      "step": 9655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.154810195459988e-05,
+      "loss": 1.2264,
+      "step": 9656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1546436804258065e-05,
+      "loss": 0.3442,
+      "step": 9657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1544771516802925e-05,
+      "loss": 0.4819,
+      "step": 9658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.154310609224504e-05,
+      "loss": 0.2766,
+      "step": 9659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1541440530595035e-05,
+      "loss": 1.5943,
+      "step": 9660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.153977483186349e-05,
+      "loss": 0.8354,
+      "step": 9661
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.153810899606102e-05,
+      "loss": 0.549,
+      "step": 9662
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.153644302319821e-05,
+      "loss": 1.1099,
+      "step": 9663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.153477691328567e-05,
+      "loss": 0.5344,
+      "step": 9664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1533110666334e-05,
+      "loss": 0.9287,
+      "step": 9665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.153144428235382e-05,
+      "loss": 0.824,
+      "step": 9666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.152977776135571e-05,
+      "loss": 0.3883,
+      "step": 9667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.152811110335028e-05,
+      "loss": 1.6209,
+      "step": 9668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.152644430834815e-05,
+      "loss": 0.9873,
+      "step": 9669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1524777376359915e-05,
+      "loss": 0.4897,
+      "step": 9670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.152311030739617e-05,
+      "loss": 0.309,
+      "step": 9671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.152144310146756e-05,
+      "loss": 1.4091,
+      "step": 9672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.151977575858466e-05,
+      "loss": 0.8556,
+      "step": 9673
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.151810827875811e-05,
+      "loss": 1.1161,
+      "step": 9674
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.151644066199849e-05,
+      "loss": 1.4445,
+      "step": 9675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1514772908316425e-05,
+      "loss": 0.2701,
+      "step": 9676
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.151310501772254e-05,
+      "loss": 0.6194,
+      "step": 9677
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1511436990227426e-05,
+      "loss": 1.0162,
+      "step": 9678
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.150976882584172e-05,
+      "loss": 0.5695,
+      "step": 9679
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1508100524576024e-05,
+      "loss": 0.3182,
+      "step": 9680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1506432086440965e-05,
+      "loss": 1.2526,
+      "step": 9681
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.150476351144715e-05,
+      "loss": 0.5867,
+      "step": 9682
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.15030947996052e-05,
+      "loss": 1.233,
+      "step": 9683
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1501425950925736e-05,
+      "loss": 1.1774,
+      "step": 9684
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.149975696541937e-05,
+      "loss": 0.5524,
+      "step": 9685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1498087843096736e-05,
+      "loss": 0.9604,
+      "step": 9686
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1496418583968454e-05,
+      "loss": 0.4201,
+      "step": 9687
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.149474918804514e-05,
+      "loss": 1.0797,
+      "step": 9688
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.14930796553374e-05,
+      "loss": 0.8316,
+      "step": 9689
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1491409985855905e-05,
+      "loss": 1.201,
+      "step": 9690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.148974017961124e-05,
+      "loss": 0.6438,
+      "step": 9691
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1488070236614036e-05,
+      "loss": 0.9138,
+      "step": 9692
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.148640015687494e-05,
+      "loss": 0.508,
+      "step": 9693
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1484729940404565e-05,
+      "loss": 0.9358,
+      "step": 9694
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1483059587213535e-05,
+      "loss": 1.2207,
+      "step": 9695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.14813890973125e-05,
+      "loss": 1.7415,
+      "step": 9696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1479718470712055e-05,
+      "loss": 0.9441,
+      "step": 9697
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1478047707422876e-05,
+      "loss": 0.2872,
+      "step": 9698
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.147637680745556e-05,
+      "loss": 0.5388,
+      "step": 9699
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1474705770820755e-05,
+      "loss": 0.6982,
+      "step": 9700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.147303459752909e-05,
+      "loss": 1.3192,
+      "step": 9701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1471363287591194e-05,
+      "loss": 0.8154,
+      "step": 9702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.146969184101773e-05,
+      "loss": 1.5027,
+      "step": 9703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1468020257819296e-05,
+      "loss": 1.675,
+      "step": 9704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.146634853800656e-05,
+      "loss": 0.599,
+      "step": 9705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.146467668159014e-05,
+      "loss": 0.3209,
+      "step": 9706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1463004688580685e-05,
+      "loss": 0.5416,
+      "step": 9707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1461332558988845e-05,
+      "loss": 0.4022,
+      "step": 9708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1459660292825235e-05,
+      "loss": 0.9296,
+      "step": 9709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.145798789010051e-05,
+      "loss": 1.7327,
+      "step": 9710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.145631535082532e-05,
+      "loss": 1.4516,
+      "step": 9711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1454642675010295e-05,
+      "loss": 1.0753,
+      "step": 9712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.145296986266609e-05,
+      "loss": 1.1785,
+      "step": 9713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.145129691380335e-05,
+      "loss": 0.7905,
+      "step": 9714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1449623828432706e-05,
+      "loss": 1.5552,
+      "step": 9715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.144795060656482e-05,
+      "loss": 0.6411,
+      "step": 9716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.144627724821034e-05,
+      "loss": 1.1137,
+      "step": 9717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.144460375337991e-05,
+      "loss": 1.4045,
+      "step": 9718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.144293012208418e-05,
+      "loss": 1.407,
+      "step": 9719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1441256354333795e-05,
+      "loss": 1.4755,
+      "step": 9720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.143958245013941e-05,
+      "loss": 0.7295,
+      "step": 9721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1437908409511685e-05,
+      "loss": 0.5378,
+      "step": 9722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.143623423246127e-05,
+      "loss": 0.899,
+      "step": 9723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.14345599189988e-05,
+      "loss": 0.6188,
+      "step": 9724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.143288546913496e-05,
+      "loss": 0.4155,
+      "step": 9725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1431210882880375e-05,
+      "loss": 1.0904,
+      "step": 9726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.142953616024573e-05,
+      "loss": 1.8238,
+      "step": 9727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.142786130124166e-05,
+      "loss": 0.3747,
+      "step": 9728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.142618630587884e-05,
+      "loss": 0.5401,
+      "step": 9729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1424511174167916e-05,
+      "loss": 1.4254,
+      "step": 9730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1422835906119554e-05,
+      "loss": 1.3384,
+      "step": 9731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.142116050174442e-05,
+      "loss": 0.694,
+      "step": 9732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1419484961053164e-05,
+      "loss": 0.7597,
+      "step": 9733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.141780928405645e-05,
+      "loss": 1.4272,
+      "step": 9734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.141613347076495e-05,
+      "loss": 0.5274,
+      "step": 9735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.141445752118932e-05,
+      "loss": 1.6496,
+      "step": 9736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.141278143534024e-05,
+      "loss": 0.1233,
+      "step": 9737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1411105213228354e-05,
+      "loss": 1.281,
+      "step": 9738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.140942885486435e-05,
+      "loss": 1.2136,
+      "step": 9739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1407752360258875e-05,
+      "loss": 0.4499,
+      "step": 9740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1406075729422606e-05,
+      "loss": 0.3481,
+      "step": 9741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.140439896236623e-05,
+      "loss": 1.1807,
+      "step": 9742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.140272205910038e-05,
+      "loss": 0.5802,
+      "step": 9743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.140104501963576e-05,
+      "loss": 1.3998,
+      "step": 9744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.139936784398304e-05,
+      "loss": 0.9448,
+      "step": 9745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.139769053215287e-05,
+      "loss": 0.4285,
+      "step": 9746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1396013084155945e-05,
+      "loss": 0.6406,
+      "step": 9747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.139433550000293e-05,
+      "loss": 0.7449,
+      "step": 9748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1392657779704496e-05,
+      "loss": 0.5227,
+      "step": 9749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.139097992327134e-05,
+      "loss": 0.8275,
+      "step": 9750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1389301930714106e-05,
+      "loss": 0.6038,
+      "step": 9751
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.138762380204351e-05,
+      "loss": 1.592,
+      "step": 9752
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.13859455372702e-05,
+      "loss": 0.9661,
+      "step": 9753
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.138426713640488e-05,
+      "loss": 1.2595,
+      "step": 9754
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.138258859945822e-05,
+      "loss": 0.6822,
+      "step": 9755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.138090992644089e-05,
+      "loss": 0.841,
+      "step": 9756
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.137923111736358e-05,
+      "loss": 1.3556,
+      "step": 9757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1377552172236983e-05,
+      "loss": 1.2377,
+      "step": 9758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.137587309107177e-05,
+      "loss": 0.9584,
+      "step": 9759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.137419387387864e-05,
+      "loss": 0.6075,
+      "step": 9760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1372514520668274e-05,
+      "loss": 1.4536,
+      "step": 9761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.137083503145135e-05,
+      "loss": 1.0504,
+      "step": 9762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.136915540623857e-05,
+      "loss": 0.9709,
+      "step": 9763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1367475645040605e-05,
+      "loss": 0.8049,
+      "step": 9764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.136579574786816e-05,
+      "loss": 1.2269,
+      "step": 9765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.136411571473191e-05,
+      "loss": 0.4046,
+      "step": 9766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.136243554564256e-05,
+      "loss": 0.4088,
+      "step": 9767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.136075524061079e-05,
+      "loss": 0.7947,
+      "step": 9768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.135907479964732e-05,
+      "loss": 0.3527,
+      "step": 9769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.13573942227628e-05,
+      "loss": 1.0743,
+      "step": 9770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1355713509967966e-05,
+      "loss": 1.2227,
+      "step": 9771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.135403266127348e-05,
+      "loss": 0.8219,
+      "step": 9772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.135235167669006e-05,
+      "loss": 0.9886,
+      "step": 9773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1350670556228404e-05,
+      "loss": 0.5332,
+      "step": 9774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.134898929989919e-05,
+      "loss": 0.7607,
+      "step": 9775
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.134730790771314e-05,
+      "loss": 0.4543,
+      "step": 9776
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1345626379680936e-05,
+      "loss": 1.5176,
+      "step": 9777
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.134394471581329e-05,
+      "loss": 1.4569,
+      "step": 9778
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.13422629161209e-05,
+      "loss": 0.2274,
+      "step": 9779
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.134058098061447e-05,
+      "loss": 0.4708,
+      "step": 9780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1338898909304694e-05,
+      "loss": 0.5952,
+      "step": 9781
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.133721670220229e-05,
+      "loss": 1.3616,
+      "step": 9782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.133553435931795e-05,
+      "loss": 1.3742,
+      "step": 9783
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.133385188066239e-05,
+      "loss": 1.3254,
+      "step": 9784
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.133216926624631e-05,
+      "loss": 0.6484,
+      "step": 9785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.133048651608042e-05,
+      "loss": 0.7062,
+      "step": 9786
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.132880363017543e-05,
+      "loss": 0.5231,
+      "step": 9787
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.132712060854205e-05,
+      "loss": 1.3441,
+      "step": 9788
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.132543745119098e-05,
+      "loss": 0.6204,
+      "step": 9789
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1323754158132946e-05,
+      "loss": 1.3462,
+      "step": 9790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.132207072937865e-05,
+      "loss": 1.0065,
+      "step": 9791
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1320387164938804e-05,
+      "loss": 0.9118,
+      "step": 9792
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.131870346482413e-05,
+      "loss": 0.7603,
+      "step": 9793
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.131701962904534e-05,
+      "loss": 1.0233,
+      "step": 9794
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.131533565761313e-05,
+      "loss": 1.1877,
+      "step": 9795
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.131365155053825e-05,
+      "loss": 0.5582,
+      "step": 9796
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.13119673078314e-05,
+      "loss": 1.5619,
+      "step": 9797
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.131028292950329e-05,
+      "loss": 2.6783,
+      "step": 9798
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.130859841556465e-05,
+      "loss": 0.6939,
+      "step": 9799
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.130691376602619e-05,
+      "loss": 0.5525,
+      "step": 9800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.130522898089864e-05,
+      "loss": 0.4851,
+      "step": 9801
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.130354406019272e-05,
+      "loss": 1.4872,
+      "step": 9802
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.130185900391915e-05,
+      "loss": 0.6113,
+      "step": 9803
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1300173812088644e-05,
+      "loss": 1.0235,
+      "step": 9804
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.129848848471194e-05,
+      "loss": 0.388,
+      "step": 9805
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.129680302179976e-05,
+      "loss": 1.3518,
+      "step": 9806
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1295117423362825e-05,
+      "loss": 1.1895,
+      "step": 9807
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.129343168941186e-05,
+      "loss": 1.2792,
+      "step": 9808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1291745819957605e-05,
+      "loss": 0.3232,
+      "step": 9809
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.129005981501078e-05,
+      "loss": 1.1598,
+      "step": 9810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.12883736745821e-05,
+      "loss": 0.4241,
+      "step": 9811
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1286687398682324e-05,
+      "loss": 0.8171,
+      "step": 9812
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1285000987322164e-05,
+      "loss": 1.8208,
+      "step": 9813
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.128331444051234e-05,
+      "loss": 0.7778,
+      "step": 9814
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.128162775826362e-05,
+      "loss": 0.7202,
+      "step": 9815
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.127994094058671e-05,
+      "loss": 0.7287,
+      "step": 9816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1278253987492346e-05,
+      "loss": 0.7039,
+      "step": 9817
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.127656689899127e-05,
+      "loss": 0.9916,
+      "step": 9818
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1274879675094226e-05,
+      "loss": 0.376,
+      "step": 9819
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1273192315811925e-05,
+      "loss": 1.0307,
+      "step": 9820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.127150482115513e-05,
+      "loss": 0.689,
+      "step": 9821
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.126981719113457e-05,
+      "loss": 0.4435,
+      "step": 9822
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.126812942576098e-05,
+      "loss": 0.3826,
+      "step": 9823
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1266441525045115e-05,
+      "loss": 1.8374,
+      "step": 9824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.12647534889977e-05,
+      "loss": 1.6093,
+      "step": 9825
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1263065317629486e-05,
+      "loss": 2.2007,
+      "step": 9826
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.12613770109512e-05,
+      "loss": 0.4435,
+      "step": 9827
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.125968856897362e-05,
+      "loss": 0.1576,
+      "step": 9828
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1257999991707455e-05,
+      "loss": 0.5218,
+      "step": 9829
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.125631127916347e-05,
+      "loss": 0.1557,
+      "step": 9830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.125462243135241e-05,
+      "loss": 1.3982,
+      "step": 9831
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1252933448285013e-05,
+      "loss": 1.3734,
+      "step": 9832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1251244329972034e-05,
+      "loss": 0.9552,
+      "step": 9833
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.124955507642423e-05,
+      "loss": 1.7075,
+      "step": 9834
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.124786568765232e-05,
+      "loss": 1.5997,
+      "step": 9835
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1246176163667086e-05,
+      "loss": 0.7346,
+      "step": 9836
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.124448650447927e-05,
+      "loss": 1.4396,
+      "step": 9837
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.124279671009963e-05,
+      "loss": 1.1977,
+      "step": 9838
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1241106780538905e-05,
+      "loss": 1.2189,
+      "step": 9839
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.123941671580786e-05,
+      "loss": 0.9315,
+      "step": 9840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.123772651591725e-05,
+      "loss": 0.671,
+      "step": 9841
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.123603618087782e-05,
+      "loss": 0.9388,
+      "step": 9842
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.123434571070034e-05,
+      "loss": 0.8185,
+      "step": 9843
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1232655105395556e-05,
+      "loss": 0.5843,
+      "step": 9844
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1230964364974245e-05,
+      "loss": 0.7122,
+      "step": 9845
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.122927348944714e-05,
+      "loss": 1.5117,
+      "step": 9846
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1227582478825016e-05,
+      "loss": 0.4915,
+      "step": 9847
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.122589133311863e-05,
+      "loss": 0.9005,
+      "step": 9848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1224200052338765e-05,
+      "loss": 0.7782,
+      "step": 9849
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.122250863649615e-05,
+      "loss": 1.7332,
+      "step": 9850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.122081708560156e-05,
+      "loss": 1.0507,
+      "step": 9851
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1219125399665775e-05,
+      "loss": 1.1134,
+      "step": 9852
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1217433578699536e-05,
+      "loss": 0.2872,
+      "step": 9853
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.121574162271363e-05,
+      "loss": 1.1051,
+      "step": 9854
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1214049531718816e-05,
+      "loss": 1.1271,
+      "step": 9855
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1212357305725856e-05,
+      "loss": 0.4226,
+      "step": 9856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1210664944745525e-05,
+      "loss": 1.2312,
+      "step": 9857
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.120897244878858e-05,
+      "loss": 0.3025,
+      "step": 9858
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.120727981786583e-05,
+      "loss": 0.4082,
+      "step": 9859
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1205587051987997e-05,
+      "loss": 1.3968,
+      "step": 9860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1203894151165874e-05,
+      "loss": 1.3312,
+      "step": 9861
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.120220111541024e-05,
+      "loss": 1.0046,
+      "step": 9862
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.120050794473187e-05,
+      "loss": 0.7127,
+      "step": 9863
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.119881463914153e-05,
+      "loss": 0.2777,
+      "step": 9864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.119712119865e-05,
+      "loss": 0.7565,
+      "step": 9865
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.119542762326805e-05,
+      "loss": 0.6742,
+      "step": 9866
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.119373391300646e-05,
+      "loss": 0.2829,
+      "step": 9867
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1192040067876006e-05,
+      "loss": 0.5579,
+      "step": 9868
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1190346087887484e-05,
+      "loss": 1.3076,
+      "step": 9869
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.118865197305165e-05,
+      "loss": 0.062,
+      "step": 9870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.11869577233793e-05,
+      "loss": 1.8487,
+      "step": 9871
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.118526333888122e-05,
+      "loss": 1.1424,
+      "step": 9872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.118356881956817e-05,
+      "loss": 1.1942,
+      "step": 9873
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1181874165450946e-05,
+      "loss": 0.419,
+      "step": 9874
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.118017937654034e-05,
+      "loss": 0.9841,
+      "step": 9875
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1178484452847125e-05,
+      "loss": 0.4205,
+      "step": 9876
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.117678939438209e-05,
+      "loss": 1.1439,
+      "step": 9877
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.117509420115603e-05,
+      "loss": 0.8497,
+      "step": 9878
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.117339887317972e-05,
+      "loss": 1.1761,
+      "step": 9879
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1171703410463965e-05,
+      "loss": 2.2265,
+      "step": 9880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1170007813019525e-05,
+      "loss": 1.2195,
+      "step": 9881
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1168312080857225e-05,
+      "loss": 1.1706,
+      "step": 9882
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.116661621398784e-05,
+      "loss": 1.5886,
+      "step": 9883
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1164920212422145e-05,
+      "loss": 0.7485,
+      "step": 9884
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1163224076170966e-05,
+      "loss": 1.3228,
+      "step": 9885
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.116152780524507e-05,
+      "loss": 1.3356,
+      "step": 9886
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.115983139965527e-05,
+      "loss": 0.3603,
+      "step": 9887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.115813485941235e-05,
+      "loss": 0.485,
+      "step": 9888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1156438184527104e-05,
+      "loss": 0.7495,
+      "step": 9889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.115474137501033e-05,
+      "loss": 1.3115,
+      "step": 9890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.115304443087284e-05,
+      "loss": 1.76,
+      "step": 9891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.115134735212542e-05,
+      "loss": 0.3462,
+      "step": 9892
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1149650138778867e-05,
+      "loss": 0.4644,
+      "step": 9893
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1147952790843994e-05,
+      "loss": 1.3975,
+      "step": 9894
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1146255308331586e-05,
+      "loss": 0.6397,
+      "step": 9895
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.114455769125246e-05,
+      "loss": 1.1477,
+      "step": 9896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.114285993961741e-05,
+      "loss": 0.9682,
+      "step": 9897
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.114116205343724e-05,
+      "loss": 1.2461,
+      "step": 9898
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.113946403272276e-05,
+      "loss": 0.3756,
+      "step": 9899
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.113776587748478e-05,
+      "loss": 0.67,
+      "step": 9900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.113606758773409e-05,
+      "loss": 1.4258,
+      "step": 9901
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.113436916348151e-05,
+      "loss": 1.0243,
+      "step": 9902
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.113267060473784e-05,
+      "loss": 0.8218,
+      "step": 9903
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.113097191151389e-05,
+      "loss": 1.6141,
+      "step": 9904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.112927308382049e-05,
+      "loss": 0.9089,
+      "step": 9905
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.112757412166842e-05,
+      "loss": 1.1069,
+      "step": 9906
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.112587502506851e-05,
+      "loss": 0.6788,
+      "step": 9907
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.112417579403157e-05,
+      "loss": 0.9826,
+      "step": 9908
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.11224764285684e-05,
+      "loss": 0.2788,
+      "step": 9909
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1120776928689845e-05,
+      "loss": 1.2236,
+      "step": 9910
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1119077294406684e-05,
+      "loss": 1.1181,
+      "step": 9911
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.111737752572976e-05,
+      "loss": 0.6004,
+      "step": 9912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.111567762266987e-05,
+      "loss": 0.709,
+      "step": 9913
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.111397758523784e-05,
+      "loss": 1.0871,
+      "step": 9914
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.111227741344449e-05,
+      "loss": 1.0037,
+      "step": 9915
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.111057710730065e-05,
+      "loss": 0.3645,
+      "step": 9916
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.110887666681711e-05,
+      "loss": 1.6155,
+      "step": 9917
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1107176092004724e-05,
+      "loss": 1.3604,
+      "step": 9918
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1105475382874293e-05,
+      "loss": 1.1534,
+      "step": 9919
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1103774539436645e-05,
+      "loss": 0.6965,
+      "step": 9920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.110207356170261e-05,
+      "loss": 0.5282,
+      "step": 9921
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1100372449682995e-05,
+      "loss": 0.3575,
+      "step": 9922
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1098671203388636e-05,
+      "loss": 1.3258,
+      "step": 9923
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1096969822830374e-05,
+      "loss": 0.5754,
+      "step": 9924
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1095268308019016e-05,
+      "loss": 1.2384,
+      "step": 9925
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.10935666589654e-05,
+      "loss": 1.2308,
+      "step": 9926
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1091864875680346e-05,
+      "loss": 0.3503,
+      "step": 9927
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1090162958174685e-05,
+      "loss": 0.4054,
+      "step": 9928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.108846090645925e-05,
+      "loss": 1.3902,
+      "step": 9929
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1086758720544865e-05,
+      "loss": 1.4759,
+      "step": 9930
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1085056400442384e-05,
+      "loss": 0.8938,
+      "step": 9931
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.108335394616261e-05,
+      "loss": 0.8889,
+      "step": 9932
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.108165135771641e-05,
+      "loss": 1.2016,
+      "step": 9933
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1079948635114584e-05,
+      "loss": 0.56,
+      "step": 9934
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1078245778368e-05,
+      "loss": 1.4449,
+      "step": 9935
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.107654278748746e-05,
+      "loss": 1.6206,
+      "step": 9936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1074839662483836e-05,
+      "loss": 1.081,
+      "step": 9937
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.107313640336794e-05,
+      "loss": 0.4163,
+      "step": 9938
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.107143301015063e-05,
+      "loss": 0.4586,
+      "step": 9939
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.106972948284273e-05,
+      "loss": 1.1839,
+      "step": 9940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.106802582145509e-05,
+      "loss": 1.3409,
+      "step": 9941
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.106632202599855e-05,
+      "loss": 0.3297,
+      "step": 9942
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.106461809648394e-05,
+      "loss": 0.6593,
+      "step": 9943
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1062914032922124e-05,
+      "loss": 0.7712,
+      "step": 9944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1061209835323926e-05,
+      "loss": 0.8544,
+      "step": 9945
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.10595055037002e-05,
+      "loss": 0.7551,
+      "step": 9946
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.105780103806181e-05,
+      "loss": 0.9756,
+      "step": 9947
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.105609643841957e-05,
+      "loss": 0.6333,
+      "step": 9948
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.105439170478434e-05,
+      "loss": 0.3182,
+      "step": 9949
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.105268683716698e-05,
+      "loss": 0.5325,
+      "step": 9950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1050981835578335e-05,
+      "loss": 0.2913,
+      "step": 9951
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1049276700029234e-05,
+      "loss": 0.8316,
+      "step": 9952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.104757143053055e-05,
+      "loss": 1.41,
+      "step": 9953
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1045866027093134e-05,
+      "loss": 1.0803,
+      "step": 9954
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.104416048972783e-05,
+      "loss": 0.5318,
+      "step": 9955
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1042454818445496e-05,
+      "loss": 0.9802,
+      "step": 9956
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.104074901325698e-05,
+      "loss": 1.267,
+      "step": 9957
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1039043074173145e-05,
+      "loss": 0.3265,
+      "step": 9958
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1037337001204835e-05,
+      "loss": 1.1303,
+      "step": 9959
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1035630794362926e-05,
+      "loss": 0.805,
+      "step": 9960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.103392445365826e-05,
+      "loss": 0.3073,
+      "step": 9961
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.103221797910169e-05,
+      "loss": 0.8807,
+      "step": 9962
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.103051137070411e-05,
+      "loss": 0.6903,
+      "step": 9963
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1028804628476335e-05,
+      "loss": 0.6977,
+      "step": 9964
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.102709775242926e-05,
+      "loss": 0.232,
+      "step": 9965
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.102539074257373e-05,
+      "loss": 1.3969,
+      "step": 9966
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1023683598920616e-05,
+      "loss": 1.1918,
+      "step": 9967
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1021976321480765e-05,
+      "loss": 1.3124,
+      "step": 9968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.102026891026506e-05,
+      "loss": 0.3084,
+      "step": 9969
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.101856136528437e-05,
+      "loss": 0.8225,
+      "step": 9970
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.101685368654954e-05,
+      "loss": 0.4507,
+      "step": 9971
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.101514587407145e-05,
+      "loss": 1.734,
+      "step": 9972
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.101343792786097e-05,
+      "loss": 0.4456,
+      "step": 9973
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.1011729847928957e-05,
+      "loss": 1.531,
+      "step": 9974
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.101002163428629e-05,
+      "loss": 0.5912,
+      "step": 9975
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.100831328694385e-05,
+      "loss": 1.4128,
+      "step": 9976
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 39392,
+  "num_train_epochs": 4,
+  "save_steps": 500,
+  "total_flos": 7.568072040382464e+17,
+  "trial_name": null,
+  "trial_params": null
+}